{"dirty": [{"country": " seychelles ", "city": "hemmingen", "amount": "7.426,31", "company": "Wonka Co "}, {"country": "REPUBLIC OF  SEYCHELLES", "city": "Hemmingen", "amount": "$1,416.82", "company": " Hooli "}, {"country": "SUR", "city": "Montreal ", "amount": "2.029,24", "company": "   Initech"}, {"country": "Lux", "city": "HEMMINGEN   ", "amount": "3.134,52", "company": "Hooli  "}, {"country": "saint martin (french part)", "city": "hemmingen", "amount": "3.996,74", "company": "Umbrella"}, {"country": "SUINAME", "city": "   hemmingen", "amount": "$2,685.25", "company": "Umbrella"}, {"country": "suriname", "city": "   HEMMINGEN", "amount": "5.582,07", "company": "Initech"}, {"country": "SR", "city": "hemmingen", "amount": "$3,664.18", "company": "Vehement  "}, {"country": "Seychelles", "city": "HEMMINGEN", "amount": "#N/A", "company": "   Cyberdyne"}, {"country": "Grand Duchy of Luxembourg", "city": "montreal", "amount": "-100", "company": "Acme Inc "}, {"country": "Luxembourg", "city": "montreal", "amount": "4.812,79", "company": " Cyberdyne "}, {"country": "Republic of Seychelles", "city": "MONTREAL   ", "amount": "5.381,82", "company": "Umbrella"}, {"country": "Republic of Guinea", "city": "Hemmingen   ", "amount": "$629.82", "company": " Hooli "}, {"country": "REPUBLIC OF GUINEA", "city": "MONTREAL", "amount": "$7,094.24", "company": "Hooli"}, {"country": "Luxembourg   ", "city": "Hemmingen", "amount": "--", "company": "Hooli"}, {"country": "Seychelles", "city": "HEMMINGEN", "amount": "#N/A", "company": "   Cyberdyne"}, {"country": "", "city": "", "amount": "", "company": ""}], "clean": [{"country": "Seychelles", "city": "Hemmingen", "amount": 7426.31, "company": "Wonka Co"}, {"country": "Seychelles", "city": "Hemmingen", "amount": 1416.82, "company": "Hooli"}, {"country": "Suriname", "city": "Montreal", "amount": 2029.24, "company": "Initech"}, {"country": "Luxembourg", "city": "Hemmingen", "amount": 3134.52, "company": "Hooli"}, {"country": "Saint Martin (French part)", "city": "Hemmingen", "amount": 3996.74, "company": "Umbrella"}, {"country": "Suriname", "city": "Hemmingen", "amount": 2685.25, "company": "Umbrella"}, {"country": "Suriname", "city": "Hemmingen", "amount": 5582.07, "company": "Initech"}, {"country": "Suriname", "city": "Hemmingen", "amount": 3664.18, "company": "Vehement"}, {"country": "Seychelles", "city": "Hemmingen", "amount": NaN, "company": "Cyberdyne"}, {"country": "Luxembourg", "city": "Montreal", "amount": -100.0, "company": "Acme Inc"}, {"country": "Luxembourg", "city": "Montreal", "amount": 4812.79, "company": "Cyberdyne"}, {"country": "Seychelles", "city": "Montreal", "amount": 5381.82, "company": "Umbrella"}, {"country": "Guinea", "city": "Hemmingen", "amount": 629.82, "company": "Hooli"}, {"country": "Guinea", "city": "Montreal", "amount": 7094.24, "company": "Hooli"}, {"country": "Luxembourg", "city": "Hemmingen", "amount": NaN, "company": "Hooli"}], "dirty_cols": ["country", "city", "amount", "company"], "clean_cols": ["country", "city", "amount", "company"], "plan": {"dataset_summary": "17 rows × 4 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"seychelles": "Seychelles", "REPUBLIC OF SEYCHELLES": "Seychelles", "SUR": "Suriname", "Lux": "Luxembourg", "saint martin (french part)": "Saint Martin (French part)", "SUINAME": "Suriname", "suriname": "Suriname", "SR": "Suriname", "Grand Duchy of Luxembourg": "Luxembourg", "Republic of Seychelles": "Seychelles", "Republic of Guinea": "Guinea", "REPUBLIC OF GUINEA": "Guinea"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"hemmingen": "Hemmingen", "HEMMINGEN": "Hemmingen", "montreal": "Montreal", "MONTREAL": "Montreal"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"unit": "   liter", "job_title": "acct exec", "currency": "WIIR Franc", "email": "HOEC@MAIL.COM ", "is_active": "F", "name": " Heidi Adams", "signup_date": "1/23/2023", "notes2": ""}, {"unit": "IN", "job_title": "ENGINEERING MGR  ", "currency": "bermudian dollar", "email": "mnlp@corp.io", "is_active": "false", "name": "Heidi  Novak", "signup_date": "44949", "notes2": ""}, {"unit": "litre ", "job_title": "Acct Exec", "currency": "WIR FRANC", "email": "PEKAPAE@MAIL.COM", "is_active": "Y", "name": "Bob Cruz  ", "signup_date": "44962", "notes2": ""}, {"unit": "inch", "job_title": "acct exec", "currency": "bmd", "email": "nmcdfjk@example.com", "is_active": "0", "name": "Heidi Moore   ", "signup_date": "45186", "notes2": ""}, {"unit": "   inches", "job_title": "ENGINEERING  MGR", "currency": "WIR FRANC", "email": "eboln@example.com", "is_active": "true", "name": "Mona Ali", "signup_date": "1 Feb 2023", "notes2": ""}, {"unit": "l", "job_title": "Eng Manager", "currency": "Bermudian Dollar", "email": " AJKPJFF@EXAMPLE.COM ", "is_active": "1", "name": "Carol Khan   ", "signup_date": "2023-11-13", "notes2": ""}, {"unit": "inch", "job_title": "Engineering Mgr", "currency": "bmd", "email": "EADONHE@CORP.IO", "is_active": "false", "name": "Bob Adams  ", "signup_date": "10/24/2023", "notes2": ""}, {"unit": "Litre", "job_title": "acct  exec", "currency": "WIR FRANC", "email": "EHJB@TEST.ORG", "is_active": "TRUE", "name": "Priya  Ali", "signup_date": "12 Aug 2023", "notes2": ""}, {"unit": "in", "job_title": "ACCOUNT EXECUTIVE", "currency": "BMD", "email": " NKMGO@TEST.ORG ", "is_active": "true", "name": "Karl Adams ", "signup_date": "5/12/2023", "notes2": ""}, {"unit": "liter", "job_title": "Engineering Manager", "currency": "BND", "email": " oddoja@mail.com", "is_active": "false", "name": "David Park", "signup_date": "5/11/2023", "notes2": ""}, {"unit": "LITRE", "job_title": "Account Executive ", "currency": "BMD", "email": "LBHJ@MAIL.COM", "is_active": "F", "name": "David Lee", "signup_date": "2/1/2023", "notes2": ""}, {"unit": "  l", "job_title": "eng  mgr", "currency": "Brunei Dollar", "email": " DCPBNEM@TEST.ORG", "is_active": "Yes", "name": " Ivan Johnson", "signup_date": "2023-01-06", "notes2": ""}, {"unit": "INCH", "job_title": "Acct Exec", "currency": "BMD", "email": "DLKBCGL@MAIL.COM", "is_active": "T", "name": "Karl Moore", "signup_date": "3 Sep 2023", "notes2": ""}, {"unit": "liters", "job_title": "Ae", "currency": "BND", "email": "HPGBHKI@TEST.ORG", "is_active": "Yes", "name": "   Heidi Petrov", "signup_date": "5/10/2023", "notes2": ""}, {"unit": "   Ilches", "job_title": "eng manager", "currency": "BMD", "email": "BJELC@MAIL.COM", "is_active": "TRUE", "name": "  Bob Cruz", "signup_date": "45206", "notes2": ""}, {"unit": "in", "job_title": "Acct  Exec", "currency": "BERMUDIJN DOLLAR", "email": "gafj@example.com", "is_active": "true", "name": "Eve Khan  ", "signup_date": "14 Feb 2023", "notes2": ""}, {"unit": "ltier", "job_title": "ENGINEERINGG MANAGER", "currency": "BMD", "email": "FNJPMKI@TEST.ORG   ", "is_active": "T", "name": "Ivan Cruz", "signup_date": "44927", "notes2": ""}, {"unit": "litre", "job_title": "acct exec", "currency": "chw", "email": "GBIGK@CORP.IO", "is_active": "1", "name": "Grace Khan", "signup_date": "8 Apr 2023", "notes2": ""}, {"unit": "\"", "job_title": "eng manger", "currency": "chw", "email": "dibli@corp.io", "is_active": "true", "name": "Frank Lee ", "signup_date": "2023-01-22", "notes2": ""}, {"unit": "in", "job_title": "AE", "currency": "BRUNEI DOLLAR", "email": " OJMPEHC@MAIL.COM ", "is_active": "FALSE", "name": "Grace Ali", "signup_date": "45013", "notes2": ""}, {"unit": "inches", "job_title": "Acct Exec", "currency": "WIR Franc ", "email": "MEDNBG@TEST.ORG ", "is_active": "0", "name": "Ivan Lee ", "signup_date": "2023-08-14", "notes2": ""}, {"unit": "\"", "job_title": "acct exec", "currency": "CHW", "email": "okgdpl@corp.io", "is_active": "false", "name": "Mona Lee   ", "signup_date": "2023-10-17", "notes2": ""}, {"unit": "inchs", "job_title": "AE", "currency": "Bermudian Dollar", "email": "  hpbn@example.com", "is_active": "T", "name": "Lena Diaz", "signup_date": "23 Jul 2023", "notes2": ""}, {"unit": "inch", "job_title": "engineering mgr", "currency": "   CHW", "email": "   CLNGD@TEST.ORG", "is_active": "T", "name": "Priya  Cruz", "signup_date": "2023-04-04", "notes2": ""}, {"unit": " \"", "job_title": "Acct Exec", "currency": "chw", "email": "BNJEIB@TEST.ORG", "is_active": "T", "name": " Judy Adams", "signup_date": "45020", "notes2": ""}, {"unit": "liters", "job_title": "Account Executive", "currency": "brunei dollar", "email": "   GJABFP@EXAMPLE.COM", "is_active": "0", "name": " Lena Brandt", "signup_date": "3/13/2023", "notes2": ""}, {"unit": "litre", "job_title": "Engineering Manager", "currency": " Bnd ", "email": "cfdcicl@corp.io", "is_active": "T", "name": "Bob Cruz", "signup_date": "11 Apr 2023", "notes2": ""}, {"unit": " liter ", "job_title": "Engineering Mgr", "currency": "BRUNEI DOLLAR", "email": "DNHP@TEST.ORG", "is_active": "0", "name": "Karl Moore", "signup_date": "44950", "notes2": ""}, {"unit": "l", "job_title": "ENG MGR", "currency": "bermudian dolalr", "email": "BCMMEN@CORP.IO", "is_active": "true", "name": "Karl  Petrov", "signup_date": "11/4/2023", "notes2": ""}, {"unit": " \" ", "job_title": "Acct Exec", "currency": "wir franc", "email": " IJOOO@TEST.ORG", "is_active": "0", "name": "   Mona Khan", "signup_date": "2023-01-23", "notes2": ""}, {"unit": "\"", "job_title": "AE", "currency": " WIR FRANC", "email": "nigole@mail.com", "is_active": "No", "name": "Sara Fischer", "signup_date": "45274", "notes2": ""}, {"unit": " \" ", "job_title": "Acct  Exec", "currency": "bermudian  dollar", "email": " emnek@example.com ", "is_active": "Y", "name": "   Carol Adams", "signup_date": "45181", "notes2": ""}, {"unit": "inches", "job_title": "Engineering Mgr", "currency": "wir franc", "email": "mdmflm@corp.io", "is_active": "Yes", "name": "Mona Khan", "signup_date": "45183", "notes2": ""}, {"unit": "L", "job_title": "eng mgr", "currency": "bmd", "email": " JMOGCND@TEST.ORG ", "is_active": "0", "name": " Judy Ali", "signup_date": "7 Jul 2023", "notes2": ""}, {"unit": "liter", "job_title": "Engineering  Manager", "currency": "WIR  FRANC", "email": "kmmm@example.com", "is_active": "true", "name": "Mona Moore", "signup_date": "6/19/2023", "notes2": ""}, {"unit": "Liters", "job_title": "Account Executive", "currency": "BMD", "email": "IAAGBB@MAIL.COM", "is_active": "false", "name": "Grace Fischer", "signup_date": "45165", "notes2": ""}], "clean": [{"unit": "L", "job_title": "Account Executive", "currency": "CHW", "email": "hoec@mail.com", "is_active": false, "name": "Heidi Adams", "signup_date": "2023-01-23"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "BMD", "email": "mnlp@corp.io", "is_active": false, "name": "Heidi Novak", "signup_date": "2023-01-23"}, {"unit": "L", "job_title": "Account Executive", "currency": "CHW", "email": "pekapae@mail.com", "is_active": true, "name": "Bob Cruz", "signup_date": "2023-02-05"}, {"unit": "in", "job_title": "Account Executive", "currency": "BMD", "email": "nmcdfjk@example.com", "is_active": false, "name": "Heidi Moore", "signup_date": "2023-09-17"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "CHW", "email": "eboln@example.com", "is_active": true, "name": "Mona Ali", "signup_date": "2023-02-01"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BMD", "email": "ajkpjff@example.com", "is_active": true, "name": "Carol Khan", "signup_date": "2023-11-13"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "BMD", "email": "eadonhe@corp.io", "is_active": false, "name": "Bob Adams", "signup_date": "2023-10-24"}, {"unit": "L", "job_title": "Account Executive", "currency": "CHW", "email": "ehjb@test.org", "is_active": true, "name": "Priya Ali", "signup_date": "2023-08-12"}, {"unit": "in", "job_title": "Account Executive", "currency": "BMD", "email": "nkmgo@test.org", "is_active": true, "name": "Karl Adams", "signup_date": "2023-05-12"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BND", "email": "oddoja@mail.com", "is_active": false, "name": "David Park", "signup_date": "2023-05-11"}, {"unit": "L", "job_title": "Account Executive", "currency": "BMD", "email": "lbhj@mail.com", "is_active": false, "name": "David Lee", "signup_date": "2023-02-01"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BND", "email": "dcpbnem@test.org", "is_active": true, "name": "Ivan Johnson", "signup_date": "2023-01-06"}, {"unit": "in", "job_title": "Account Executive", "currency": "BMD", "email": "dlkbcgl@mail.com", "is_active": true, "name": "Karl Moore", "signup_date": "2023-09-03"}, {"unit": "L", "job_title": "Account Executive", "currency": "BND", "email": "hpgbhki@test.org", "is_active": true, "name": "Heidi Petrov", "signup_date": "2023-05-10"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "BMD", "email": "bjelc@mail.com", "is_active": true, "name": "Bob Cruz", "signup_date": "2023-10-07"}, {"unit": "in", "job_title": "Account Executive", "currency": "BMD", "email": "gafj@example.com", "is_active": true, "name": "Eve Khan", "signup_date": "2023-02-14"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BMD", "email": "fnjpmki@test.org", "is_active": true, "name": "Ivan Cruz", "signup_date": "2023-01-01"}, {"unit": "L", "job_title": "Account Executive", "currency": "CHW", "email": "gbigk@corp.io", "is_active": true, "name": "Grace Khan", "signup_date": "2023-04-08"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "CHW", "email": "dibli@corp.io", "is_active": true, "name": "Frank Lee", "signup_date": "2023-01-22"}, {"unit": "in", "job_title": "Account Executive", "currency": "BND", "email": "ojmpehc@mail.com", "is_active": false, "name": "Grace Ali", "signup_date": "2023-03-28"}, {"unit": "in", "job_title": "Account Executive", "currency": "CHW", "email": "mednbg@test.org", "is_active": false, "name": "Ivan Lee", "signup_date": "2023-08-14"}, {"unit": "in", "job_title": "Account Executive", "currency": "CHW", "email": "okgdpl@corp.io", "is_active": false, "name": "Mona Lee", "signup_date": "2023-10-17"}, {"unit": "in", "job_title": "Account Executive", "currency": "BMD", "email": "hpbn@example.com", "is_active": true, "name": "Lena Diaz", "signup_date": "2023-07-23"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "CHW", "email": "clngd@test.org", "is_active": true, "name": "Priya Cruz", "signup_date": "2023-04-04"}, {"unit": "in", "job_title": "Account Executive", "currency": "CHW", "email": "bnjeib@test.org", "is_active": true, "name": "Judy Adams", "signup_date": "2023-04-04"}, {"unit": "L", "job_title": "Account Executive", "currency": "BND", "email": "gjabfp@example.com", "is_active": false, "name": "Lena Brandt", "signup_date": "2023-03-13"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BND", "email": "cfdcicl@corp.io", "is_active": true, "name": "Bob Cruz", "signup_date": "2023-04-11"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BND", "email": "dnhp@test.org", "is_active": false, "name": "Karl Moore", "signup_date": "2023-01-24"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BMD", "email": "bcmmen@corp.io", "is_active": true, "name": "Karl Petrov", "signup_date": "2023-11-04"}, {"unit": "in", "job_title": "Account Executive", "currency": "CHW", "email": "ijooo@test.org", "is_active": false, "name": "Mona Khan", "signup_date": "2023-01-23"}, {"unit": "in", "job_title": "Account Executive", "currency": "CHW", "email": "nigole@mail.com", "is_active": false, "name": "Sara Fischer", "signup_date": "2023-12-14"}, {"unit": "in", "job_title": "Account Executive", "currency": "BMD", "email": "emnek@example.com", "is_active": true, "name": "Carol Adams", "signup_date": "2023-09-12"}, {"unit": "in", "job_title": "Engineering Manager", "currency": "CHW", "email": "mdmflm@corp.io", "is_active": true, "name": "Mona Khan", "signup_date": "2023-09-14"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "BMD", "email": "jmogcnd@test.org", "is_active": false, "name": "Judy Ali", "signup_date": "2023-07-07"}, {"unit": "L", "job_title": "Engineering Manager", "currency": "CHW", "email": "kmmm@example.com", "is_active": true, "name": "Mona Moore", "signup_date": "2023-06-19"}, {"unit": "L", "job_title": "Account Executive", "currency": "BMD", "email": "iaagbb@mail.com", "is_active": false, "name": "Grace Fischer", "signup_date": "2023-08-27"}], "dirty_cols": ["unit", "job_title", "currency", "email", "is_active", "name", "signup_date", "notes2"], "clean_cols": ["unit", "job_title", "currency", "email", "is_active", "name", "signup_date"], "plan": {"dataset_summary": "36 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liter": "L", "IN": "in", "litre": "L", "inch": "in", "inches": "in", "l": "L", "Litre": "L", "LITRE": "L", "INCH": "in", "liters": "L", "Ilches": "in", "ltier": "L", "\"": "in", "inchs": "in", "Liters": "L"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"acct exec": "Account Executive", "ENGINEERING MGR": "Engineering Manager", "Acct Exec": "Account Executive", "Eng Manager": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "ACCOUNT EXECUTIVE": "Account Executive", "eng mgr": "Engineering Manager", "Ae": "Account Executive", "eng manager": "Engineering Manager", "ENGINEERINGG MANAGER": "Engineering Manager", "eng manger": "Engineering Manager", "AE": "Account Executive", "engineering mgr": "Engineering Manager", "ENG MGR": "Engineering Manager"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WIIR Franc": "CHW", "bermudian dollar": "BMD", "WIR FRANC": "CHW", "bmd": "BMD", "Bermudian Dollar": "BMD", "Brunei Dollar": "BND", "BERMUDIJN DOLLAR": "BMD", "chw": "CHW", "BRUNEI DOLLAR": "BND", "WIR Franc": "CHW", "brunei dollar": "BND", "Bnd": "BND", "bermudian dolalr": "BMD", "wir franc": "CHW"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"phone": "(547)716-8068", "industry": "travel", "status": "free trial", "unnamed": ""}, {"phone": "5104916447", "industry": "PropTech", "status": "ACTIVE", "unnamed": ""}, {"phone": "509-642-5175", "industry": "   Telco", "status": "   Active", "unnamed": ""}, {"phone": "(597)310-1142", "industry": "REAL ESTATE", "status": "FREE TRIAL", "unnamed": ""}, {"phone": "533.594.8713", "industry": "Oil & Gas   ", "status": "paused", "unnamed": ""}, {"phone": "(515)020-8321", "industry": "TRAVEL", "status": "CHURNED", "unnamed": ""}, {"phone": "562.421.8580", "industry": "energy", "status": "actwve", "unnamed": ""}, {"phone": "536.307.2522", "industry": "Oil & Gas", "status": "  churned", "unnamed": ""}, {"phone": "570.779.6294", "industry": "Property", "status": "churned", "unnamed": ""}, {"phone": "(533)601-4989", "industry": " TELECOM", "status": "cancelled", "unnamed": ""}, {"phone": "554-257-2068", "industry": "hotels &  travel", "status": "canceled", "unnamed": ""}, {"phone": "594.342.5004", "industry": "Property", "status": "Actiive", "unnamed": ""}, {"phone": "597-248-9448", "industry": "property", "status": "   CHUNED", "unnamed": ""}, {"phone": "527.370.6062", "industry": " PROPTECH", "status": "active", "unnamed": ""}, {"phone": "527.370.6062", "industry": " PROPTECH", "status": "active", "unnamed": ""}], "clean": [{"phone": "(547) 716-8068", "industry": "Hospitality", "status": "Trial"}, {"phone": "(510) 491-6447", "industry": "Real Estate", "status": "Active"}, {"phone": "(509) 642-5175", "industry": "Telecommunications", "status": "Active"}, {"phone": "(597) 310-1142", "industry": "Real Estate", "status": "Trial"}, {"phone": "(533) 594-8713", "industry": "Energy", "status": "Paused"}, {"phone": "(515) 020-8321", "industry": "Hospitality", "status": "Churned"}, {"phone": "(562) 421-8580", "industry": "Energy", "status": "Active"}, {"phone": "(536) 307-2522", "industry": "Energy", "status": "Churned"}, {"phone": "(570) 779-6294", "industry": "Real Estate", "status": "Churned"}, {"phone": "(533) 601-4989", "industry": "Telecommunications", "status": "Churned"}, {"phone": "(554) 257-2068", "industry": "Hospitality", "status": "Churned"}, {"phone": "(594) 342-5004", "industry": "Real Estate", "status": "Active"}, {"phone": "(597) 248-9448", "industry": "Real Estate", "status": "Churned"}, {"phone": "(527) 370-6062", "industry": "Real Estate", "status": "Active"}], "dirty_cols": ["phone", "industry", "status", "unnamed"], "clean_cols": ["phone", "industry", "status"], "plan": {"dataset_summary": "15 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"travel": "Hospitality", "PropTech": "Real Estate", "Telco": "Telecommunications", "REAL ESTATE": "Real Estate", "Oil & Gas": "Energy", "TRAVEL": "Hospitality", "energy": "Energy", "Property": "Real Estate", "TELECOM": "Telecommunications", "hotels & travel": "Hospitality", "property": "Real Estate", "PROPTECH": "Real Estate"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"free trial": "Trial", "ACTIVE": "Active", "FREE TRIAL": "Trial", "paused": "Paused", "CHURNED": "Churned", "actwve": "Active", "churned": "Churned", "cancelled": "Churned", "canceled": "Churned", "Actiive": "Active", "CHUNED": "Churned", "active": "Active"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "2/13/2023", "currency": "BOV", "country": "isle of man", "state": "GU", "company": "Umbrella", "phone": "510.084.7147", "col_x": ""}, {"signup_date": "2023-04-25", "currency": "BWP  ", "country": "imn", "state": "guam", "company": "Wonka  Co", "phone": "594.354.7616", "col_x": ""}, {"signup_date": "20 Mar 2023", "currency": "brazilian real", "country": "IMN", "state": "in", "company": "Wonka Co", "phone": "(524)102-7580", "col_x": ""}, {"signup_date": "45226", "currency": "Unidad de Valor Real", "country": "republic of guinea", "state": "COLORADO", "company": "Vehement   ", "phone": "5137076168", "col_x": ""}, {"signup_date": "10 Jan 2023", "currency": "Mvdol", "country": "Anguilla", "state": "co", "company": "Hooli", "phone": "515.008.9701", "col_x": ""}, {"signup_date": "6/11/2023", "currency": "BOV", "country": "im", "state": "CO", "company": "Umbrella   ", "phone": "559.664.4634", "col_x": ""}, {"signup_date": "2023-06-10", "currency": "Unidad de Valor Real", "country": " IO", "state": "   GU", "company": "Hooli", "phone": "(571)300-7229", "col_x": ""}, {"signup_date": "4 Oct 2023", "currency": "   Brl", "country": "gn", "state": "co", "company": " Initech ", "phone": "(551)556-2759", "col_x": ""}, {"signup_date": "1 Oct 2023", "currency": "bov", "country": "gn", "state": "coloado", "company": " Acme Inc", "phone": "(572)798-5300", "col_x": ""}, {"signup_date": "2023-02-24", "currency": "brazilian real", "country": "isle of man", "state": "IDIANA", "company": "  Wonka Co", "phone": "526.848.7502", "col_x": ""}, {"signup_date": "45226", "currency": "Unidad de Valor Real", "country": "republic of guinea", "state": "COLORADO", "company": "Vehement   ", "phone": "5137076168", "col_x": ""}], "clean": [{"signup_date": "2023-02-13", "currency": "BOV", "country": "Isle of Man", "state": "Guam", "company": "Umbrella", "phone": "(510) 084-7147"}, {"signup_date": "2023-04-25", "currency": "BWP", "country": "Isle of Man", "state": "Guam", "company": "Wonka Co", "phone": "(594) 354-7616"}, {"signup_date": "2023-03-20", "currency": "BRL", "country": "Isle of Man", "state": "Indiana", "company": "Wonka Co", "phone": "(524) 102-7580"}, {"signup_date": "2023-10-27", "currency": "COU", "country": "Guinea", "state": "Colorado", "company": "Vehement", "phone": "(513) 707-6168"}, {"signup_date": "2023-01-10", "currency": "BOV", "country": "Anguilla", "state": "Colorado", "company": "Hooli", "phone": "(515) 008-9701"}, {"signup_date": "2023-06-11", "currency": "BOV", "country": "Isle of Man", "state": "Colorado", "company": "Umbrella", "phone": "(559) 664-4634"}, {"signup_date": "2023-06-10", "currency": "COU", "country": "British Indian Ocean Territory", "state": "Guam", "company": "Hooli", "phone": "(571) 300-7229"}, {"signup_date": "2023-10-04", "currency": "BRL", "country": "Guinea", "state": "Colorado", "company": "Initech", "phone": "(551) 556-2759"}, {"signup_date": "2023-10-01", "currency": "BOV", "country": "Guinea", "state": "Colorado", "company": "Acme Inc", "phone": "(572) 798-5300"}, {"signup_date": "2023-02-24", "currency": "BRL", "country": "Isle of Man", "state": "Indiana", "company": "Wonka Co", "phone": "(526) 848-7502"}], "dirty_cols": ["signup_date", "currency", "country", "state", "company", "phone", "col_x"], "clean_cols": ["signup_date", "currency", "country", "state", "company", "phone"], "plan": {"dataset_summary": "11 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"brazilian real": "BRL", "Unidad de Valor Real": "COU", "Mvdol": "BOV", "Brl": "BRL", "bov": "BOV"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"isle of man": "Isle of Man", "imn": "Isle of Man", "IMN": "Isle of Man", "republic of guinea": "Guinea", "im": "Isle of Man", "IO": "British Indian Ocean Territory", "gn": "Guinea"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GU": "Guam", "guam": "Guam", "in": "Indiana", "COLORADO": "Colorado", "co": "Colorado", "CO": "Colorado", "coloado": "Colorado", "IDIANA": "Indiana"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"industry": " Oil & Gas", "status": "cancelled", "currency": "CUP", "is_active": "Y", "city": "kirguli"}, {"industry": "utilities", "status": "cancelled", "currency": "cuban peso", "is_active": "N", "city": "Sanxi"}, {"industry": "Oil & Gas", "status": " Active ", "currency": "cuban peso", "is_active": "TRUE", "city": "church-yonge corridor"}, {"industry": "Technology  ", "status": "Paused", "currency": "GIP ", "is_active": "TRUE", "city": "TURFRU"}, {"industry": "EDUCATION", "status": "Paused", "currency": "cuban peso", "is_active": "T", "city": "Turrau"}, {"industry": "  IT", "status": "CANCELLED ", "currency": "Bahraini Dinar", "is_active": "T", "city": "  turaru"}, {"industry": " Utilities ", "status": "ACTIVE", "currency": "Gibraltar Pound", "is_active": "F", "city": "church-yonge corridor"}, {"industry": " technology", "status": "churned", "currency": "bahraini dikar", "is_active": "No", "city": "belize city"}, {"industry": "TECHNOLOGY", "status": "Active", "currency": "Cuban  Peso", "is_active": "false", "city": "Turaru"}, {"industry": "enegy", "status": "paused", "currency": "BHD", "is_active": "TRUE", "city": "Turaru"}, {"industry": " IT ", "status": "active", "currency": "CUP", "is_active": "No", "city": "  Kirguli"}, {"industry": "ED", "status": "on hold", "currency": "bahraini dinar", "is_active": "N", "city": "TUURARU   "}, {"industry": "Energy", "status": "paused", "currency": "cuban peso", "is_active": "FALSE", "city": "Turaru"}, {"industry": "oil & ges", "status": "paused", "currency": "CUP  ", "is_active": "FALSE", "city": "Kiirguli "}, {"industry": "utifities", "status": "Churned", "currency": "GIP", "is_active": "F", "city": "Kirguli"}, {"industry": "tech", "status": "Active", "currency": " Gip ", "is_active": "T", "city": "sanxi"}, {"industry": "Oil & Gas", "status": "Active", "currency": "BHD", "is_active": "FALSE", "city": " Sanxi"}, {"industry": "Utiities", "status": " Active ", "currency": "Cuban Peso", "is_active": "Yes", "city": "Belize City"}, {"industry": "Ed", "status": "on hold", "currency": "   Bahraini Dinar", "is_active": "false", "city": "church-yonge corriodr"}, {"industry": "oil & ges", "status": "paused", "currency": "CUP  ", "is_active": "FALSE", "city": "Kiirguli "}, {"industry": "", "status": "", "currency": "", "is_active": "", "city": ""}], "clean": [{"industry": "Energy", "status": "Churned", "currency": "CUP", "is_active": true, "city": "Kirguli"}, {"industry": "Energy", "status": "Churned", "currency": "CUP", "is_active": false, "city": "Sanxi"}, {"industry": "Energy", "status": "Active", "currency": "CUP", "is_active": true, "city": "Church-Yonge Corridor"}, {"industry": "Technology", "status": "Paused", "currency": "GIP", "is_active": true, "city": "Turaru"}, {"industry": "Education", "status": "Paused", "currency": "CUP", "is_active": true, "city": "Turaru"}, {"industry": "Technology", "status": "Churned", "currency": "BHD", "is_active": true, "city": "Turaru"}, {"industry": "Energy", "status": "Active", "currency": "GIP", "is_active": false, "city": "Church-Yonge Corridor"}, {"industry": "Technology", "status": "Churned", "currency": "BHD", "is_active": false, "city": "Belize City"}, {"industry": "Technology", "status": "Active", "currency": "CUP", "is_active": false, "city": "Turaru"}, {"industry": "Energy", "status": "Paused", "currency": "BHD", "is_active": true, "city": "Turaru"}, {"industry": "Technology", "status": "Active", "currency": "CUP", "is_active": false, "city": "Kirguli"}, {"industry": "Education", "status": "Paused", "currency": "BHD", "is_active": false, "city": "Turaru"}, {"industry": "Energy", "status": "Paused", "currency": "CUP", "is_active": false, "city": "Turaru"}, {"industry": "Energy", "status": "Paused", "currency": "CUP", "is_active": false, "city": "Kirguli"}, {"industry": "Energy", "status": "Churned", "currency": "GIP", "is_active": false, "city": "Kirguli"}, {"industry": "Technology", "status": "Active", "currency": "GIP", "is_active": true, "city": "Sanxi"}, {"industry": "Energy", "status": "Active", "currency": "BHD", "is_active": false, "city": "Sanxi"}, {"industry": "Energy", "status": "Active", "currency": "CUP", "is_active": true, "city": "Belize City"}, {"industry": "Education", "status": "Paused", "currency": "BHD", "is_active": false, "city": "Church-Yonge Corridor"}], "dirty_cols": ["industry", "status", "currency", "is_active", "city"], "clean_cols": ["industry", "status", "currency", "is_active", "city"], "plan": {"dataset_summary": "21 rows × 5 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Oil & Gas": "Energy", "utilities": "Energy", "EDUCATION": "Education", "IT": "Technology", "Utilities": "Energy", "technology": "Technology", "TECHNOLOGY": "Technology", "enegy": "Energy", "ED": "Education", "oil & ges": "Energy", "utifities": "Energy", "tech": "Technology", "Utiities": "Energy", "Ed": "Education"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cancelled": "Churned", "CANCELLED": "Churned", "ACTIVE": "Active", "churned": "Churned", "paused": "Paused", "active": "Active", "on hold": "Paused"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cuban peso": "CUP", "Bahraini Dinar": "BHD", "Gibraltar Pound": "GIP", "bahraini dikar": "BHD", "Cuban Peso": "CUP", "bahraini dinar": "BHD", "Gip": "GIP"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kirguli": "Kirguli", "church-yonge corridor": "Church-Yonge Corridor", "TURFRU": "Turaru", "Turrau": "Turaru", "turaru": "Turaru", "belize city": "Belize City", "TUURARU": "Turaru", "Kiirguli": "Kirguli", "sanxi": "Sanxi", "church-yonge corriodr": "Church-Yonge Corridor"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "(551)642-3335", "company": " Umbrella ", "job_title": " Eng Mgr", "city": " Darnytskyi Masyv", "unit": "mls  "}, {"phone": "587-085-9947", "company": "Acme  Inc", "job_title": "Engineering Mgr", "city": "Maulidn", "unit": "ml"}, {"phone": "(557)232-6508", "company": "Cyberdyne ", "job_title": "C.T.O.", "city": "Mauldin", "unit": "km"}, {"phone": "592-031-7999", "company": "Stark Industries", "job_title": "AE", "city": "temara", "unit": "KM"}, {"phone": "585.444.1060", "company": "   Acme Inc", "job_title": " engineering manager", "city": "Fossano", "unit": "ML"}, {"phone": "5451877191", "company": " Wonka Co", "job_title": "Engineering MManager   ", "city": "darnytskyi masyv", "unit": "kilometers"}, {"phone": "506.826.0633", "company": " Globex ", "job_title": "c.t.o.", "city": "mauldin", "unit": "MILLILITER"}, {"phone": "568.728.2103", "company": "Stark Industries", "job_title": "C.T.O.   ", "city": "Mauldin", "unit": "  km"}, {"phone": "(560)242-6032", "company": "   Stark Industries", "job_title": "Account Executive", "city": "darnytskyi masyv", "unit": "kilogram"}, {"phone": "5030972821", "company": "Umbrella", "job_title": "Eng Manager", "city": "Fosano", "unit": "Kg"}, {"phone": "(509)361-6688", "company": "Acme Inc ", "job_title": "C.T.O.", "city": "Temara  ", "unit": "mL"}, {"phone": "(549)207-6833", "company": "Umbrella ", "job_title": "ENGINEERING MGR", "city": "Darnytskyi Masyv", "unit": "lire  "}, {"phone": "5388375145", "company": "Stark Industries", "job_title": "Acct Exec", "city": "Temara", "unit": "KILOGRAMS"}, {"phone": "508.103.1707", "company": "Umbrella", "job_title": "C.T.O.", "city": "darnytskyi masyv", "unit": "KG"}, {"phone": "543-704-7173", "company": "Cyberdyne", "job_title": "acct exec", "city": "DARNYTSKYI MASYV", "unit": "kilometers"}, {"phone": "5641750791", "company": "Stark Industries", "job_title": "Engvneering Manager", "city": "FAIR  OAKS", "unit": "kilometers"}, {"phone": "577.374.2359", "company": "Soylent Corp", "job_title": "Eng  Manager", "city": "fossano", "unit": "liters"}, {"phone": "5974954326", "company": "Acme Inc", "job_title": "CTO", "city": "temara", "unit": "MILLILITER"}, {"phone": "537.208.6943", "company": "Umbrella", "job_title": "Eng Manager", "city": "  FOSSANO", "unit": "KG"}, {"phone": "(585)117-8200", "company": "Hooli", "job_title": "Acct Exec", "city": "Fossano", "unit": "KILOGRAMS"}, {"phone": "536-319-9313", "company": "Vehement", "job_title": "chief technology officer", "city": "Fair Oaks", "unit": " km"}, {"phone": "538-670-1311", "company": "Vehement", "job_title": "ACCT EXEC", "city": "Mauldin", "unit": "liter "}, {"phone": "(570)555-8596", "company": " Initech", "job_title": "Eng Mgr", "city": "Darnytskyi Masyv", "unit": "mls"}, {"phone": "5979552169", "company": "Wonka Co", "job_title": "cto", "city": "fossano", "unit": "km"}, {"phone": "506-513-7744", "company": "  Umbrella", "job_title": " CTO ", "city": "FAIR OAKS", "unit": " kg"}, {"phone": "572.421.8799", "company": " Vehement ", "job_title": "C.T.O.", "city": "darnytskyi masyv", "unit": "kilograis"}, {"phone": "542.469.6946", "company": "Stark Industries", "job_title": "eng manager ", "city": "Faib  Oaks", "unit": "liter"}, {"phone": "531.278.0062", "company": "Umbrella", "job_title": "Ae", "city": "  Fair Oaks", "unit": " km"}, {"phone": "541.791.4004", "company": "  Umbrella", "job_title": "CTO", "city": "darnytskyi masyv", "unit": "KILOGRAMS"}, {"phone": "590.323.7536", "company": "Globex", "job_title": "CTO", "city": "Mauldin", "unit": "mls"}, {"phone": "564.056.1950", "company": "Wonka Co", "job_title": "CTO", "city": "Temara", "unit": "  MILLILITER"}, {"phone": "506-257-2226", "company": "Soylent Corp", "job_title": "ENG MAOAGER", "city": "Darnytskkyi Masyv", "unit": "km"}, {"phone": "(595)908-1317", "company": "Wonka Co  ", "job_title": "C.T.O.", "city": "Darnytskyi Masyv", "unit": "ml"}, {"phone": "5457796591", "company": "   Umbrella", "job_title": " cto ", "city": " teara ", "unit": "liteps"}, {"phone": "(538)419-8989", "company": "Hooli   ", "job_title": "eng mgr", "city": "mauldin  ", "unit": "Kilogram"}, {"phone": "515.802.1115", "company": "  Umbrella", "job_title": "Acct Exec", "city": "Fair Oaks", "unit": "kilograms"}, {"phone": "557.720.1385", "company": "Vehement", "job_title": "Eng Manager ", "city": "Temara", "unit": " kilograms "}, {"phone": "(591)250-0449", "company": "Acme Inc  ", "job_title": "engineering manager", "city": "Darnytskyi Masyv", "unit": "litve"}, {"phone": "5505723569", "company": "Initech", "job_title": "Engineering Mgr", "city": "Darnytskyi Masyv", "unit": "kilograms"}, {"phone": "528.278.1431", "company": " Initech ", "job_title": "Engineering Mgr", "city": "Temzra", "unit": "l"}, {"phone": "534-282-7622", "company": "  Umbrella", "job_title": "Eng  Manager", "city": " Tefara ", "unit": "kilometers"}, {"phone": "(509)361-6688", "company": "Acme Inc ", "job_title": "C.T.O.", "city": "Temara  ", "unit": "mL"}, {"phone": "531.278.0062", "company": "Umbrella", "job_title": "Ae", "city": "  Fair Oaks", "unit": " km"}, {"phone": "", "company": "", "job_title": "", "city": "", "unit": ""}], "clean": [{"phone": "(551) 642-3335", "company": "Umbrella", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "mL"}, {"phone": "(587) 085-9947", "company": "Acme Inc", "job_title": "Engineering Manager", "city": "Mauldin", "unit": "mL"}, {"phone": "(557) 232-6508", "company": "Cyberdyne", "job_title": "Chief Technology Officer", "city": "Mauldin", "unit": "km"}, {"phone": "(592) 031-7999", "company": "Stark Industries", "job_title": "Account Executive", "city": "Temara", "unit": "km"}, {"phone": "(585) 444-1060", "company": "Acme Inc", "job_title": "Engineering Manager", "city": "Fossano", "unit": "mL"}, {"phone": "(545) 187-7191", "company": "Wonka Co", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "km"}, {"phone": "(506) 826-0633", "company": "Globex", "job_title": "Chief Technology Officer", "city": "Mauldin", "unit": "mL"}, {"phone": "(568) 728-2103", "company": "Stark Industries", "job_title": "Chief Technology Officer", "city": "Mauldin", "unit": "km"}, {"phone": "(560) 242-6032", "company": "Stark Industries", "job_title": "Account Executive", "city": "Darnytskyi Masyv", "unit": "kg"}, {"phone": "(503) 097-2821", "company": "Umbrella", "job_title": "Engineering Manager", "city": "Fossano", "unit": "kg"}, {"phone": "(509) 361-6688", "company": "Acme Inc", "job_title": "Chief Technology Officer", "city": "Temara", "unit": "mL"}, {"phone": "(549) 207-6833", "company": "Umbrella", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "L"}, {"phone": "(538) 837-5145", "company": "Stark Industries", "job_title": "Account Executive", "city": "Temara", "unit": "kg"}, {"phone": "(508) 103-1707", "company": "Umbrella", "job_title": "Chief Technology Officer", "city": "Darnytskyi Masyv", "unit": "kg"}, {"phone": "(543) 704-7173", "company": "Cyberdyne", "job_title": "Account Executive", "city": "Darnytskyi Masyv", "unit": "km"}, {"phone": "(564) 175-0791", "company": "Stark Industries", "job_title": "Engineering Manager", "city": "Fair Oaks", "unit": "km"}, {"phone": "(577) 374-2359", "company": "Soylent Corp", "job_title": "Engineering Manager", "city": "Fossano", "unit": "L"}, {"phone": "(597) 495-4326", "company": "Acme Inc", "job_title": "Chief Technology Officer", "city": "Temara", "unit": "mL"}, {"phone": "(537) 208-6943", "company": "Umbrella", "job_title": "Engineering Manager", "city": "Fossano", "unit": "kg"}, {"phone": "(585) 117-8200", "company": "Hooli", "job_title": "Account Executive", "city": "Fossano", "unit": "kg"}, {"phone": "(536) 319-9313", "company": "Vehement", "job_title": "Chief Technology Officer", "city": "Fair Oaks", "unit": "km"}, {"phone": "(538) 670-1311", "company": "Vehement", "job_title": "Account Executive", "city": "Mauldin", "unit": "L"}, {"phone": "(570) 555-8596", "company": "Initech", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "mL"}, {"phone": "(597) 955-2169", "company": "Wonka Co", "job_title": "Chief Technology Officer", "city": "Fossano", "unit": "km"}, {"phone": "(506) 513-7744", "company": "Umbrella", "job_title": "Chief Technology Officer", "city": "Fair Oaks", "unit": "kg"}, {"phone": "(572) 421-8799", "company": "Vehement", "job_title": "Chief Technology Officer", "city": "Darnytskyi Masyv", "unit": "kg"}, {"phone": "(542) 469-6946", "company": "Stark Industries", "job_title": "Engineering Manager", "city": "Fair Oaks", "unit": "L"}, {"phone": "(531) 278-0062", "company": "Umbrella", "job_title": "Account Executive", "city": "Fair Oaks", "unit": "km"}, {"phone": "(541) 791-4004", "company": "Umbrella", "job_title": "Chief Technology Officer", "city": "Darnytskyi Masyv", "unit": "kg"}, {"phone": "(590) 323-7536", "company": "Globex", "job_title": "Chief Technology Officer", "city": "Mauldin", "unit": "mL"}, {"phone": "(564) 056-1950", "company": "Wonka Co", "job_title": "Chief Technology Officer", "city": "Temara", "unit": "mL"}, {"phone": "(506) 257-2226", "company": "Soylent Corp", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "km"}, {"phone": "(595) 908-1317", "company": "Wonka Co", "job_title": "Chief Technology Officer", "city": "Darnytskyi Masyv", "unit": "mL"}, {"phone": "(545) 779-6591", "company": "Umbrella", "job_title": "Chief Technology Officer", "city": "Temara", "unit": "L"}, {"phone": "(538) 419-8989", "company": "Hooli", "job_title": "Engineering Manager", "city": "Mauldin", "unit": "kg"}, {"phone": "(515) 802-1115", "company": "Umbrella", "job_title": "Account Executive", "city": "Fair Oaks", "unit": "kg"}, {"phone": "(557) 720-1385", "company": "Vehement", "job_title": "Engineering Manager", "city": "Temara", "unit": "kg"}, {"phone": "(591) 250-0449", "company": "Acme Inc", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "L"}, {"phone": "(550) 572-3569", "company": "Initech", "job_title": "Engineering Manager", "city": "Darnytskyi Masyv", "unit": "kg"}, {"phone": "(528) 278-1431", "company": "Initech", "job_title": "Engineering Manager", "city": "Temara", "unit": "L"}, {"phone": "(534) 282-7622", "company": "Umbrella", "job_title": "Engineering Manager", "city": "Temara", "unit": "km"}], "dirty_cols": ["phone", "company", "job_title", "city", "unit"], "clean_cols": ["phone", "company", "job_title", "city", "unit"], "plan": {"dataset_summary": "44 rows × 5 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Eng Mgr": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "C.T.O.": "Chief Technology Officer", "AE": "Account Executive", "engineering manager": "Engineering Manager", "Engineering MManager": "Engineering Manager", "c.t.o.": "Chief Technology Officer", "Eng Manager": "Engineering Manager", "ENGINEERING MGR": "Engineering Manager", "Acct Exec": "Account Executive", "acct exec": "Account Executive", "Engvneering Manager": "Engineering Manager", "CTO": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "ACCT EXEC": "Account Executive", "cto": "Chief Technology Officer", "eng manager": "Engineering Manager", "Ae": "Account Executive", "ENG MAOAGER": "Engineering Manager", "eng mgr": "Engineering Manager"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Maulidn": "Mauldin", "temara": "Temara", "darnytskyi masyv": "Darnytskyi Masyv", "mauldin": "Mauldin", "Fosano": "Fossano", "DARNYTSKYI MASYV": "Darnytskyi Masyv", "FAIR OAKS": "Fair Oaks", "fossano": "Fossano", "FOSSANO": "Fossano", "Faib Oaks": "Fair Oaks", "Darnytskkyi Masyv": "Darnytskyi Masyv", "teara": "Temara", "Temzra": "Temara", "Tefara": "Temara"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mls": "mL", "ml": "mL", "KM": "km", "ML": "mL", "kilometers": "km", "MILLILITER": "mL", "kilogram": "kg", "Kg": "kg", "lire": "L", "KILOGRAMS": "kg", "KG": "kg", "liters": "L", "liter": "L", "kilograis": "kg", "liteps": "L", "Kilogram": "kg", "kilograms": "kg", "litve": "L", "l": "L"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "7.413,09", "state": "GUAM", "unit": "Cm  ", "country": "AZE", "status": "HIGH", "company": "Umbrella"}, {"amount": "1.380,29", "state": " Guam ", "unit": "in", "country": "az", "status": "HIGH", "company": "Globex"}, {"amount": "3.517,84", "state": "ak", "unit": "inches", "country": "Republic Of Azerbaijan ", "status": "HIGH", "company": "Acme Inc"}, {"amount": "296,36", "state": "Alaska ", "unit": "centimeter", "country": "azerbaijan", "status": "high", "company": "Wonka Co"}, {"amount": "$2,449.67", "state": " ak ", "unit": "INCHES", "country": "AZ", "status": "low", "company": "Wonka  Co"}, {"amount": "5.371,07", "state": "GU", "unit": "in", "country": "JAPAN", "status": " L ", "company": "Soylent Corp"}, {"amount": "5.812,20", "state": "gu", "unit": "\"", "country": "aze", "status": "low", "company": " Wonka Co"}, {"amount": "$3,274.22", "state": "Guam", "unit": "inches  ", "country": "AZ", "status": "p3", "company": "  Wonka Co"}, {"amount": "2.464,96", "state": "AK", "unit": "centimeter", "country": "jpn", "status": "High", "company": "Initech  "}, {"amount": "1.833,34", "state": "AK", "unit": "centimeters", "country": "jpn", "status": "high", "company": " Hooli"}, {"amount": "3.038,05", "state": "Alaska  ", "unit": "Cm", "country": "aze", "status": "l", "company": "Vehement"}, {"amount": "6.934,86", "state": "   Alaska", "unit": "cm   ", "country": "Japan", "status": "low", "company": "   Initech"}, {"amount": "TBD", "state": "gu", "unit": "  Cm", "country": "Azerbaijan", "status": "   p3", "company": "   Acme Inc"}, {"amount": "5.346,66", "state": "Guam", "unit": "  cm", "country": "republic of azerbaijan", "status": "high", "company": "Wonka Co"}, {"amount": "$3,227.14", "state": "GU", "unit": "in", "country": "Jpn", "status": "l", "company": "Globex"}, {"amount": "1.616,12", "state": "AK", "unit": "centimeters", "country": "azerbaijan", "status": "P1", "company": "   Stark Industries"}, {"amount": "1.750,23", "state": "  GUAM", "unit": "INCHES", "country": "   JPN", "status": "P3", "company": "  Globex"}, {"amount": "2.129,56", "state": "Alaska", "unit": "inches", "country": "AZ", "status": "H", "company": "Soylent Corp "}, {"amount": "$3,100.07", "state": "AK", "unit": "centimeters", "country": " JP ", "status": "Hih", "company": " Globex"}, {"amount": "1.356,42", "state": "GUAM", "unit": "Cm", "country": "jp", "status": "low", "company": "Cyberdyne"}, {"amount": "$7,507.21", "state": "Guam ", "unit": "centimeters", "country": "JPN", "status": "high", "company": "Initech"}, {"amount": "$986.82", "state": "Guam ", "unit": "\"", "country": "jp", "status": "HIGH", "company": "  Hooli"}, {"amount": "5.071,36", "state": "guam", "unit": "inch", "country": "japan", "status": "Low", "company": "Acme Inc"}, {"amount": "$5,693.78", "state": "Alaska", "unit": " \"", "country": "Repulbic  of Azerbaijan", "status": "P3", "company": "  Initech"}, {"amount": "2.792,40", "state": "Alaska", "unit": "Inch", "country": "Japan", "status": "P3", "company": "Umbrella   "}, {"amount": "$6,944.94", "state": "  Guam", "unit": "cm", "country": "Azerbaijan", "status": "high", "company": "Vehement"}, {"amount": "2.919,93", "state": "guam", "unit": "Cm", "country": "AZE", "status": "low", "company": "Vehement"}, {"amount": "3.473,10", "state": "  AK", "unit": "IN", "country": "Jp", "status": " high", "company": "Hooli"}, {"amount": "6.748,33", "state": "GU", "unit": "cm ", "country": "JPN", "status": "  h", "company": "Umbrella"}, {"amount": "$476.84", "state": "Alaska", "unit": "inch", "country": "japan", "status": "High", "company": "Initech"}, {"amount": "3.906,27", "state": "Alaska", "unit": "centimeter  ", "country": "jpn", "status": "P1", "company": "Stark Industries"}, {"amount": "na", "state": "Ak", "unit": "cm", "country": "Republic of Azerbaijan", "status": "high", "company": "Acme Inc"}, {"amount": "6.114,92", "state": "AK", "unit": "Cm", "country": "requblic of azerbaijan", "status": "L", "company": "Soylent Corp"}, {"amount": "$6,717.65", "state": "   Guam", "unit": "Cm", "country": "AZ", "status": " Low ", "company": "  Cyberdyne"}, {"amount": "$8,292.37", "state": "guam", "unit": " cm ", "country": "Azerbaijan", "status": "Low", "company": "Soylent  Corp"}, {"amount": "8.108,45", "state": "ak", "unit": " centimeter ", "country": "   AZE", "status": "LOW", "company": "Cyberdyne"}, {"amount": "$1,057.92", "state": "guam   ", "unit": "inch", "country": " jpn ", "status": "LOW", "company": "Soylent Corp"}, {"amount": "1.037,65", "state": "alaska", "unit": "CM ", "country": "AZ", "status": "P1", "company": "Acme Inc"}], "clean": [{"amount": 7413.09, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Umbrella"}, {"amount": 1380.29, "state": "Guam", "unit": "in", "country": "Azerbaijan", "status": "High", "company": "Globex"}, {"amount": 3517.84, "state": "Alaska", "unit": "in", "country": "Azerbaijan", "status": "High", "company": "Acme Inc"}, {"amount": 296.36, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Wonka Co"}, {"amount": 2449.67, "state": "Alaska", "unit": "in", "country": "Azerbaijan", "status": "Low", "company": "Wonka Co"}, {"amount": 5371.07, "state": "Guam", "unit": "in", "country": "Japan", "status": "Low", "company": "Soylent Corp"}, {"amount": 5812.2, "state": "Guam", "unit": "in", "country": "Azerbaijan", "status": "Low", "company": "Wonka Co"}, {"amount": 3274.22, "state": "Guam", "unit": "in", "country": "Azerbaijan", "status": "Low", "company": "Wonka Co"}, {"amount": 2464.96, "state": "Alaska", "unit": "cm", "country": "Japan", "status": "High", "company": "Initech"}, {"amount": 1833.34, "state": "Alaska", "unit": "cm", "country": "Japan", "status": "High", "company": "Hooli"}, {"amount": 3038.05, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Vehement"}, {"amount": 6934.86, "state": "Alaska", "unit": "cm", "country": "Japan", "status": "Low", "company": "Initech"}, {"amount": NaN, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Acme Inc"}, {"amount": 5346.66, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Wonka Co"}, {"amount": 3227.14, "state": "Guam", "unit": "in", "country": "Japan", "status": "Low", "company": "Globex"}, {"amount": 1616.12, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Stark Industries"}, {"amount": 1750.23, "state": "Guam", "unit": "in", "country": "Japan", "status": "Low", "company": "Globex"}, {"amount": 2129.56, "state": "Alaska", "unit": "in", "country": "Azerbaijan", "status": "High", "company": "Soylent Corp"}, {"amount": 3100.07, "state": "Alaska", "unit": "cm", "country": "Japan", "status": "High", "company": "Globex"}, {"amount": 1356.42, "state": "Guam", "unit": "cm", "country": "Japan", "status": "Low", "company": "Cyberdyne"}, {"amount": 7507.21, "state": "Guam", "unit": "cm", "country": "Japan", "status": "High", "company": "Initech"}, {"amount": 986.82, "state": "Guam", "unit": "in", "country": "Japan", "status": "High", "company": "Hooli"}, {"amount": 5071.36, "state": "Guam", "unit": "in", "country": "Japan", "status": "Low", "company": "Acme Inc"}, {"amount": 5693.78, "state": "Alaska", "unit": "in", "country": "Azerbaijan", "status": "Low", "company": "Initech"}, {"amount": 2792.4, "state": "Alaska", "unit": "in", "country": "Japan", "status": "Low", "company": "Umbrella"}, {"amount": 6944.94, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Vehement"}, {"amount": 2919.93, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Vehement"}, {"amount": 3473.1, "state": "Alaska", "unit": "in", "country": "Japan", "status": "High", "company": "Hooli"}, {"amount": 6748.33, "state": "Guam", "unit": "cm", "country": "Japan", "status": "High", "company": "Umbrella"}, {"amount": 476.84, "state": "Alaska", "unit": "in", "country": "Japan", "status": "High", "company": "Initech"}, {"amount": 3906.27, "state": "Alaska", "unit": "cm", "country": "Japan", "status": "High", "company": "Stark Industries"}, {"amount": NaN, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Acme Inc"}, {"amount": 6114.92, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Soylent Corp"}, {"amount": 6717.65, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Cyberdyne"}, {"amount": 8292.37, "state": "Guam", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Soylent Corp"}, {"amount": 8108.45, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "Low", "company": "Cyberdyne"}, {"amount": 1057.92, "state": "Guam", "unit": "in", "country": "Japan", "status": "Low", "company": "Soylent Corp"}, {"amount": 1037.65, "state": "Alaska", "unit": "cm", "country": "Azerbaijan", "status": "High", "company": "Acme Inc"}], "dirty_cols": ["amount", "state", "unit", "country", "status", "company"], "clean_cols": ["amount", "state", "unit", "country", "status", "company"], "plan": {"dataset_summary": "38 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GUAM": "Guam", "ak": "Alaska", "GU": "Guam", "gu": "Guam", "AK": "Alaska", "guam": "Guam", "Ak": "Alaska", "alaska": "Alaska"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cm": "cm", "inches": "in", "centimeter": "cm", "INCHES": "in", "\"": "in", "centimeters": "cm", "inch": "in", "Inch": "in", "IN": "in", "CM": "cm"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AZE": "Azerbaijan", "az": "Azerbaijan", "Republic Of Azerbaijan": "Azerbaijan", "azerbaijan": "Azerbaijan", "AZ": "Azerbaijan", "JAPAN": "Japan", "aze": "Azerbaijan", "jpn": "Japan", "republic of azerbaijan": "Azerbaijan", "Jpn": "Japan", "JPN": "Japan", "JP": "Japan", "jp": "Japan", "japan": "Japan", "Repulbic of Azerbaijan": "Azerbaijan", "Jp": "Japan", "Republic of Azerbaijan": "Azerbaijan", "requblic of azerbaijan": "Azerbaijan"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HIGH": "High", "high": "High", "low": "Low", "L": "Low", "p3": "Low", "l": "Low", "P1": "High", "P3": "Low", "H": "High", "Hih": "High", "h": "High", "LOW": "Low"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"state": "WV ", "status": "New", "unit": "mi", "name": " Mona Fischer", "signup_date": "2023-08-27"}, {"state": "ga", "status": "Closed LLost", "unit": "   Kg", "name": " Karl Johnson", "signup_date": "2023-02-13"}, {"state": "georgia", "status": "open", "unit": "kilogram", "name": "Judy Brandt", "signup_date": "6/21/2023"}, {"state": "Virgiia", "status": "Lost", "unit": "kg", "name": "Carol Smith", "signup_date": "7/18/2023"}, {"state": " virginia", "status": "Open", "unit": "Mi", "name": "Mona  Reyes", "signup_date": "2023-10-25"}, {"state": "  Vigrinia", "status": "OPEN", "unit": "MI", "name": "Priya Fischer", "signup_date": "2023-06-08"}, {"state": "georgia ", "status": "Open", "unit": "Mi", "name": "Sara Johnson   ", "signup_date": "2023-03-16"}, {"state": "GA", "status": "lost", "unit": "KG", "name": "Ivan Johnson   ", "signup_date": "6/23/2023"}, {"state": "West  Viryinia", "status": "Lost", "unit": "Mi  ", "name": "Alice Wong", "signup_date": "2023-07-12"}, {"state": "GA", "status": "Lost", "unit": "mipe", "name": "Priya Cruz", "signup_date": "11/12/2023"}, {"state": "Wevt  Virginia", "status": " new", "unit": " mile", "name": "Mona Petrov   ", "signup_date": "13 Mar 2023"}, {"state": " virginia ", "status": "closed lost", "unit": " miles ", "name": "  Grace Fischer", "signup_date": "2023-05-08"}, {"state": "", "status": "", "unit": "", "name": "", "signup_date": ""}], "clean": [{"state": "West Virginia", "status": "Open", "unit": "mi", "name": "Mona Fischer", "signup_date": "2023-08-27"}, {"state": "Georgia", "status": "Lost", "unit": "kg", "name": "Karl Johnson", "signup_date": "2023-02-13"}, {"state": "Georgia", "status": "Open", "unit": "kg", "name": "Judy Brandt", "signup_date": "2023-06-21"}, {"state": "Virginia", "status": "Lost", "unit": "kg", "name": "Carol Smith", "signup_date": "2023-07-18"}, {"state": "Virginia", "status": "Open", "unit": "mi", "name": "Mona Reyes", "signup_date": "2023-10-25"}, {"state": "Virginia", "status": "Open", "unit": "mi", "name": "Priya Fischer", "signup_date": "2023-06-08"}, {"state": "Georgia", "status": "Open", "unit": "mi", "name": "Sara Johnson", "signup_date": "2023-03-16"}, {"state": "Georgia", "status": "Lost", "unit": "kg", "name": "Ivan Johnson", "signup_date": "2023-06-23"}, {"state": "West Virginia", "status": "Lost", "unit": "mi", "name": "Alice Wong", "signup_date": "2023-07-12"}, {"state": "Georgia", "status": "Lost", "unit": "mi", "name": "Priya Cruz", "signup_date": "2023-11-12"}, {"state": "West Virginia", "status": "Open", "unit": "mi", "name": "Mona Petrov", "signup_date": "2023-03-13"}, {"state": "Virginia", "status": "Lost", "unit": "mi", "name": "Grace Fischer", "signup_date": "2023-05-08"}], "dirty_cols": ["state", "status", "unit", "name", "signup_date"], "clean_cols": ["state", "status", "unit", "name", "signup_date"], "plan": {"dataset_summary": "13 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WV": "West Virginia", "ga": "Georgia", "georgia": "Georgia", "Virgiia": "Virginia", "virginia": "Virginia", "Vigrinia": "Virginia", "GA": "Georgia", "West Viryinia": "West Virginia", "Wevt Virginia": "West Virginia"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"New": "Open", "Closed LLost": "Lost", "open": "Open", "OPEN": "Open", "lost": "Lost", "new": "Open", "closed lost": "Lost"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Kg": "kg", "kilogram": "kg", "Mi": "mi", "MI": "mi", "KG": "kg", "mipe": "mi", "mile": "mi", "miles": "mi"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"currency": "Swiss Franc", "city": "Monterotondo", "signup_date": "45216", "job_title": "CTO", "phone": "575-994-0441", "notes2": ""}, {"currency": "ETB", "city": "san diego", "signup_date": "2/16/2023", "job_title": "c.t.o.", "phone": "5952215005", "notes2": ""}, {"currency": "ETHIOPIAN BIRR", "city": "monterotondo", "signup_date": "45244", "job_title": "senior engineer", "phone": "5711548857", "notes2": ""}, {"currency": "Etb", "city": " MONTEROTONDDO ", "signup_date": "2023-08-12", "job_title": "Sr  Engineer", "phone": "576-725-9453", "notes2": ""}, {"currency": "Ethioipan Birr", "city": "Bukit Tambun", "signup_date": "5/16/2023", "job_title": "chief techlology officer", "phone": "558.059.1930", "notes2": ""}, {"currency": "etb", "city": "monterotondo", "signup_date": "45177", "job_title": "sr engineer", "phone": "5637632517", "notes2": ""}, {"currency": "CHF", "city": "monterotondo", "signup_date": "44946", "job_title": " Sr Engineer", "phone": "(541)899-8178", "notes2": ""}, {"currency": "  swiss franc", "city": "Bukit Tambun  ", "signup_date": "45050", "job_title": "Sr EEngineer", "phone": "574-790-5333", "notes2": ""}, {"currency": "ETB", "city": "San Diego ", "signup_date": "2023-09-21", "job_title": "   CTO", "phone": "5607941793", "notes2": ""}, {"currency": " ETB ", "city": "Monterotoondo", "signup_date": "25 Oct 2023", "job_title": "c.t.o.", "phone": "(546)241-8973", "notes2": ""}, {"currency": "Swiss  Franc", "city": "San Diego  ", "signup_date": "3 Mar 2023", "job_title": "chief technology officer", "phone": "(598)964-5367", "notes2": ""}, {"currency": "chf", "city": "  Bukit Tambun", "signup_date": "10/18/2023", "job_title": "Snr  Engineer", "phone": "(508)499-9843", "notes2": ""}, {"currency": "Ethiopian Birr", "city": "monterotondo", "signup_date": "1/15/2023", "job_title": "Sr. Engineer", "phone": "(589)504-1043", "notes2": ""}, {"currency": "swiss frac   ", "city": "Montjrotondo", "signup_date": "1 May 2023", "job_title": "SR. ENGINEER", "phone": "574.531.6430", "notes2": ""}, {"currency": "Swiss Franc", "city": "bukit tambun", "signup_date": "45033", "job_title": "Chief Technology fficer", "phone": "505.001.9612", "notes2": ""}, {"currency": "etb", "city": "bukit tambun", "signup_date": "45118", "job_title": "cto", "phone": "5565981051", "notes2": ""}, {"currency": " ETB ", "city": "Monterotondo", "signup_date": "2023-02-28", "job_title": "Sr Engineer", "phone": "552.461.6332", "notes2": ""}, {"currency": "Swiss Franc   ", "city": "bukit tambun", "signup_date": "28 Oct 2023", "job_title": "sr engineer", "phone": "(552)244-6568", "notes2": ""}, {"currency": "ethiopian birr", "city": "San Miego", "signup_date": "26 Jun 2023", "job_title": "cto", "phone": "(506)991-2914", "notes2": ""}, {"currency": "ETHIOPIAN BIRR", "city": "Bukit Tambun", "signup_date": "45181", "job_title": "chief tehcnology officer", "phone": "543.037.8742", "notes2": ""}, {"currency": "CHF", "city": "MONTEROTONDO", "signup_date": "11/27/2023", "job_title": "cto", "phone": "5629930801", "notes2": ""}, {"currency": "Ethiopian Birr", "city": "Bukit ambun", "signup_date": "2023-03-05", "job_title": "Chief Technology Officer", "phone": "5329369361", "notes2": ""}, {"currency": "ethiopian birr", "city": "San  Diego", "signup_date": "9 Mar 2023", "job_title": "Senior Eng", "phone": "(509)136-5170", "notes2": ""}, {"currency": "   Ethiopian Birr", "city": "Bukit Tambun", "signup_date": "11/8/2023", "job_title": "Sr. Engineer", "phone": "525-256-7447", "notes2": ""}, {"currency": "CHF", "city": "Bkit Tambun", "signup_date": "45056", "job_title": "Chief Tehnology Officer", "phone": "5495889788", "notes2": ""}, {"currency": "ethiopain birr", "city": "Bukit Tambun", "signup_date": "45041", "job_title": "c.t.o.", "phone": "(586)153-2502", "notes2": ""}, {"currency": " ETB ", "city": "BUKIT TAMBUN", "signup_date": "45040", "job_title": "CTO", "phone": "585.455.2948", "notes2": ""}, {"currency": "CHF", "city": "  Monterotondo", "signup_date": "45282", "job_title": "Sr. Enginer", "phone": "(526)837-1717", "notes2": ""}, {"currency": "etb", "city": "SAN DIEGO", "signup_date": "2023-05-10", "job_title": "Chief Technology Officer", "phone": "543-044-2384", "notes2": ""}, {"currency": "ETHIOPIAN BIRR", "city": "San Digeo", "signup_date": "13 Aug 2023", "job_title": "SENIOR ENG", "phone": "(518)239-3882", "notes2": ""}, {"currency": "Chf", "city": "Monterotondo", "signup_date": "3 Sep 2023", "job_title": "Sr Engineer", "phone": "5373429990", "notes2": ""}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "3/4/2023", "job_title": "C.T.O.", "phone": "583.672.0169", "notes2": ""}, {"currency": "Swiss Franc ", "city": "Bukit Tabun", "signup_date": "11/7/2023", "job_title": "C.T.O.  ", "phone": "(590)424-7338", "notes2": ""}, {"currency": "Ethiopian Birr", "city": "San Diego", "signup_date": "44927", "job_title": "cto", "phone": "585-235-8727", "notes2": ""}, {"currency": "ethiopian birr", "city": "Bukit Tambun", "signup_date": "8/25/2023", "job_title": "chief technology officer", "phone": "557.262.7978", "notes2": ""}, {"currency": "Ethiopian Birr", "city": "   Bukit Tambun", "signup_date": "44964", "job_title": "cto", "phone": "508-779-4877", "notes2": ""}, {"currency": "ETHIOPIAN BIRR", "city": "  Monterotondo", "signup_date": "2023-07-09", "job_title": "cto", "phone": "(567)160-9021", "notes2": ""}, {"currency": "Ethiopian Birr", "city": "Bukit Tambun", "signup_date": "10/16/2023", "job_title": "Chief Technology Officer", "phone": "559-576-5114", "notes2": ""}, {"currency": "Ethiopian Birr", "city": "Bukt Tambun", "signup_date": "45094", "job_title": "C.T.O.", "phone": "578-882-0957", "notes2": ""}], "clean": [{"currency": "CHF", "city": "Monterotondo", "signup_date": "2023-10-17", "job_title": "Chief Technology Officer", "phone": "(575) 994-0441"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-02-16", "job_title": "Chief Technology Officer", "phone": "(595) 221-5005"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-11-14", "job_title": "Senior Engineer", "phone": "(571) 154-8857"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-08-12", "job_title": "Senior Engineer", "phone": "(576) 725-9453"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-05-16", "job_title": "Chief Technology Officer", "phone": "(558) 059-1930"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-09-08", "job_title": "Senior Engineer", "phone": "(563) 763-2517"}, {"currency": "CHF", "city": "Monterotondo", "signup_date": "2023-01-20", "job_title": "Senior Engineer", "phone": "(541) 899-8178"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-05-04", "job_title": "Senior Engineer", "phone": "(574) 790-5333"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-09-21", "job_title": "Chief Technology Officer", "phone": "(560) 794-1793"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-10-25", "job_title": "Chief Technology Officer", "phone": "(546) 241-8973"}, {"currency": "CHF", "city": "San Diego", "signup_date": "2023-03-03", "job_title": "Chief Technology Officer", "phone": "(598) 964-5367"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-10-18", "job_title": "Senior Engineer", "phone": "(508) 499-9843"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-01-15", "job_title": "Senior Engineer", "phone": "(589) 504-1043"}, {"currency": "CHF", "city": "Monterotondo", "signup_date": "2023-05-01", "job_title": "Senior Engineer", "phone": "(574) 531-6430"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-04-17", "job_title": "Chief Technology Officer", "phone": "(505) 001-9612"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-07-11", "job_title": "Chief Technology Officer", "phone": "(556) 598-1051"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-02-28", "job_title": "Senior Engineer", "phone": "(552) 461-6332"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-10-28", "job_title": "Senior Engineer", "phone": "(552) 244-6568"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-06-26", "job_title": "Chief Technology Officer", "phone": "(506) 991-2914"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-09-12", "job_title": "Chief Technology Officer", "phone": "(543) 037-8742"}, {"currency": "CHF", "city": "Monterotondo", "signup_date": "2023-11-27", "job_title": "Chief Technology Officer", "phone": "(562) 993-0801"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-03-05", "job_title": "Chief Technology Officer", "phone": "(532) 936-9361"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-03-09", "job_title": "Senior Engineer", "phone": "(509) 136-5170"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-11-08", "job_title": "Senior Engineer", "phone": "(525) 256-7447"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-05-10", "job_title": "Chief Technology Officer", "phone": "(549) 588-9788"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-04-25", "job_title": "Chief Technology Officer", "phone": "(586) 153-2502"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-04-24", "job_title": "Chief Technology Officer", "phone": "(585) 455-2948"}, {"currency": "CHF", "city": "Monterotondo", "signup_date": "2023-12-22", "job_title": "Senior Engineer", "phone": "(526) 837-1717"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-05-10", "job_title": "Chief Technology Officer", "phone": "(543) 044-2384"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-08-13", "job_title": "Senior Engineer", "phone": "(518) 239-3882"}, {"currency": "CHF", "city": "Monterotondo", "signup_date": "2023-09-03", "job_title": "Senior Engineer", "phone": "(537) 342-9990"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-03-04", "job_title": "Chief Technology Officer", "phone": "(583) 672-0169"}, {"currency": "CHF", "city": "Bukit Tambun", "signup_date": "2023-11-07", "job_title": "Chief Technology Officer", "phone": "(590) 424-7338"}, {"currency": "ETB", "city": "San Diego", "signup_date": "2023-01-01", "job_title": "Chief Technology Officer", "phone": "(585) 235-8727"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-08-25", "job_title": "Chief Technology Officer", "phone": "(557) 262-7978"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-02-07", "job_title": "Chief Technology Officer", "phone": "(508) 779-4877"}, {"currency": "ETB", "city": "Monterotondo", "signup_date": "2023-07-09", "job_title": "Chief Technology Officer", "phone": "(567) 160-9021"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-10-16", "job_title": "Chief Technology Officer", "phone": "(559) 576-5114"}, {"currency": "ETB", "city": "Bukit Tambun", "signup_date": "2023-06-17", "job_title": "Chief Technology Officer", "phone": "(578) 882-0957"}], "dirty_cols": ["currency", "city", "signup_date", "job_title", "phone", "notes2"], "clean_cols": ["currency", "city", "signup_date", "job_title", "phone"], "plan": {"dataset_summary": "39 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Swiss Franc": "CHF", "ETHIOPIAN BIRR": "ETB", "Etb": "ETB", "Ethioipan Birr": "ETB", "etb": "ETB", "swiss franc": "CHF", "chf": "CHF", "Ethiopian Birr": "ETB", "swiss frac": "CHF", "ethiopian birr": "ETB", "ethiopain birr": "ETB", "Chf": "CHF"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"san diego": "San Diego", "monterotondo": "Monterotondo", "MONTEROTONDDO": "Monterotondo", "Monterotoondo": "Monterotondo", "Montjrotondo": "Monterotondo", "bukit tambun": "Bukit Tambun", "San Miego": "San Diego", "MONTEROTONDO": "Monterotondo", "Bukit ambun": "Bukit Tambun", "Bkit Tambun": "Bukit Tambun", "BUKIT TAMBUN": "Bukit Tambun", "SAN DIEGO": "San Diego", "San Digeo": "San Diego", "Bukit Tabun": "Bukit Tambun", "Bukt Tambun": "Bukit Tambun"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CTO": "Chief Technology Officer", "c.t.o.": "Chief Technology Officer", "senior engineer": "Senior Engineer", "Sr Engineer": "Senior Engineer", "chief techlology officer": "Chief Technology Officer", "sr engineer": "Senior Engineer", "Sr EEngineer": "Senior Engineer", "chief technology officer": "Chief Technology Officer", "Snr Engineer": "Senior Engineer", "Sr. Engineer": "Senior Engineer", "SR. ENGINEER": "Senior Engineer", "Chief Technology fficer": "Chief Technology Officer", "cto": "Chief Technology Officer", "chief tehcnology officer": "Chief Technology Officer", "Senior Eng": "Senior Engineer", "Chief Tehnology Officer": "Chief Technology Officer", "Sr. Enginer": "Senior Engineer", "SENIOR ENG": "Senior Engineer", "C.T.O.": "Chief Technology Officer"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"currency": "Bahraini  Dinar", "industry": "  Health Care", "rate": "63.3%", "extra": ""}, {"currency": "BHD", "industry": "   EDUCATOIN", "rate": "46.0%", "extra": ""}, {"currency": "unidad de valor real", "industry": "realestate", "rate": "75.2%", "extra": ""}, {"currency": "bahraini dinar", "industry": "Health Care", "rate": "20.4%", "extra": ""}, {"currency": "BHD", "industry": "Health", "rate": "43.2%", "extra": ""}, {"currency": "Cou", "industry": "Healthcare", "rate": "41.6%", "extra": ""}, {"currency": " COU", "industry": "Education", "rate": "39.3%", "extra": ""}, {"currency": "  unidad de valor real", "industry": "EDUCATION", "rate": "40.6%", "extra": ""}, {"currency": "unidad de valor real  ", "industry": "Health Care", "rate": "76.3%", "extra": ""}, {"currency": "Bahrain iDinar ", "industry": "Educatoin", "rate": "3.3%", "extra": ""}, {"currency": "bhd", "industry": "Health", "rate": "49.4%", "extra": ""}, {"currency": "BAHRAINI DINAR", "industry": " education", "rate": "95.8%", "extra": ""}], "clean": [{"currency": "BHD", "industry": "Healthcare", "rate": 0.633}, {"currency": "BHD", "industry": "Education", "rate": 0.46}, {"currency": "COU", "industry": "Real Estate", "rate": 0.752}, {"currency": "BHD", "industry": "Healthcare", "rate": 0.204}, {"currency": "BHD", "industry": "Healthcare", "rate": 0.43200000000000005}, {"currency": "COU", "industry": "Healthcare", "rate": 0.41600000000000004}, {"currency": "COU", "industry": "Education", "rate": 0.39299999999999996}, {"currency": "COU", "industry": "Education", "rate": 0.406}, {"currency": "COU", "industry": "Healthcare", "rate": 0.763}, {"currency": "BHD", "industry": "Education", "rate": 0.033}, {"currency": "BHD", "industry": "Healthcare", "rate": 0.494}, {"currency": "BHD", "industry": "Education", "rate": 0.958}], "dirty_cols": ["currency", "industry", "rate", "extra"], "clean_cols": ["currency", "industry", "rate"], "plan": {"dataset_summary": "12 rows × 4 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Bahraini Dinar": "BHD", "unidad de valor real": "COU", "bahraini dinar": "BHD", "Cou": "COU", "Bahrain iDinar": "BHD", "bhd": "BHD", "BAHRAINI DINAR": "BHD"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Health Care": "Healthcare", "EDUCATOIN": "Education", "realestate": "Real Estate", "Health": "Healthcare", "EDUCATION": "Education", "Educatoin": "Education", "education": "Education"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"amount": "7.499,05", "rate": "25.1%", "currency": "Beermudian Dollar", "signup_date": "2023-05-25", "phone": "(583)545-4430", "company": "Stark  Industries"}, {"amount": "4.911,12", "rate": "6.1%", "currency": "lempira", "signup_date": "2023-12-18", "phone": "5501509937", "company": "Hooli   "}, {"amount": "$3,415.53", "rate": "67.8%", "currency": "Lempira", "signup_date": "45050", "phone": "567-367-7740", "company": "  Vehement"}, {"amount": "$7,498.37", "rate": "18.1%", "currency": "lempira", "signup_date": "2023-07-06", "phone": "576.733.6116", "company": "Soylent Corp"}, {"amount": "$2,887.15", "rate": "64.2%", "currency": " hnl ", "signup_date": "2023-04-17", "phone": "546.895.6314", "company": "Cyberdyne"}, {"amount": "TBD", "rate": "64.3%", "currency": "Ars  ", "signup_date": "2023-03-09", "phone": "(579)545-9618", "company": "Cyberdyne  "}, {"amount": "8.499,18", "rate": "11.1%", "currency": " HNL ", "signup_date": "6 Aug 2023", "phone": "500-170-2439", "company": "Cyberdyne"}, {"amount": "$5,927.10", "rate": "57.1%", "currency": " Lempia ", "signup_date": "45063", "phone": "5401765024", "company": "Globex"}, {"amount": "1.489,13", "rate": "63.0%", "currency": " Argentine Peso", "signup_date": "26 Aug 2023", "phone": "5935955717", "company": "Stark  Industries"}, {"amount": "$1,070.20", "rate": "97.9%", "currency": "LEMPIRA", "signup_date": "22 Sep 2023", "phone": "583-179-5813", "company": "Globex   "}, {"amount": "$8,862.86", "rate": "92.6%", "currency": "hnl", "signup_date": "45245", "phone": "5946341468", "company": "Wonka  Co"}, {"amount": "4.671,17", "rate": "20.1%", "currency": "Argentine Peso", "signup_date": "23 Sep 2023", "phone": "543-625-1062", "company": "Globex "}, {"amount": "4.736,06", "rate": "3.5%", "currency": "Argentine Peso", "signup_date": "16 Apr 2023", "phone": "5903968155", "company": " Hooli "}, {"amount": "$1,356.50", "rate": "50.5%", "currency": "BMD", "signup_date": "45029", "phone": "5734894961", "company": " Acme Inc"}, {"amount": "None", "rate": "65.8%", "currency": "Argentine Peso", "signup_date": "21 Jun 2023", "phone": "5330768437", "company": "   Soylent Corp"}, {"amount": "N/A", "rate": "9.5%", "currency": " ars", "signup_date": "18 Nov 2023", "phone": "521.687.4572", "company": "Stark Industries"}, {"amount": "$911.59", "rate": "93.6%", "currency": "Lempira", "signup_date": "20 Jan 2023", "phone": "(512)095-5274", "company": "Wonka Co"}], "clean": [{"amount": 7499.05, "rate": 0.251, "currency": "BMD", "signup_date": "2023-05-25", "phone": "(583) 545-4430", "company": "Stark Industries"}, {"amount": 4911.12, "rate": 0.061, "currency": "HNL", "signup_date": "2023-12-18", "phone": "(550) 150-9937", "company": "Hooli"}, {"amount": 3415.53, "rate": 0.6779999999999999, "currency": "HNL", "signup_date": "2023-05-04", "phone": "(567) 367-7740", "company": "Vehement"}, {"amount": 7498.37, "rate": 0.18100000000000002, "currency": "HNL", "signup_date": "2023-07-06", "phone": "(576) 733-6116", "company": "Soylent Corp"}, {"amount": 2887.15, "rate": 0.642, "currency": "HNL", "signup_date": "2023-04-17", "phone": "(546) 895-6314", "company": "Cyberdyne"}, {"amount": NaN, "rate": 0.643, "currency": "ARS", "signup_date": "2023-03-09", "phone": "(579) 545-9618", "company": "Cyberdyne"}, {"amount": 8499.18, "rate": 0.111, "currency": "HNL", "signup_date": "2023-08-06", "phone": "(500) 170-2439", "company": "Cyberdyne"}, {"amount": 5927.1, "rate": 0.5710000000000001, "currency": "HNL", "signup_date": "2023-05-17", "phone": "(540) 176-5024", "company": "Globex"}, {"amount": 1489.13, "rate": 0.63, "currency": "ARS", "signup_date": "2023-08-26", "phone": "(593) 595-5717", "company": "Stark Industries"}, {"amount": 1070.2, "rate": 0.9790000000000001, "currency": "HNL", "signup_date": "2023-09-22", "phone": "(583) 179-5813", "company": "Globex"}, {"amount": 8862.86, "rate": 0.9259999999999999, "currency": "HNL", "signup_date": "2023-11-15", "phone": "(594) 634-1468", "company": "Wonka Co"}, {"amount": 4671.17, "rate": 0.201, "currency": "ARS", "signup_date": "2023-09-23", "phone": "(543) 625-1062", "company": "Globex"}, {"amount": 4736.06, "rate": 0.035, "currency": "ARS", "signup_date": "2023-04-16", "phone": "(590) 396-8155", "company": "Hooli"}, {"amount": 1356.5, "rate": 0.505, "currency": "BMD", "signup_date": "2023-04-13", "phone": "(573) 489-4961", "company": "Acme Inc"}, {"amount": NaN, "rate": 0.6579999999999999, "currency": "ARS", "signup_date": "2023-06-21", "phone": "(533) 076-8437", "company": "Soylent Corp"}, {"amount": NaN, "rate": 0.095, "currency": "ARS", "signup_date": "2023-11-18", "phone": "(521) 687-4572", "company": "Stark Industries"}, {"amount": 911.59, "rate": 0.9359999999999999, "currency": "HNL", "signup_date": "2023-01-20", "phone": "(512) 095-5274", "company": "Wonka Co"}], "dirty_cols": ["amount", "rate", "currency", "signup_date", "phone", "company"], "clean_cols": ["amount", "rate", "currency", "signup_date", "phone", "company"], "plan": {"dataset_summary": "17 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Beermudian Dollar": "BMD", "lempira": "HNL", "Lempira": "HNL", "hnl": "HNL", "Ars": "ARS", "Lempia": "HNL", "Argentine Peso": "ARS", "LEMPIRA": "HNL", "ars": "ARS"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"state": "RI  ", "country": "Korea,  Deocratic People's Republic of", "signup_date": "45256", "currency": "Bhd", "city": "church-yonge corridor  ", "unit": " Pounds ", "col_x": ""}, {"state": "pr", "country": "Democratic Peoples  Republic of Korea", "signup_date": "3/2/2023", "currency": "Swiss  Franc", "city": "Harlingen", "unit": "  cm", "col_x": ""}, {"state": "Pr", "country": "FIJI   ", "signup_date": "2023-05-20", "currency": "CHF", "city": "Triprangod", "unit": " cenhimeters ", "col_x": ""}, {"state": "PR", "country": "FJ", "signup_date": "17 Dec 2023", "currency": "bhd", "city": "Church-Yonge Corridor", "unit": "LB", "col_x": ""}, {"state": "PR", "country": "Isle of Man", "signup_date": "2023-04-12", "currency": "Bahraini Dinar", "city": " Draguignan ", "unit": "cm", "col_x": ""}, {"state": "Puerto  Rico", "country": "denmark", "signup_date": "2023-12-16", "currency": "Aed", "city": "  Triprangod", "unit": "inch", "col_x": ""}, {"state": "PR", "country": "Korea, Dmocratic People's Republic of", "signup_date": "4/15/2023", "currency": "Bhd", "city": "Harlingen", "unit": "pounds", "col_x": ""}, {"state": "  puerto rico", "country": " IMN", "signup_date": "20 Apr 2023", "currency": "BHD   ", "city": "CHURCH-YONGE CORRIDOR", "unit": "LB", "col_x": ""}, {"state": "rhode island", "country": "Fiji", "signup_date": "2023-11-11", "currency": "CHF", "city": "draguignan", "unit": "lb", "col_x": ""}, {"state": "PuertoiRico  ", "country": " Isle of Man", "signup_date": "15 Apr 2023", "currency": "UAE Dirham", "city": "draguignan", "unit": "Cm", "col_x": ""}, {"state": "Rhde Island", "country": "IM", "signup_date": "2023-05-06", "currency": " Chf", "city": "Harlingen", "unit": "lb ", "col_x": ""}, {"state": "Puerto Rico", "country": "Federative Republic of Brazil", "signup_date": "3/8/2023", "currency": "chf", "city": "   Triprangod", "unit": " lb ", "col_x": ""}, {"state": "RI  ", "country": "FJ", "signup_date": "44971", "currency": "Bahraini  Dinar", "city": "Harlingen", "unit": " inch", "col_x": ""}, {"state": "ri", "country": "DEMOCRATIC PEOPLE'S REPUBLIC OF KOREA", "signup_date": "7/19/2023", "currency": "bhd", "city": "Triprangod", "unit": "cm", "col_x": ""}, {"state": "puerto rico", "country": "IMN", "signup_date": "44934", "currency": "BHD", "city": "Harlingen", "unit": "inch", "col_x": ""}, {"state": "Puerto Rico  ", "country": "BRA", "signup_date": "1/6/2023", "currency": " Bhd ", "city": "Draguignan", "unit": "Cm", "col_x": ""}, {"state": "Puerto  Rico", "country": "denmark", "signup_date": "2023-12-16", "currency": "Aed", "city": "  Triprangod", "unit": "inch", "col_x": ""}, {"state": "RI  ", "country": "Korea,  Deocratic People's Republic of", "signup_date": "45256", "currency": "Bhd", "city": "church-yonge corridor  ", "unit": " Pounds ", "col_x": ""}, {"state": "", "country": "", "signup_date": "", "currency": "", "city": "", "unit": "", "col_x": ""}], "clean": [{"state": "Rhode Island", "country": "Korea, Democratic People's Republic of", "signup_date": "2023-11-26", "currency": "BHD", "city": "Church-Yonge Corridor", "unit": "lb"}, {"state": "Puerto Rico", "country": "Korea, Democratic People's Republic of", "signup_date": "2023-03-02", "currency": "CHF", "city": "Harlingen", "unit": "cm"}, {"state": "Puerto Rico", "country": "Fiji", "signup_date": "2023-05-20", "currency": "CHF", "city": "Triprangod", "unit": "cm"}, {"state": "Puerto Rico", "country": "Fiji", "signup_date": "2023-12-17", "currency": "BHD", "city": "Church-Yonge Corridor", "unit": "lb"}, {"state": "Puerto Rico", "country": "Isle of Man", "signup_date": "2023-04-12", "currency": "BHD", "city": "Draguignan", "unit": "cm"}, {"state": "Puerto Rico", "country": "Denmark", "signup_date": "2023-12-16", "currency": "AED", "city": "Triprangod", "unit": "in"}, {"state": "Puerto Rico", "country": "Korea, Democratic People's Republic of", "signup_date": "2023-04-15", "currency": "BHD", "city": "Harlingen", "unit": "lb"}, {"state": "Puerto Rico", "country": "Isle of Man", "signup_date": "2023-04-20", "currency": "BHD", "city": "Church-Yonge Corridor", "unit": "lb"}, {"state": "Rhode Island", "country": "Fiji", "signup_date": "2023-11-11", "currency": "CHF", "city": "Draguignan", "unit": "lb"}, {"state": "Puerto Rico", "country": "Isle of Man", "signup_date": "2023-04-15", "currency": "AED", "city": "Draguignan", "unit": "cm"}, {"state": "Rhode Island", "country": "Isle of Man", "signup_date": "2023-05-06", "currency": "CHF", "city": "Harlingen", "unit": "lb"}, {"state": "Puerto Rico", "country": "Brazil", "signup_date": "2023-03-08", "currency": "CHF", "city": "Triprangod", "unit": "lb"}, {"state": "Rhode Island", "country": "Fiji", "signup_date": "2023-02-14", "currency": "BHD", "city": "Harlingen", "unit": "in"}, {"state": "Rhode Island", "country": "Korea, Democratic People's Republic of", "signup_date": "2023-07-19", "currency": "BHD", "city": "Triprangod", "unit": "cm"}, {"state": "Puerto Rico", "country": "Isle of Man", "signup_date": "2023-01-08", "currency": "BHD", "city": "Harlingen", "unit": "in"}, {"state": "Puerto Rico", "country": "Brazil", "signup_date": "2023-01-06", "currency": "BHD", "city": "Draguignan", "unit": "cm"}], "dirty_cols": ["state", "country", "signup_date", "currency", "city", "unit", "col_x"], "clean_cols": ["state", "country", "signup_date", "currency", "city", "unit"], "plan": {"dataset_summary": "19 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"RI": "Rhode Island", "pr": "Puerto Rico", "Pr": "Puerto Rico", "PR": "Puerto Rico", "puerto rico": "Puerto Rico", "rhode island": "Rhode Island", "PuertoiRico": "Puerto Rico", "Rhde Island": "Rhode Island", "ri": "Rhode Island"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Korea, Deocratic People's Republic of": "Korea, Democratic People's Republic of", "Democratic Peoples Republic of Korea": "Korea, Democratic People's Republic of", "FIJI": "Fiji", "FJ": "Fiji", "denmark": "Denmark", "Korea, Dmocratic People's Republic of": "Korea, Democratic People's Republic of", "IMN": "Isle of Man", "IM": "Isle of Man", "Federative Republic of Brazil": "Brazil", "DEMOCRATIC PEOPLE'S REPUBLIC OF KOREA": "Korea, Democratic People's Republic of", "BRA": "Brazil"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Bhd": "BHD", "Swiss Franc": "CHF", "bhd": "BHD", "Bahraini Dinar": "BHD", "Aed": "AED", "UAE Dirham": "AED", "Chf": "CHF", "chf": "CHF"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"church-yonge corridor": "Church-Yonge Corridor", "CHURCH-YONGE CORRIDOR": "Church-Yonge Corridor", "draguignan": "Draguignan"}, "rationale": "Unified 3 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Pounds": "lb", "cenhimeters": "cm", "LB": "lb", "inch": "in", "pounds": "lb", "Cm": "cm"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "Bob Ali", "phone": "(543)611-6641", "country": "Federal Republic Of Germany", "job_title": "ENGINEERING MANAGER", "rate": "31.0%", "company": "Acme Inc", "industry": "Energy", "state": "WA"}, {"name": "   Heidi Brandt", "phone": "(518)656-7466", "country": "gm", "job_title": "accuunt executive", "rate": "13.4%", "company": "  Hooli", "industry": "Energy", "state": "NEVKDA"}, {"name": "Ivan Wong", "phone": "5864895089", "country": "Republic of the Gambia", "job_title": "Acctt Exec", "rate": "6.9%", "company": " Wonka Co", "industry": "binking", "state": "NV"}, {"name": "Bob Park", "phone": "558.876.1053", "country": "GAMBIA", "job_title": "Acct Exec", "rate": "46.0%", "company": "Soylent Corp", "industry": "Financial Services", "state": "NV"}, {"name": "Mona  Petrov", "phone": "5889677614", "country": "GM", "job_title": "Account Executive", "rate": "56.1%", "company": "Wonka  Co", "industry": " banking", "state": "delaware  "}, {"name": "Eve Fischer", "phone": "579.157.0016", "country": "deu", "job_title": "ae", "rate": "3.8%", "company": "Wonka  Co", "industry": "ENERYG", "state": "id"}, {"name": " Carol Smith", "phone": "556.022.6632", "country": "lc", "job_title": "account executive", "rate": "34.3%", "company": "Umbrella   ", "industry": "enregy", "state": "DE"}, {"name": "Eve Wong ", "phone": "5796145311", "country": "GERMANY", "job_title": "Account  Executive", "rate": "68.0%", "company": "Initech", "industry": "Financial Services", "state": "Delaware"}, {"name": "Sara  Reyes", "phone": "571.990.8461", "country": " TUR", "job_title": "acct exec", "rate": "47.0%", "company": "Globex  ", "industry": " Utilities ", "state": " NEVADA "}, {"name": "Carol Fischer", "phone": "5576957251", "country": "LC", "job_title": "Account Kxecutive", "rate": "89.1%", "company": "Initech", "industry": "Financial Services", "state": "Delaware"}, {"name": "Eve Wong", "phone": "(510)826-0154", "country": "  Türkiye", "job_title": "Engineering Manager", "rate": "37.4%", "company": "Stark Industries", "industry": "Energy  ", "state": "Idaho"}, {"name": "Ivan  Wong", "phone": "5421128555", "country": "DE   ", "job_title": "Acct Exce", "rate": "84.5%", "company": "Cyberdyne", "industry": "FINTECH", "state": " Washington "}, {"name": "Alice Wong", "phone": "(561)229-2655", "country": "Federal Republic of Germany", "job_title": "Engineering Mgr", "rate": "86.2%", "company": " Vehement ", "industry": "Oil & Gas", "state": "nevada"}, {"name": "Priya Adams  ", "phone": "5516107042", "country": " lca", "job_title": "Ae", "rate": "34.3%", "company": "Acme  Inc", "industry": "energy", "state": "Idaho"}, {"name": "Carol Smith", "phone": "5182996818", "country": " Germany ", "job_title": "ae", "rate": "80.9%", "company": "   Globex", "industry": "Finanec", "state": "Nevada"}, {"name": "Eve Moore", "phone": "5436744030", "country": "TUR   ", "job_title": "engineering mgr", "rate": "6.0%", "company": " Stark Industries", "industry": " Fintech", "state": "wa"}, {"name": "Mona Diaz", "phone": "5614814084", "country": "  federal republic of germany", "job_title": "Account Executive", "rate": "42.2%", "company": "Hooli", "industry": "Finance", "state": "washington"}, {"name": "Bob Adams", "phone": "537-810-2939", "country": "gm", "job_title": "Eng Mgr", "rate": "99.8%", "company": "Soylent Corp", "industry": "finance", "state": " Nevada "}, {"name": "Frank Khan", "phone": "569-216-6985", "country": "Tr", "job_title": "Eng Manaer", "rate": "59.0%", "company": " Acme Inc", "industry": "FinServ", "state": "Nevada"}, {"name": "   Heidi Johnson", "phone": "560-348-0761", "country": "GMB", "job_title": "eng gr", "rate": "43.4%", "company": "Vehement", "industry": " energy ", "state": "washington"}, {"name": "Omar Diaz", "phone": "552.330.1058", "country": "saint lcia", "job_title": "Account Executive", "rate": "37.3%", "company": "Stark Industries", "industry": "  Energy", "state": "washington"}, {"name": "Carol Park", "phone": "5197602783", "country": "saint lucia", "job_title": " AE", "rate": "58.8%", "company": "   Umbrella", "industry": "  Fintech", "state": "Idaho"}, {"name": "Bob Wong", "phone": "597.190.7678", "country": "Republic Of Türkiye", "job_title": "Eng Mgr", "rate": "73.5%", "company": " Globex ", "industry": "finserv", "state": "  Wa"}, {"name": "Heidi Lee", "phone": "(592)142-7628", "country": " Gambia", "job_title": "Eng Manager", "rate": "6.9%", "company": "Acme  Inc", "industry": " Finance ", "state": "ID"}, {"name": "Karl Novak", "phone": "(532)635-6981", "country": "Germany", "job_title": "Engineering Manager", "rate": "45.1%", "company": "   Stark Industries", "industry": " Finance ", "state": "Delaware"}, {"name": "Alice Wong", "phone": "584-880-6961", "country": "FEDEAL REPUBLIC OF GERMANY", "job_title": "Eng Manager", "rate": "42.8%", "company": "Wonka Co ", "industry": "Utilities", "state": "idaaho"}, {"name": "Mona Cruz", "phone": "(545)585-0573", "country": "Gm", "job_title": "   Engineering Mgr", "rate": "49.1%", "company": "Cyberdyne", "industry": "oil & gas", "state": "wa"}, {"name": " Alice Ali", "phone": "510.258.5589", "country": " gm ", "job_title": " AE ", "rate": "46.1%", "company": "Globex", "industry": "  Utilities", "state": "DE"}, {"name": "  Heidi Cruz", "phone": "572-480-4352", "country": "REPUBLIC OF TÜRKIYE", "job_title": "Account Executive", "rate": "51.1%", "company": "  Wonka Co", "industry": "Fianncial Services", "state": "Idaho"}, {"name": "Judy Fischer", "phone": "560-512-7790", "country": "Republic of the Gambia", "job_title": "Acct Exec", "rate": "47.8%", "company": "Globex", "industry": "energy", "state": "DE"}, {"name": "Alice Moore", "phone": "(518)523-5565", "country": "republic of the gambia", "job_title": "Acct Exec", "rate": "20.9%", "company": "   Wonka Co", "industry": "Banking", "state": "DE"}, {"name": "Frank Diaz", "phone": "514.671.1145", "country": "Saint Lucia", "job_title": "ENG MGGR", "rate": "78.4%", "company": "  Vehement", "industry": "ENERGY", "state": "wa"}, {"name": "Karl Petrov", "phone": "568-088-5044", "country": "Türkiye", "job_title": "Engineering Mgr", "rate": "22.6%", "company": "Cyberdyne", "industry": "Oil & Gas", "state": "NV"}, {"name": "Frank  Adams", "phone": "(595)041-1981", "country": "GM", "job_title": "Eng Manager", "rate": "15.0%", "company": "Umbrella", "industry": "oil & gas", "state": "Idaho"}, {"name": "Karl Fischer", "phone": "(534)313-0983", "country": "   lc", "job_title": "account executive", "rate": "55.1%", "company": "Globex  ", "industry": "energy", "state": " delaware"}, {"name": "Heidi Novak", "phone": "501.271.9169", "country": "GM", "job_title": "ENGINEERING MGR", "rate": "25.6%", "company": "Initech", "industry": "eneryg", "state": "Washington"}, {"name": "Omar Lee", "phone": "536-106-5673", "country": "Saint  Lucia", "job_title": "Engineeriing Manager", "rate": "70.5%", "company": "Acme Inc", "industry": "Energy   ", "state": "Delaware  "}, {"name": "  Grace Wong", "phone": "5562075639", "country": "republic of türkiye", "job_title": "engineering mgr", "rate": "49.9%", "company": "Stark Industries", "industry": "Finjech", "state": "NV"}, {"name": "Sara  Petrov", "phone": "(535)837-4119", "country": "GAMBIA", "job_title": "Acct Exec", "rate": "50.0%", "company": " Wonka Co", "industry": "UTILITIES", "state": "IDAHO"}, {"name": "Frank Fischer ", "phone": "(532)179-3037", "country": "TÜRKIIE", "job_title": "AE", "rate": "76.6%", "company": "Stark Industries ", "industry": "Finserv", "state": " NV "}, {"name": "David Wong", "phone": "517.225.1563", "country": "REPUBLIC OF TRKIYE", "job_title": "account executive", "rate": "58.5%", "company": "Hooli ", "industry": " Energy ", "state": "  IDAHO"}, {"name": "Ivan Moore   ", "phone": "595-812-7653", "country": " LC ", "job_title": "Acct Exec", "rate": "37.7%", "company": "Initech", "industry": "Utilities  ", "state": "   Wa"}, {"name": "Judy Moore", "phone": "563.550.2710", "country": "lca", "job_title": "AE", "rate": "55.4%", "company": "  Cyberdyne", "industry": "FinServ", "state": "nevada   "}, {"name": "Bob Park ", "phone": "542.134.5596", "country": "republic of trkiye", "job_title": "eng manager", "rate": "79.6%", "company": "Acme  Inc", "industry": "Financial Services", "state": "Idaro"}, {"name": "Karl  Wong", "phone": "5840330804", "country": "Saint  Lucia", "job_title": "ae", "rate": "5.0%", "company": " Cyberdyne ", "industry": "Financial Services", "state": "wa"}, {"name": "Lena Diaz", "phone": "5678772379", "country": "REPUBLIC OF TÜRKIYE", "job_title": "   Engineering Mgr", "rate": "79.9%", "company": "Wonka  Co", "industry": "financial services", "state": "Igaho"}, {"name": "Grace Park", "phone": "578-566-4426", "country": "TR", "job_title": "Engineering Mgr", "rate": "28.9%", "company": " Wonka Co", "industry": "fintech", "state": "Nv"}, {"name": "Bob Wong", "phone": "551-417-0972", "country": "  TR", "job_title": "eng mgr", "rate": "59.7%", "company": "Umbrella  ", "industry": "fintech ", "state": "IDAHO"}, {"name": "   Mona Johnson", "phone": "559-085-7043", "country": "republic of  the gambia", "job_title": "Acct Exec", "rate": "50.8%", "company": "Stark Industries", "industry": "  Energy", "state": "DE"}, {"name": "Ivan Diaz", "phone": "550-395-5613", "country": "Deu", "job_title": "  acct exec", "rate": "95.6%", "company": "Wonka  Co", "industry": " Energy ", "state": "Id"}, {"name": "Heidi Cruz   ", "phone": "503-609-8524", "country": "   Gambia", "job_title": " ENG MGR", "rate": "96.2%", "company": "Acme Inc  ", "industry": "Fintech", "state": "Wahsington"}, {"name": "", "phone": "", "country": "", "job_title": "", "rate": "", "company": "", "industry": "", "state": ""}], "clean": [{"name": "Bob Ali", "phone": "(543) 611-6641", "country": "Germany", "job_title": "Engineering Manager", "rate": 0.31, "company": "Acme Inc", "industry": "Energy", "state": "Washington"}, {"name": "Heidi Brandt", "phone": "(518) 656-7466", "country": "Gambia", "job_title": "Account Executive", "rate": 0.134, "company": "Hooli", "industry": "Energy", "state": "Nevada"}, {"name": "Ivan Wong", "phone": "(586) 489-5089", "country": "Gambia", "job_title": "Account Executive", "rate": 0.069, "company": "Wonka Co", "industry": "Financial Services", "state": "Nevada"}, {"name": "Bob Park", "phone": "(558) 876-1053", "country": "Gambia", "job_title": "Account Executive", "rate": 0.46, "company": "Soylent Corp", "industry": "Financial Services", "state": "Nevada"}, {"name": "Mona Petrov", "phone": "(588) 967-7614", "country": "Gambia", "job_title": "Account Executive", "rate": 0.561, "company": "Wonka Co", "industry": "Financial Services", "state": "Delaware"}, {"name": "Eve Fischer", "phone": "(579) 157-0016", "country": "Germany", "job_title": "Account Executive", "rate": 0.038, "company": "Wonka Co", "industry": "Energy", "state": "Idaho"}, {"name": "Carol Smith", "phone": "(556) 022-6632", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.34299999999999997, "company": "Umbrella", "industry": "Energy", "state": "Delaware"}, {"name": "Eve Wong", "phone": "(579) 614-5311", "country": "Germany", "job_title": "Account Executive", "rate": 0.68, "company": "Initech", "industry": "Financial Services", "state": "Delaware"}, {"name": "Sara Reyes", "phone": "(571) 990-8461", "country": "Türkiye", "job_title": "Account Executive", "rate": 0.47, "company": "Globex", "industry": "Energy", "state": "Nevada"}, {"name": "Carol Fischer", "phone": "(557) 695-7251", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.8909999999999999, "company": "Initech", "industry": "Financial Services", "state": "Delaware"}, {"name": "Eve Wong", "phone": "(510) 826-0154", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.374, "company": "Stark Industries", "industry": "Energy", "state": "Idaho"}, {"name": "Ivan Wong", "phone": "(542) 112-8555", "country": "Germany", "job_title": "Account Executive", "rate": 0.845, "company": "Cyberdyne", "industry": "Financial Services", "state": "Washington"}, {"name": "Alice Wong", "phone": "(561) 229-2655", "country": "Germany", "job_title": "Engineering Manager", "rate": 0.862, "company": "Vehement", "industry": "Energy", "state": "Nevada"}, {"name": "Priya Adams", "phone": "(551) 610-7042", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.34299999999999997, "company": "Acme Inc", "industry": "Energy", "state": "Idaho"}, {"name": "Carol Smith", "phone": "(518) 299-6818", "country": "Germany", "job_title": "Account Executive", "rate": 0.809, "company": "Globex", "industry": "Financial Services", "state": "Nevada"}, {"name": "Eve Moore", "phone": "(543) 674-4030", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.06, "company": "Stark Industries", "industry": "Financial Services", "state": "Washington"}, {"name": "Mona Diaz", "phone": "(561) 481-4084", "country": "Germany", "job_title": "Account Executive", "rate": 0.42200000000000004, "company": "Hooli", "industry": "Financial Services", "state": "Washington"}, {"name": "Bob Adams", "phone": "(537) 810-2939", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.998, "company": "Soylent Corp", "industry": "Financial Services", "state": "Nevada"}, {"name": "Frank Khan", "phone": "(569) 216-6985", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.59, "company": "Acme Inc", "industry": "Financial Services", "state": "Nevada"}, {"name": "Heidi Johnson", "phone": "(560) 348-0761", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.434, "company": "Vehement", "industry": "Energy", "state": "Washington"}, {"name": "Omar Diaz", "phone": "(552) 330-1058", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.373, "company": "Stark Industries", "industry": "Energy", "state": "Washington"}, {"name": "Carol Park", "phone": "(519) 760-2783", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.588, "company": "Umbrella", "industry": "Financial Services", "state": "Idaho"}, {"name": "Bob Wong", "phone": "(597) 190-7678", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.735, "company": "Globex", "industry": "Financial Services", "state": "Washington"}, {"name": "Heidi Lee", "phone": "(592) 142-7628", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.069, "company": "Acme Inc", "industry": "Financial Services", "state": "Idaho"}, {"name": "Karl Novak", "phone": "(532) 635-6981", "country": "Germany", "job_title": "Engineering Manager", "rate": 0.451, "company": "Stark Industries", "industry": "Financial Services", "state": "Delaware"}, {"name": "Alice Wong", "phone": "(584) 880-6961", "country": "Germany", "job_title": "Engineering Manager", "rate": 0.428, "company": "Wonka Co", "industry": "Energy", "state": "Idaho"}, {"name": "Mona Cruz", "phone": "(545) 585-0573", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.491, "company": "Cyberdyne", "industry": "Energy", "state": "Washington"}, {"name": "Alice Ali", "phone": "(510) 258-5589", "country": "Gambia", "job_title": "Account Executive", "rate": 0.461, "company": "Globex", "industry": "Energy", "state": "Delaware"}, {"name": "Heidi Cruz", "phone": "(572) 480-4352", "country": "Türkiye", "job_title": "Account Executive", "rate": 0.511, "company": "Wonka Co", "industry": "Financial Services", "state": "Idaho"}, {"name": "Judy Fischer", "phone": "(560) 512-7790", "country": "Gambia", "job_title": "Account Executive", "rate": 0.478, "company": "Globex", "industry": "Energy", "state": "Delaware"}, {"name": "Alice Moore", "phone": "(518) 523-5565", "country": "Gambia", "job_title": "Account Executive", "rate": 0.209, "company": "Wonka Co", "industry": "Financial Services", "state": "Delaware"}, {"name": "Frank Diaz", "phone": "(514) 671-1145", "country": "Saint Lucia", "job_title": "Engineering Manager", "rate": 0.784, "company": "Vehement", "industry": "Energy", "state": "Washington"}, {"name": "Karl Petrov", "phone": "(568) 088-5044", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.226, "company": "Cyberdyne", "industry": "Energy", "state": "Nevada"}, {"name": "Frank Adams", "phone": "(595) 041-1981", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.15, "company": "Umbrella", "industry": "Energy", "state": "Idaho"}, {"name": "Karl Fischer", "phone": "(534) 313-0983", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.551, "company": "Globex", "industry": "Energy", "state": "Delaware"}, {"name": "Heidi Novak", "phone": "(501) 271-9169", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.256, "company": "Initech", "industry": "Energy", "state": "Washington"}, {"name": "Omar Lee", "phone": "(536) 106-5673", "country": "Saint Lucia", "job_title": "Engineering Manager", "rate": 0.705, "company": "Acme Inc", "industry": "Energy", "state": "Delaware"}, {"name": "Grace Wong", "phone": "(556) 207-5639", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.499, "company": "Stark Industries", "industry": "Financial Services", "state": "Nevada"}, {"name": "Sara Petrov", "phone": "(535) 837-4119", "country": "Gambia", "job_title": "Account Executive", "rate": 0.5, "company": "Wonka Co", "industry": "Energy", "state": "Idaho"}, {"name": "Frank Fischer", "phone": "(532) 179-3037", "country": "Türkiye", "job_title": "Account Executive", "rate": 0.7659999999999999, "company": "Stark Industries", "industry": "Financial Services", "state": "Nevada"}, {"name": "David Wong", "phone": "(517) 225-1563", "country": "Türkiye", "job_title": "Account Executive", "rate": 0.585, "company": "Hooli", "industry": "Energy", "state": "Idaho"}, {"name": "Ivan Moore", "phone": "(595) 812-7653", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.377, "company": "Initech", "industry": "Energy", "state": "Washington"}, {"name": "Judy Moore", "phone": "(563) 550-2710", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.5539999999999999, "company": "Cyberdyne", "industry": "Financial Services", "state": "Nevada"}, {"name": "Bob Park", "phone": "(542) 134-5596", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.7959999999999999, "company": "Acme Inc", "industry": "Financial Services", "state": "Idaho"}, {"name": "Karl Wong", "phone": "(584) 033-0804", "country": "Saint Lucia", "job_title": "Account Executive", "rate": 0.05, "company": "Cyberdyne", "industry": "Financial Services", "state": "Washington"}, {"name": "Lena Diaz", "phone": "(567) 877-2379", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.799, "company": "Wonka Co", "industry": "Financial Services", "state": "Idaho"}, {"name": "Grace Park", "phone": "(578) 566-4426", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.289, "company": "Wonka Co", "industry": "Financial Services", "state": "Nevada"}, {"name": "Bob Wong", "phone": "(551) 417-0972", "country": "Türkiye", "job_title": "Engineering Manager", "rate": 0.597, "company": "Umbrella", "industry": "Financial Services", "state": "Idaho"}, {"name": "Mona Johnson", "phone": "(559) 085-7043", "country": "Gambia", "job_title": "Account Executive", "rate": 0.508, "company": "Stark Industries", "industry": "Energy", "state": "Delaware"}, {"name": "Ivan Diaz", "phone": "(550) 395-5613", "country": "Germany", "job_title": "Account Executive", "rate": 0.956, "company": "Wonka Co", "industry": "Energy", "state": "Idaho"}, {"name": "Heidi Cruz", "phone": "(503) 609-8524", "country": "Gambia", "job_title": "Engineering Manager", "rate": 0.9620000000000001, "company": "Acme Inc", "industry": "Financial Services", "state": "Washington"}], "dirty_cols": ["name", "phone", "country", "job_title", "rate", "company", "industry", "state"], "clean_cols": ["name", "phone", "country", "job_title", "rate", "company", "industry", "state"], "plan": {"dataset_summary": "52 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Federal Republic Of Germany": "Germany", "gm": "Gambia", "Republic of the Gambia": "Gambia", "GAMBIA": "Gambia", "GM": "Gambia", "deu": "Germany", "lc": "Saint Lucia", "GERMANY": "Germany", "TUR": "Türkiye", "LC": "Saint Lucia", "DE": "Germany", "Federal Republic of Germany": "Germany", "lca": "Saint Lucia", "federal republic of germany": "Germany", "Tr": "Türkiye", "GMB": "Gambia", "saint lcia": "Saint Lucia", "saint lucia": "Saint Lucia", "Republic Of Türkiye": "Türkiye", "FEDEAL REPUBLIC OF GERMANY": "Germany", "Gm": "Gambia", "REPUBLIC OF TÜRKIYE": "Türkiye", "republic of the gambia": "Gambia", "Saint  Lucia": "Saint Lucia", "republic of türkiye": "Türkiye", "TÜRKIIE": "Türkiye", "REPUBLIC OF TRKIYE": "Türkiye", "republic of trkiye": "Türkiye", "TR": "Türkiye", "Deu": "Germany"}, "rationale": "Unified 30 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ENGINEERING MANAGER": "Engineering Manager", "accuunt executive": "Account Executive", "Acctt Exec": "Account Executive", "Acct Exec": "Account Executive", "ae": "Account Executive", "account executive": "Account Executive", "acct exec": "Account Executive", "Account Kxecutive": "Account Executive", "Acct Exce": "Account Executive", "Engineering Mgr": "Engineering Manager", "Ae": "Account Executive", "engineering mgr": "Engineering Manager", "Eng Mgr": "Engineering Manager", "Eng Manaer": "Engineering Manager", "eng gr": "Engineering Manager", "AE": "Account Executive", "Eng Manager": "Engineering Manager", "ENG MGGR": "Engineering Manager", "ENGINEERING MGR": "Engineering Manager", "Engineeriing Manager": "Engineering Manager", "eng manager": "Engineering Manager", "eng mgr": "Engineering Manager", "ENG MGR": "Engineering Manager"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"binking": "Financial Services", "banking": "Financial Services", "ENERYG": "Energy", "enregy": "Energy", "Utilities": "Energy", "FINTECH": "Financial Services", "Oil & Gas": "Energy", "energy": "Energy", "Finanec": "Financial Services", "Fintech": "Financial Services", "Finance": "Financial Services", "finance": "Financial Services", "FinServ": "Financial Services", "finserv": "Financial Services", "oil & gas": "Energy", "Fianncial Services": "Financial Services", "Banking": "Financial Services", "ENERGY": "Energy", "eneryg": "Energy", "Finjech": "Financial Services", "UTILITIES": "Energy", "Finserv": "Financial Services", "financial services": "Financial Services", "fintech": "Financial Services"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WA": "Washington", "NEVKDA": "Nevada", "NV": "Nevada", "delaware": "Delaware", "id": "Idaho", "DE": "Delaware", "NEVADA": "Nevada", "nevada": "Nevada", "wa": "Washington", "washington": "Washington", "Wa": "Washington", "ID": "Idaho", "idaaho": "Idaho", "IDAHO": "Idaho", "Idaro": "Idaho", "Igaho": "Idaho", "Nv": "Nevada", "Id": "Idaho", "Wahsington": "Washington"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": "  Sales", "amount": "5.249,22", "signup_date": "45048", "col_x": ""}, {"department": "IT", "amount": "$3,769.19", "signup_date": "2023-07-21", "col_x": ""}, {"department": "People Ops  ", "amount": "$430.74", "signup_date": "9/25/2023", "col_x": ""}, {"department": "People Ops", "amount": "$3,548.93", "signup_date": "8/15/2023", "col_x": ""}, {"department": "OPERATIONS", "amount": "9999999", "signup_date": "2023-07-09", "col_x": ""}, {"department": "Human Resources", "amount": "--", "signup_date": "2/10/2023", "col_x": ""}, {"department": "Ops   ", "amount": "$2,850.83", "signup_date": "25 Aug 2023", "col_x": ""}, {"department": "operations", "amount": "TBD", "signup_date": "2023-03-21", "col_x": ""}, {"department": "OPS", "amount": "$1,007.80", "signup_date": "1/28/2023", "col_x": ""}, {"department": "HR", "amount": "$5,923.18", "signup_date": "16 May 2023", "col_x": ""}, {"department": "Ops", "amount": "268,79", "signup_date": "3/4/2023", "col_x": ""}, {"department": "Information Technology", "amount": "2.467,94", "signup_date": "2023-08-26", "col_x": ""}, {"department": "information technology", "amount": "$5,092.81", "signup_date": "2023-11-27", "col_x": ""}, {"department": "ops", "amount": "$8,513.82", "signup_date": "2023-10-24", "col_x": ""}, {"department": "Sales", "amount": "1.198,21", "signup_date": "12/27/2023", "col_x": ""}, {"department": "it", "amount": "2.384,84", "signup_date": "45116", "col_x": ""}, {"department": "Ops", "amount": "$3,303.50", "signup_date": "3/11/2023", "col_x": ""}, {"department": "ops", "amount": "$4,942.05", "signup_date": "2023-07-24", "col_x": ""}, {"department": "ops", "amount": "296,03", "signup_date": "2023-05-10", "col_x": ""}, {"department": "I.T.", "amount": "1.456,06", "signup_date": "2023-06-07", "col_x": ""}], "clean": [{"department": "Sales", "amount": 5249.22, "signup_date": "2023-05-02"}, {"department": "Information Technology", "amount": 3769.19, "signup_date": "2023-07-21"}, {"department": "Human Resources", "amount": 430.74, "signup_date": "2023-09-25"}, {"department": "Human Resources", "amount": 3548.93, "signup_date": "2023-08-15"}, {"department": "Operations", "amount": 9999999.0, "signup_date": "2023-07-09"}, {"department": "Human Resources", "amount": NaN, "signup_date": "2023-02-10"}, {"department": "Operations", "amount": 2850.83, "signup_date": "2023-08-25"}, {"department": "Operations", "amount": NaN, "signup_date": "2023-03-21"}, {"department": "Operations", "amount": 1007.8, "signup_date": "2023-01-28"}, {"department": "Human Resources", "amount": 5923.18, "signup_date": "2023-05-16"}, {"department": "Operations", "amount": 268.79, "signup_date": "2023-03-04"}, {"department": "Information Technology", "amount": 2467.94, "signup_date": "2023-08-26"}, {"department": "Information Technology", "amount": 5092.81, "signup_date": "2023-11-27"}, {"department": "Operations", "amount": 8513.82, "signup_date": "2023-10-24"}, {"department": "Sales", "amount": 1198.21, "signup_date": "2023-12-27"}, {"department": "Information Technology", "amount": 2384.84, "signup_date": "2023-07-09"}, {"department": "Operations", "amount": 3303.5, "signup_date": "2023-03-11"}, {"department": "Operations", "amount": 4942.05, "signup_date": "2023-07-24"}, {"department": "Operations", "amount": 296.03, "signup_date": "2023-05-10"}, {"department": "Information Technology", "amount": 1456.06, "signup_date": "2023-06-07"}], "dirty_cols": ["department", "amount", "signup_date", "col_x"], "clean_cols": ["department", "amount", "signup_date"], "plan": {"dataset_summary": "20 rows × 4 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "People Ops": "Human Resources", "OPERATIONS": "Operations", "Ops": "Operations", "operations": "Operations", "OPS": "Operations", "HR": "Human Resources", "information technology": "Information Technology", "ops": "Operations", "it": "Information Technology", "I.T.": "Information Technology"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"unit": "  pounds", "country": "Lithuania", "company": "Acme Inc", "status": "late"}, {"unit": "  Km", "country": "ncl", "company": "Globex", "status": "Paid"}, {"unit": "Km", "country": "SJ", "company": "Wonka  Co", "status": "piad"}, {"unit": "In", "country": "TG   ", "company": " Cyberdyne ", "status": "overduue"}, {"unit": "lbs", "country": "SJM", "company": "Vehement", "status": "overdue"}, {"unit": "inches", "country": "New Caledonia", "company": "Vehement", "status": "PAID"}, {"unit": "inch", "country": "NC", "company": "Globex", "status": "Paid"}, {"unit": "KILOMETERS", "country": "SJM", "company": "   Umbrella", "status": "Refunded"}, {"unit": "Km", "country": "SJM", "company": "  Umbrella", "status": "Overdue"}, {"unit": "kilometers", "country": "TOGO", "company": "   Acme Inc", "status": " Overdue "}, {"unit": "kilometer", "country": " sjm", "company": "Stark Industries", "status": "refunded"}, {"unit": "inches ", "country": "Togo", "company": "Vehement", "status": " Paid "}], "clean": [{"unit": "lb", "country": "Lithuania", "company": "Acme Inc", "status": "Overdue"}, {"unit": "km", "country": "New Caledonia", "company": "Globex", "status": "Paid"}, {"unit": "km", "country": "Svalbard and Jan Mayen", "company": "Wonka Co", "status": "Paid"}, {"unit": "in", "country": "Togo", "company": "Cyberdyne", "status": "Overdue"}, {"unit": "lb", "country": "Svalbard and Jan Mayen", "company": "Vehement", "status": "Overdue"}, {"unit": "in", "country": "New Caledonia", "company": "Vehement", "status": "Paid"}, {"unit": "in", "country": "New Caledonia", "company": "Globex", "status": "Paid"}, {"unit": "km", "country": "Svalbard and Jan Mayen", "company": "Umbrella", "status": "Refunded"}, {"unit": "km", "country": "Svalbard and Jan Mayen", "company": "Umbrella", "status": "Overdue"}, {"unit": "km", "country": "Togo", "company": "Acme Inc", "status": "Overdue"}, {"unit": "km", "country": "Svalbard and Jan Mayen", "company": "Stark Industries", "status": "Refunded"}, {"unit": "in", "country": "Togo", "company": "Vehement", "status": "Paid"}], "dirty_cols": ["unit", "country", "company", "status"], "clean_cols": ["unit", "country", "company", "status"], "plan": {"dataset_summary": "12 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pounds": "lb", "Km": "km", "In": "in", "lbs": "lb", "inches": "in", "inch": "in", "KILOMETERS": "km", "kilometers": "km", "kilometer": "km"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ncl": "New Caledonia", "SJ": "Svalbard and Jan Mayen", "TG": "Togo", "SJM": "Svalbard and Jan Mayen", "NC": "New Caledonia", "TOGO": "Togo", "sjm": "Svalbard and Jan Mayen"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"late": "Overdue", "piad": "Paid", "overduue": "Overdue", "overdue": "Overdue", "PAID": "Paid", "refunded": "Refunded"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "4.4%", "amount": "5.871,94", "is_active": "false", "job_title": " Chief Technology Officer", "unit": " ml ", "signup_date": "45113"}, {"rate": "12.5%", "amount": "6.165,84", "is_active": "true", "job_title": "C.TT.O.   ", "unit": "LBS", "signup_date": "45001"}, {"rate": "69.7%", "amount": "$576.07", "is_active": "No", "job_title": "Chief Technology Ofvicer", "unit": "milliliter", "signup_date": "2023-12-18"}, {"rate": "96.8%", "amount": "$3,918.26", "is_active": "N", "job_title": "Product Manager", "unit": "Mi", "signup_date": "25 Aug 2023"}, {"rate": "27.8%", "amount": "3.767,04", "is_active": "No", "job_title": "prod manager ", "unit": "mL", "signup_date": "12/10/2023"}, {"rate": "41.1%", "amount": "None", "is_active": "Yes", "job_title": "prod  manager", "unit": "ml", "signup_date": "22 Oct 2023"}, {"rate": "20.8%", "amount": "8.222,69", "is_active": "No", "job_title": "CTO", "unit": "miles", "signup_date": "19 Jan 2023"}, {"rate": "40.4%", "amount": "$3,386.89", "is_active": "No", "job_title": "C.T.O.", "unit": "miles", "signup_date": "45164"}, {"rate": "97.0%", "amount": "4.176,25", "is_active": "0", "job_title": " prod mgr", "unit": "ML", "signup_date": "2023-01-26"}, {"rate": "72.4%", "amount": "$8,878.12", "is_active": "FALSE", "job_title": "productt manager", "unit": "ml", "signup_date": "2/4/2023"}, {"rate": "70.4%", "amount": "360,52", "is_active": "F", "job_title": "prod manager", "unit": "mls", "signup_date": "45128"}, {"rate": "86.0%", "amount": "$3,842.80", "is_active": "1", "job_title": "c.t.o.  ", "unit": "  LB", "signup_date": "2023-12-08"}, {"rate": "99.3%", "amount": "9999999", "is_active": "1", "job_title": "CTO", "unit": "mi", "signup_date": "15 Dec 2023"}, {"rate": "68.6%", "amount": "$2,725.56", "is_active": "true", "job_title": "Chief Technology Officer", "unit": "mls", "signup_date": "26 May 2023"}, {"rate": "83.1%", "amount": "6.190,61", "is_active": "1", "job_title": "C.T.O.", "unit": "pound  ", "signup_date": "11/16/2023"}, {"rate": "14.8%", "amount": "7.525,16", "is_active": "0", "job_title": "Chief Technology Offier ", "unit": "milliliter", "signup_date": "8/3/2023"}, {"rate": "10.0%", "amount": "2.579,27", "is_active": "false", "job_title": "Chief Technology Officer  ", "unit": "mi", "signup_date": "45008"}, {"rate": "36.7%", "amount": "3.497,85", "is_active": "TRUE", "job_title": "PM", "unit": "ml", "signup_date": "13 Jun 2023"}, {"rate": "20.9%", "amount": "$6,234.17", "is_active": "TRUE", "job_title": "chief technology officer", "unit": "poud", "signup_date": "2023-02-11"}, {"rate": "57.8%", "amount": "5.231,48", "is_active": "Yes", "job_title": " pm ", "unit": "miles", "signup_date": "2023-07-22"}, {"rate": "12.5%", "amount": "6.165,84", "is_active": "true", "job_title": "C.TT.O.   ", "unit": "LBS", "signup_date": "45001"}], "clean": [{"rate": 0.044000000000000004, "amount": 5871.94, "is_active": false, "job_title": "Chief Technology Officer", "unit": "mL", "signup_date": "2023-07-06"}, {"rate": 0.125, "amount": 6165.84, "is_active": true, "job_title": "Chief Technology Officer", "unit": "lb", "signup_date": "2023-03-16"}, {"rate": 0.6970000000000001, "amount": 576.07, "is_active": false, "job_title": "Chief Technology Officer", "unit": "mL", "signup_date": "2023-12-18"}, {"rate": 0.968, "amount": 3918.26, "is_active": false, "job_title": "Product Manager", "unit": "mi", "signup_date": "2023-08-25"}, {"rate": 0.278, "amount": 3767.04, "is_active": false, "job_title": "Product Manager", "unit": "mL", "signup_date": "2023-12-10"}, {"rate": 0.41100000000000003, "amount": NaN, "is_active": true, "job_title": "Product Manager", "unit": "mL", "signup_date": "2023-10-22"}, {"rate": 0.20800000000000002, "amount": 8222.69, "is_active": false, "job_title": "Chief Technology Officer", "unit": "mi", "signup_date": "2023-01-19"}, {"rate": 0.40399999999999997, "amount": 3386.89, "is_active": false, "job_title": "Chief Technology Officer", "unit": "mi", "signup_date": "2023-08-26"}, {"rate": 0.97, "amount": 4176.25, "is_active": false, "job_title": "Product Manager", "unit": "mL", "signup_date": "2023-01-26"}, {"rate": 0.7240000000000001, "amount": 8878.12, "is_active": false, "job_title": "Product Manager", "unit": "mL", "signup_date": "2023-02-04"}, {"rate": 0.7040000000000001, "amount": 360.52, "is_active": false, "job_title": "Product Manager", "unit": "mL", "signup_date": "2023-07-21"}, {"rate": 0.86, "amount": 3842.8, "is_active": true, "job_title": "Chief Technology Officer", "unit": "lb", "signup_date": "2023-12-08"}, {"rate": 0.993, "amount": 9999999.0, "is_active": true, "job_title": "Chief Technology Officer", "unit": "mi", "signup_date": "2023-12-15"}, {"rate": 0.6859999999999999, "amount": 2725.56, "is_active": true, "job_title": "Chief Technology Officer", "unit": "mL", "signup_date": "2023-05-26"}, {"rate": 0.831, "amount": 6190.61, "is_active": true, "job_title": "Chief Technology Officer", "unit": "lb", "signup_date": "2023-11-16"}, {"rate": 0.14800000000000002, "amount": 7525.16, "is_active": false, "job_title": "Chief Technology Officer", "unit": "mL", "signup_date": "2023-08-03"}, {"rate": 0.1, "amount": 2579.27, "is_active": false, "job_title": "Chief Technology Officer", "unit": "mi", "signup_date": "2023-03-23"}, {"rate": 0.36700000000000005, "amount": 3497.85, "is_active": true, "job_title": "Product Manager", "unit": "mL", "signup_date": "2023-06-13"}, {"rate": 0.209, "amount": 6234.17, "is_active": true, "job_title": "Chief Technology Officer", "unit": "lb", "signup_date": "2023-02-11"}, {"rate": 0.578, "amount": 5231.48, "is_active": true, "job_title": "Product Manager", "unit": "mi", "signup_date": "2023-07-22"}], "dirty_cols": ["rate", "amount", "is_active", "job_title", "unit", "signup_date"], "clean_cols": ["rate", "amount", "is_active", "job_title", "unit", "signup_date"], "plan": {"dataset_summary": "21 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"C.TT.O.": "Chief Technology Officer", "Chief Technology Ofvicer": "Chief Technology Officer", "prod manager": "Product Manager", "prod  manager": "Product Manager", "CTO": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "prod mgr": "Product Manager", "productt manager": "Product Manager", "c.t.o.": "Chief Technology Officer", "Chief Technology Offier": "Chief Technology Officer", "PM": "Product Manager", "chief technology officer": "Chief Technology Officer", "pm": "Product Manager"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ml": "mL", "LBS": "lb", "milliliter": "mL", "Mi": "mi", "miles": "mi", "ML": "mL", "mls": "mL", "LB": "lb", "pound": "lb", "poud": "lb"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"company": " Umbrella ", "department": "FINANCE", "rate": "77.0%", "city": "Daal", "industry": "ed ", "unit": "kilograms"}, {"company": "Cyberdyne ", "department": "human resources", "rate": "7.0%", "city": "Yaring", "industry": "Banking", "unit": "Mi"}, {"company": "Umbrella  ", "department": "  Finance", "rate": "64.0%", "city": "Siderno", "industry": " FINTECH ", "unit": "MILLILITER"}, {"company": " Umbrella ", "department": "   accounting", "rate": "55.9%", "city": "Yaring ", "industry": "Manufacturing", "unit": "Miwes"}, {"company": "Globex", "department": "human resources   ", "rate": "16.4%", "city": "Siderno", "industry": "Mfg  ", "unit": "mi"}, {"company": " Cyberdyne ", "department": "ops", "rate": "28.2%", "city": " Nairobi ", "industry": "Ed", "unit": "milliliter"}, {"company": "   Cyberdyne", "department": "BIZ DEV  ", "rate": "85.9%", "city": "nairobi", "industry": "Software", "unit": " MLS"}, {"company": "Hooli", "department": "Biz Dev", "rate": "25.8%", "city": " Nairobi ", "industry": "Education", "unit": " mi "}, {"company": " Cyberdyne ", "department": "people", "rate": "72.5%", "city": "siderno", "industry": "Banking", "unit": " kilogram"}, {"company": "Initech   ", "department": "operations", "rate": "89.5%", "city": "Dayal", "industry": "Software", "unit": "Mls"}, {"company": "Stark  Industries", "department": "Fin", "rate": "75.2%", "city": "YARING", "industry": "MANUFACTURING", "unit": "KG"}, {"company": " Globex ", "department": "Finance", "rate": "62.1%", "city": " Yaring ", "industry": "Banking", "unit": "miles"}, {"company": "Cyberdyne", "department": "Sales", "rate": "15.3%", "city": "Nairrobi   ", "industry": "manufacturing", "unit": "ml"}, {"company": " Soylent Corp", "department": "Operations", "rate": "38.4%", "city": "siderno   ", "industry": "EdTech", "unit": "ML"}, {"company": " Acme Inc", "department": " fin ", "rate": "50.9%", "city": "Siderno  ", "industry": "Mfg", "unit": "miles"}, {"company": "Acme  Inc", "department": "People", "rate": "43.5%", "city": "Nairobi", "industry": "finserv   ", "unit": "MI"}, {"company": "Acme  Inc", "department": "OPERATIONS", "rate": "22.1%", "city": "Nairoi", "industry": "Softlare", "unit": " Kilogram "}, {"company": "   Stark Industries", "department": "Sales", "rate": "58.3%", "city": "Yarring", "industry": "Tech", "unit": "mi"}, {"company": "Cyberdyne", "department": "Ops", "rate": "10.8%", "city": "siderno", "industry": "Educaion", "unit": "kilograms"}, {"company": "   Initech", "department": "Operations", "rate": "78.1%", "city": "Yariing", "industry": "Financial  Serices", "unit": " mi "}, {"company": "Cyberdyne", "department": "Sales", "rate": "34.5%", "city": "Nairobi", "industry": " It ", "unit": "mls"}, {"company": "Wonka  Co", "department": " ops ", "rate": "31.2%", "city": "Yaring", "industry": "Finance", "unit": "ml"}, {"company": "  Cyberdyne", "department": "ops   ", "rate": "3.3%", "city": " yaring ", "industry": "Mfg", "unit": "kilogram"}, {"company": "  Hooli", "department": "   Human Resources", "rate": "43.3%", "city": "Nairobi", "industry": "tech", "unit": "miles"}, {"company": "Soylent Corp", "department": " Ops ", "rate": "28.1%", "city": "nairobi", "industry": "   Ed", "unit": "   kilograms"}, {"company": "Initech", "department": "OPS", "rate": "18.3%", "city": "YRAING", "industry": "mfg", "unit": "mL"}, {"company": "Acme Inc", "department": "Finance", "rate": "48.0%", "city": " Dayal", "industry": "Tch", "unit": "mi"}, {"company": "Globex", "department": "  Accounting", "rate": "42.8%", "city": "Nairobi", "industry": "TECHNOLOGY", "unit": "KG   "}, {"company": "Stark Industries", "department": "Biz Dev", "rate": "26.5%", "city": "Dayal", "industry": "education", "unit": "ml"}, {"company": "Umbrella", "department": "HR", "rate": "77.4%", "city": " SIDERNO ", "industry": "Education", "unit": "ml  "}, {"company": " Initech ", "department": "Biz Dev", "rate": "40.4%", "city": "Yaring", "industry": "EDECH", "unit": " KG "}, {"company": "Acme Inc   ", "department": "sales   ", "rate": "79.0%", "city": "Dayal", "industry": "Mfg", "unit": "MILE"}, {"company": " Umbrella", "department": "Biz Dev", "rate": "13.4%", "city": "Dayal", "industry": "BANKING", "unit": "Kg"}, {"company": " Hooli ", "department": "SALES", "rate": "26.2%", "city": " Dayal", "industry": " Fintech ", "unit": "KILOGRAMS"}, {"company": "Soylent  Corp", "department": "Human Resources", "rate": "26.2%", "city": "Yaring   ", "industry": "Edtech", "unit": "mL"}, {"company": "Umbrella  ", "department": "OPS ", "rate": "77.8%", "city": "siderno", "industry": "Financial Services", "unit": "mL"}, {"company": " Umbrella", "department": "Accounting ", "rate": "98.2%", "city": "dayal", "industry": "software ", "unit": "Mi"}, {"company": "  Cyberdyne", "department": "Accounting", "rate": "90.2%", "city": "Yarnig", "industry": "TECHNOLOGY", "unit": "kilogram"}, {"company": "Acme Inc", "department": "Operations", "rate": "3.7%", "city": "Nairobi  ", "industry": "FipServ", "unit": "  MI"}, {"company": "Acme  Inc", "department": "ops", "rate": "67.0%", "city": "Nairobi  ", "industry": "Mfg", "unit": "milliliter"}, {"company": " Initech", "department": "hr", "rate": "41.7%", "city": "Yaring", "industry": "Ed", "unit": "ml"}, {"company": "Soylent Corp  ", "department": "Hr", "rate": "26.8%", "city": "SIDERNO", "industry": "Software", "unit": "mls"}, {"company": "Stark Industries", "department": "   FIN", "rate": "40.5%", "city": "Siderno", "industry": "Fintech", "unit": "kg"}, {"company": " Acme Inc", "department": " Accounting ", "rate": "94.3%", "city": "Dayal ", "industry": "industrial", "unit": "mi"}, {"company": "Hooli  ", "department": "HMUAN RESOURCES", "rate": "95.1%", "city": "Dayal", "industry": "ed ", "unit": "ml"}, {"company": "Vehement", "department": "Human Resources", "rate": "14.5%", "city": " Dayal ", "industry": "EdTech", "unit": "Mls  "}, {"company": "Hooli", "department": "Ops", "rate": "19.5%", "city": "SIDERNO", "industry": "Finance", "unit": "ml"}, {"company": "Umbrella", "department": "human resources", "rate": "33.5%", "city": "Yarig", "industry": "Ed", "unit": "milliliter"}, {"company": "Cyberdyne ", "department": "Operations", "rate": "78.9%", "city": "nairoib", "industry": "ed", "unit": "mls"}, {"company": " Vehement ", "department": "ACCOUNTING", "rate": "26.6%", "city": "Sideerno", "industry": "Tech", "unit": "   mls"}, {"company": "  Soylent Corp", "department": "   HUMAN RESOURCES", "rate": "98.1%", "city": "YARINNG", "industry": "Education", "unit": "kg"}, {"company": "Umbrella  ", "department": "Human Rosources", "rate": "59.7%", "city": "SIDERNO", "industry": "Banking", "unit": "kilograms"}, {"company": "Hooli  ", "department": "Operations", "rate": "0.5%", "city": "Siderno", "industry": "  Ed", "unit": "Kg"}, {"company": " Cyberdyne ", "department": "PEOPLE OPS", "rate": "92.7%", "city": "dyaal", "industry": "Ed  ", "unit": "kilograms"}, {"company": "   Soylent Corp", "department": "Accounting", "rate": "52.4%", "city": "Dayal", "industry": "Tech   ", "unit": " kg "}, {"company": "Hooli", "department": "Biz Dev ", "rate": "84.0%", "city": "nairobi", "industry": "it ", "unit": "kilograms"}, {"company": "Initech", "department": "ops", "rate": "75.1%", "city": " yaring ", "industry": "FinSeerv  ", "unit": "mi"}, {"company": "Initech", "department": "hr", "rate": "1.6%", "city": "   nairobi", "industry": "FinServ", "unit": "kg"}, {"company": "Globex", "department": "Operations", "rate": "25.8%", "city": "dayal", "industry": "Mfg  ", "unit": "kg"}, {"company": "Stark  Industries", "department": "OPS", "rate": "9.5%", "city": "Yaring", "industry": "Finnance", "unit": "ml"}, {"company": "   Soylent Corp", "department": "Accounting", "rate": "52.4%", "city": "Dayal", "industry": "Tech   ", "unit": " kg "}, {"company": "Hooli  ", "department": "Operations", "rate": "0.5%", "city": "Siderno", "industry": "  Ed", "unit": "Kg"}, {"company": "", "department": "", "rate": "", "city": "", "industry": "", "unit": ""}], "clean": [{"company": "Umbrella", "department": "Finance", "rate": 0.77, "city": "Dayal", "industry": "Education", "unit": "kg"}, {"company": "Cyberdyne", "department": "Human Resources", "rate": 0.07, "city": "Yaring", "industry": "Financial Services", "unit": "mi"}, {"company": "Umbrella", "department": "Finance", "rate": 0.64, "city": "Siderno", "industry": "Financial Services", "unit": "mL"}, {"company": "Umbrella", "department": "Finance", "rate": 0.5589999999999999, "city": "Yaring", "industry": "Manufacturing", "unit": "mi"}, {"company": "Globex", "department": "Human Resources", "rate": 0.16399999999999998, "city": "Siderno", "industry": "Manufacturing", "unit": "mi"}, {"company": "Cyberdyne", "department": "Operations", "rate": 0.282, "city": "Nairobi", "industry": "Education", "unit": "mL"}, {"company": "Cyberdyne", "department": "Sales", "rate": 0.8590000000000001, "city": "Nairobi", "industry": "Technology", "unit": "mL"}, {"company": "Hooli", "department": "Sales", "rate": 0.258, "city": "Nairobi", "industry": "Education", "unit": "mi"}, {"company": "Cyberdyne", "department": "Human Resources", "rate": 0.725, "city": "Siderno", "industry": "Financial Services", "unit": "kg"}, {"company": "Initech", "department": "Operations", "rate": 0.895, "city": "Dayal", "industry": "Technology", "unit": "mL"}, {"company": "Stark Industries", "department": "Finance", "rate": 0.752, "city": "Yaring", "industry": "Manufacturing", "unit": "kg"}, {"company": "Globex", "department": "Finance", "rate": 0.621, "city": "Yaring", "industry": "Financial Services", "unit": "mi"}, {"company": "Cyberdyne", "department": "Sales", "rate": 0.153, "city": "Nairobi", "industry": "Manufacturing", "unit": "mL"}, {"company": "Soylent Corp", "department": "Operations", "rate": 0.384, "city": "Siderno", "industry": "Education", "unit": "mL"}, {"company": "Acme Inc", "department": "Finance", "rate": 0.509, "city": "Siderno", "industry": "Manufacturing", "unit": "mi"}, {"company": "Acme Inc", "department": "Human Resources", "rate": 0.435, "city": "Nairobi", "industry": "Financial Services", "unit": "mi"}, {"company": "Acme Inc", "department": "Operations", "rate": 0.221, "city": "Nairobi", "industry": "Technology", "unit": "kg"}, {"company": "Stark Industries", "department": "Sales", "rate": 0.583, "city": "Yaring", "industry": "Technology", "unit": "mi"}, {"company": "Cyberdyne", "department": "Operations", "rate": 0.10800000000000001, "city": "Siderno", "industry": "Education", "unit": "kg"}, {"company": "Initech", "department": "Operations", "rate": 0.7809999999999999, "city": "Yaring", "industry": "Financial Services", "unit": "mi"}, {"company": "Cyberdyne", "department": "Sales", "rate": 0.345, "city": "Nairobi", "industry": "Technology", "unit": "mL"}, {"company": "Wonka Co", "department": "Operations", "rate": 0.312, "city": "Yaring", "industry": "Financial Services", "unit": "mL"}, {"company": "Cyberdyne", "department": "Operations", "rate": 0.033, "city": "Yaring", "industry": "Manufacturing", "unit": "kg"}, {"company": "Hooli", "department": "Human Resources", "rate": 0.433, "city": "Nairobi", "industry": "Technology", "unit": "mi"}, {"company": "Soylent Corp", "department": "Operations", "rate": 0.281, "city": "Nairobi", "industry": "Education", "unit": "kg"}, {"company": "Initech", "department": "Operations", "rate": 0.183, "city": "Yaring", "industry": "Manufacturing", "unit": "mL"}, {"company": "Acme Inc", "department": "Finance", "rate": 0.48, "city": "Dayal", "industry": "Technology", "unit": "mi"}, {"company": "Globex", "department": "Finance", "rate": 0.428, "city": "Nairobi", "industry": "Technology", "unit": "kg"}, {"company": "Stark Industries", "department": "Sales", "rate": 0.265, "city": "Dayal", "industry": "Education", "unit": "mL"}, {"company": "Umbrella", "department": "Human Resources", "rate": 0.774, "city": "Siderno", "industry": "Education", "unit": "mL"}, {"company": "Initech", "department": "Sales", "rate": 0.40399999999999997, "city": "Yaring", "industry": "Education", "unit": "kg"}, {"company": "Acme Inc", "department": "Sales", "rate": 0.79, "city": "Dayal", "industry": "Manufacturing", "unit": "mi"}, {"company": "Umbrella", "department": "Sales", "rate": 0.134, "city": "Dayal", "industry": "Financial Services", "unit": "kg"}, {"company": "Hooli", "department": "Sales", "rate": 0.262, "city": "Dayal", "industry": "Financial Services", "unit": "kg"}, {"company": "Soylent Corp", "department": "Human Resources", "rate": 0.262, "city": "Yaring", "industry": "Education", "unit": "mL"}, {"company": "Umbrella", "department": "Operations", "rate": 0.778, "city": "Siderno", "industry": "Financial Services", "unit": "mL"}, {"company": "Umbrella", "department": "Finance", "rate": 0.982, "city": "Dayal", "industry": "Technology", "unit": "mi"}, {"company": "Cyberdyne", "department": "Finance", "rate": 0.902, "city": "Yaring", "industry": "Technology", "unit": "kg"}, {"company": "Acme Inc", "department": "Operations", "rate": 0.037000000000000005, "city": "Nairobi", "industry": "Financial Services", "unit": "mi"}, {"company": "Acme Inc", "department": "Operations", "rate": 0.67, "city": "Nairobi", "industry": "Manufacturing", "unit": "mL"}, {"company": "Initech", "department": "Human Resources", "rate": 0.41700000000000004, "city": "Yaring", "industry": "Education", "unit": "mL"}, {"company": "Soylent Corp", "department": "Human Resources", "rate": 0.268, "city": "Siderno", "industry": "Technology", "unit": "mL"}, {"company": "Stark Industries", "department": "Finance", "rate": 0.405, "city": "Siderno", "industry": "Financial Services", "unit": "kg"}, {"company": "Acme Inc", "department": "Finance", "rate": 0.943, "city": "Dayal", "industry": "Manufacturing", "unit": "mi"}, {"company": "Hooli", "department": "Human Resources", "rate": 0.951, "city": "Dayal", "industry": "Education", "unit": "mL"}, {"company": "Vehement", "department": "Human Resources", "rate": 0.145, "city": "Dayal", "industry": "Education", "unit": "mL"}, {"company": "Hooli", "department": "Operations", "rate": 0.195, "city": "Siderno", "industry": "Financial Services", "unit": "mL"}, {"company": "Umbrella", "department": "Human Resources", "rate": 0.335, "city": "Yaring", "industry": "Education", "unit": "mL"}, {"company": "Cyberdyne", "department": "Operations", "rate": 0.789, "city": "Nairobi", "industry": "Education", "unit": "mL"}, {"company": "Vehement", "department": "Finance", "rate": 0.266, "city": "Siderno", "industry": "Technology", "unit": "mL"}, {"company": "Soylent Corp", "department": "Human Resources", "rate": 0.981, "city": "Yaring", "industry": "Education", "unit": "kg"}, {"company": "Umbrella", "department": "Human Resources", "rate": 0.597, "city": "Siderno", "industry": "Financial Services", "unit": "kg"}, {"company": "Hooli", "department": "Operations", "rate": 0.005, "city": "Siderno", "industry": "Education", "unit": "kg"}, {"company": "Cyberdyne", "department": "Human Resources", "rate": 0.927, "city": "Dayal", "industry": "Education", "unit": "kg"}, {"company": "Soylent Corp", "department": "Finance", "rate": 0.524, "city": "Dayal", "industry": "Technology", "unit": "kg"}, {"company": "Hooli", "department": "Sales", "rate": 0.84, "city": "Nairobi", "industry": "Technology", "unit": "kg"}, {"company": "Initech", "department": "Operations", "rate": 0.7509999999999999, "city": "Yaring", "industry": "Financial Services", "unit": "mi"}, {"company": "Initech", "department": "Human Resources", "rate": 0.016, "city": "Nairobi", "industry": "Financial Services", "unit": "kg"}, {"company": "Globex", "department": "Operations", "rate": 0.258, "city": "Dayal", "industry": "Manufacturing", "unit": "kg"}, {"company": "Stark Industries", "department": "Operations", "rate": 0.095, "city": "Yaring", "industry": "Financial Services", "unit": "mL"}], "dirty_cols": ["company", "department", "rate", "city", "industry", "unit"], "clean_cols": ["company", "department", "rate", "city", "industry", "unit"], "plan": {"dataset_summary": "63 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"FINANCE": "Finance", "human resources": "Human Resources", "accounting": "Finance", "ops": "Operations", "BIZ DEV": "Sales", "Biz Dev": "Sales", "people": "Human Resources", "operations": "Operations", "Fin": "Finance", "fin": "Finance", "People": "Human Resources", "OPERATIONS": "Operations", "Ops": "Operations", "OPS": "Operations", "Accounting": "Finance", "HR": "Human Resources", "sales": "Sales", "SALES": "Sales", "hr": "Human Resources", "Hr": "Human Resources", "FIN": "Finance", "HMUAN RESOURCES": "Human Resources", "ACCOUNTING": "Finance", "HUMAN RESOURCES": "Human Resources", "Human Rosources": "Human Resources", "PEOPLE OPS": "Human Resources"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Daal": "Dayal", "nairobi": "Nairobi", "siderno": "Siderno", "YARING": "Yaring", "Nairrobi": "Nairobi", "Nairoi": "Nairobi", "Yarring": "Yaring", "Yariing": "Yaring", "yaring": "Yaring", "YRAING": "Yaring", "SIDERNO": "Siderno", "dayal": "Dayal", "Yarnig": "Yaring", "Yarig": "Yaring", "nairoib": "Nairobi", "Sideerno": "Siderno", "YARINNG": "Yaring", "dyaal": "Dayal"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ed": "Education", "Banking": "Financial Services", "FINTECH": "Financial Services", "Mfg": "Manufacturing", "Ed": "Education", "Software": "Technology", "MANUFACTURING": "Manufacturing", "manufacturing": "Manufacturing", "EdTech": "Education", "finserv": "Financial Services", "Softlare": "Technology", "Tech": "Technology", "Educaion": "Education", "Financial Serices": "Financial Services", "It": "Technology", "Finance": "Financial Services", "tech": "Technology", "mfg": "Manufacturing", "Tch": "Technology", "TECHNOLOGY": "Technology", "education": "Education", "EDECH": "Education", "BANKING": "Financial Services", "Fintech": "Financial Services", "Edtech": "Education", "software": "Technology", "FipServ": "Financial Services", "industrial": "Manufacturing", "it": "Technology", "FinSeerv": "Financial Services", "FinServ": "Financial Services", "Finnance": "Financial Services"}, "rationale": "Unified 32 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilograms": "kg", "Mi": "mi", "MILLILITER": "mL", "Miwes": "mi", "milliliter": "mL", "MLS": "mL", "kilogram": "kg", "Mls": "mL", "KG": "kg", "miles": "mi", "ml": "mL", "ML": "mL", "MI": "mi", "Kilogram": "kg", "mls": "mL", "MILE": "mi", "Kg": "kg", "KILOGRAMS": "kg"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "sy", "rate": "49.0%", "signup_date": "9/11/2023", "city": "khandaw  ", "status": " free trial", "department": "IT", "industry": "Software", "is_active": "0", "unnamed": ""}, {"country": "NEPAL", "rate": "25.9%", "signup_date": "2023-06-11", "city": "Maple Ridge", "status": "   on old", "department": "Mkt", "industry": "Retail  ", "is_active": "1", "unnamed": ""}, {"country": "syr", "rate": "75.7%", "signup_date": "2023-01-08", "city": "Khandna", "status": "ACTIVE", "department": "Cust Support", "industry": "tech", "is_active": "true", "unnamed": ""}, {"country": "Nepal", "rate": "53.5%", "signup_date": "8/13/2023", "city": "Maple Ridge", "status": "Active", "department": "IT", "industry": " E-commerce ", "is_active": "Y", "unnamed": ""}, {"country": "SX", "rate": "65.8%", "signup_date": "2023-10-27", "city": "maple ridge   ", "status": "TIRAL", "department": "information technology", "industry": "RETAIL & E-COMMERCE", "is_active": "1", "unnamed": ""}, {"country": "NP", "rate": "2.1%", "signup_date": "5/22/2023", "city": "Khandwwa", "status": "  on hold", "department": "Information Technology", "industry": "E-commerce", "is_active": "FALSE", "unnamed": ""}, {"country": "Sint Maarten (Dutch Part) ", "rate": "93.2%", "signup_date": "8 Sep 2023", "city": "Maple Ridge", "status": "paused", "department": "CS", "industry": "RETAIL & E-COMEMRCE ", "is_active": "false", "unnamed": ""}, {"country": "SY", "rate": "88.8%", "signup_date": "3/12/2023", "city": "mapl ridge", "status": "paused", "department": "cs", "industry": "tech", "is_active": "1", "unnamed": ""}, {"country": "Syrian Arab Republic", "rate": "72.8%", "signup_date": "45114", "city": "  MAPLE RIDGE", "status": "on hold", "department": "SUPPORT", "industry": "tech", "is_active": "0", "unnamed": ""}, {"country": "   Sint Maarten (Dutch part)", "rate": "47.5%", "signup_date": "5/22/2023", "city": "MOYALE", "status": "pased", "department": " information technology", "industry": "Retail", "is_active": "T", "unnamed": ""}, {"country": "SXM", "rate": "59.1%", "signup_date": "45215", "city": "Moale", "status": "TRIAL", "department": "Mkt", "industry": "E-Commerce", "is_active": "F", "unnamed": ""}, {"country": "sxm", "rate": "24.6%", "signup_date": "2023-04-17", "city": "kdandwa   ", "status": " paused ", "department": "Information Technology", "industry": "Technology", "is_active": "Y", "unnamed": ""}, {"country": "SY", "rate": "71.7%", "signup_date": "8/12/2023", "city": "MAPLE RIDEG", "status": "Paused", "department": "Information Tecnology", "industry": "IT", "is_active": "Y", "unnamed": ""}, {"country": " Nepaal ", "rate": "67.1%", "signup_date": "5/7/2023", "city": "   Maple Ridge", "status": "trial", "department": "CS", "industry": "Retail", "is_active": "true", "unnamed": ""}, {"country": "   Sy", "rate": "99.1%", "signup_date": "2023-01-23", "city": "MAPEL RIDGE", "status": "Trial", "department": "  information technology", "industry": "RETAIL", "is_active": "Yes", "unnamed": ""}, {"country": "SXM", "rate": "9.2%", "signup_date": "17 Dec 2023", "city": "Moyale", "status": "on hold", "department": "   MKTG", "industry": "RETAIL & E-COMMERCE", "is_active": "No", "unnamed": ""}, {"country": " sy ", "rate": "84.4%", "signup_date": "12 Jan 2023", "city": "moyale", "status": "active", "department": "MKT", "industry": "Retail", "is_active": "T", "unnamed": ""}, {"country": "Nepal", "rate": "56.5%", "signup_date": "2023-01-27", "city": "Maple Rdge", "status": "active", "department": "mkt", "industry": "it", "is_active": "0", "unnamed": ""}, {"country": "DEU", "rate": "13.8%", "signup_date": "12/5/2023", "city": "Maple Ridge", "status": "Active", "department": "marketing", "industry": "E-commerce", "is_active": "T", "unnamed": ""}, {"country": "Germany ", "rate": "82.3%", "signup_date": "12/6/2023", "city": "Moyale", "status": "Active", "department": "Mkt", "industry": " tceh ", "is_active": "N", "unnamed": ""}, {"country": "deu", "rate": "62.9%", "signup_date": "18 Nov 2023", "city": "Khandwa", "status": " Active ", "department": "Inormation  Technology", "industry": "e-commecre", "is_active": "true", "unnamed": ""}, {"country": "Nepal", "rate": "85.5%", "signup_date": "45087", "city": "   moyale", "status": "free trial", "department": "CS", "industry": "TECH", "is_active": "No", "unnamed": ""}, {"country": "DEU", "rate": "46.2%", "signup_date": "15 Jan 2023", "city": " khandwa ", "status": "trial", "department": "I.T.", "industry": "Retail & E-commerce   ", "is_active": "N", "unnamed": ""}, {"country": "Sint Maarten (Dutch part)", "rate": "28.3%", "signup_date": "45160", "city": "maple ridge", "status": "TIRAL", "department": "i..", "industry": "Retail", "is_active": "T", "unnamed": ""}, {"country": "NP", "rate": "94.1%", "signup_date": "8 Jun 2023", "city": "Moyale", "status": "active", "department": "Information Technology", "industry": "TECCH", "is_active": "false", "unnamed": ""}, {"country": "NPL   ", "rate": "8.7%", "signup_date": "2023-11-09", "city": "  Maple Ridge", "status": "active", "department": "Information Technology", "industry": "Retail", "is_active": "FALSE", "unnamed": ""}, {"country": "germany", "rate": "3.9%", "signup_date": "6/10/2023", "city": "Khandwa", "status": "paused", "department": "CS   ", "industry": "IT", "is_active": "F", "unnamed": ""}, {"country": "sxm", "rate": "56.2%", "signup_date": "2023-10-11", "city": " Khandwa ", "status": "trial", "department": "I.T.", "industry": "SOFTWARE", "is_active": "Yes", "unnamed": ""}, {"country": "Sint Maarten (Dutch part)", "rate": "25.5%", "signup_date": "2023-11-05", "city": "Kahndwa", "status": "aczive", "department": " Mktg ", "industry": "E-commerce", "is_active": "Yes", "unnamed": ""}, {"country": "syrian arab republic", "rate": "59.8%", "signup_date": "2023-03-01", "city": "Maple Ridge", "status": "trial", "department": "GROWTH   ", "industry": "e-commerce", "is_active": "TRUE", "unnamed": ""}, {"country": "syr", "rate": "41.4%", "signup_date": "2023-10-07", "city": "Khandwa", "status": "trial", "department": "IT   ", "industry": "TECH", "is_active": "N", "unnamed": ""}, {"country": " NP", "rate": "90.1%", "signup_date": "3/18/2023", "city": "khandwa", "status": "active", "department": "IT", "industry": "   e-comberce", "is_active": "N", "unnamed": ""}, {"country": "SXM", "rate": "38.7%", "signup_date": "11 Nov 2023", "city": "Khandwa", "status": "on hold", "department": " I.T. ", "industry": "Retail", "is_active": "FALSE", "unnamed": ""}, {"country": "Nepal   ", "rate": "60.4%", "signup_date": "45071", "city": "Maplle Ridge", "status": " ACTIVE ", "department": "Information Technology ", "industry": " Retail", "is_active": "T", "unnamed": ""}, {"country": "Syrian Arab Republic", "rate": "28.6%", "signup_date": "9 Mar 2023", "city": "khandwa", "status": "Active", "department": "I..T.", "industry": " tech", "is_active": "true", "unnamed": ""}, {"country": "SYRIAN ARAB REPUBLIC", "rate": "48.4%", "signup_date": "10 Jun 2023", "city": " khanndwa ", "status": "ACTIVE", "department": "Mkt", "industry": "  tceh", "is_active": "FALSE", "unnamed": ""}, {"country": "Federal Demoocratic Republic of Nepal", "rate": "86.4%", "signup_date": "45240", "city": "Moyale", "status": "paused", "department": "   mkt", "industry": "tehnology", "is_active": "No", "unnamed": ""}, {"country": "NPL", "rate": "0.7%", "signup_date": "1 Mar 2023", "city": "Moyale", "status": "active", "department": "Mktg", "industry": "Retail", "is_active": "No", "unnamed": ""}, {"country": "Syrian Arab Republic", "rate": "72.8%", "signup_date": "45114", "city": "  MAPLE RIDGE", "status": "on hold", "department": "SUPPORT", "industry": "tech", "is_active": "0", "unnamed": ""}, {"country": "", "rate": "", "signup_date": "", "city": "", "status": "", "department": "", "industry": "", "is_active": "", "unnamed": ""}], "clean": [{"country": "Syrian Arab Republic", "rate": 0.49, "signup_date": "2023-09-11", "city": "Khandwa", "status": "Trial", "department": "Information Technology", "industry": "Technology", "is_active": false}, {"country": "Nepal", "rate": 0.259, "signup_date": "2023-06-11", "city": "Maple Ridge", "status": "Paused", "department": "Marketing", "industry": "Retail", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.757, "signup_date": "2023-01-08", "city": "Khandwa", "status": "Active", "department": "Customer Support", "industry": "Technology", "is_active": true}, {"country": "Nepal", "rate": 0.535, "signup_date": "2023-08-13", "city": "Maple Ridge", "status": "Active", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Sint Maarten (Dutch part)", "rate": 0.6579999999999999, "signup_date": "2023-10-27", "city": "Maple Ridge", "status": "Trial", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Nepal", "rate": 0.021, "signup_date": "2023-05-22", "city": "Khandwa", "status": "Paused", "department": "Information Technology", "industry": "Retail", "is_active": false}, {"country": "Sint Maarten (Dutch part)", "rate": 0.932, "signup_date": "2023-09-08", "city": "Maple Ridge", "status": "Paused", "department": "Customer Support", "industry": "Retail", "is_active": false}, {"country": "Syrian Arab Republic", "rate": 0.888, "signup_date": "2023-03-12", "city": "Maple Ridge", "status": "Paused", "department": "Customer Support", "industry": "Technology", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.728, "signup_date": "2023-07-07", "city": "Maple Ridge", "status": "Paused", "department": "Customer Support", "industry": "Technology", "is_active": false}, {"country": "Sint Maarten (Dutch part)", "rate": 0.475, "signup_date": "2023-05-22", "city": "Moyale", "status": "Paused", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Sint Maarten (Dutch part)", "rate": 0.591, "signup_date": "2023-10-16", "city": "Moyale", "status": "Trial", "department": "Marketing", "industry": "Retail", "is_active": false}, {"country": "Sint Maarten (Dutch part)", "rate": 0.24600000000000002, "signup_date": "2023-04-17", "city": "Khandwa", "status": "Paused", "department": "Information Technology", "industry": "Technology", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.7170000000000001, "signup_date": "2023-08-12", "city": "Maple Ridge", "status": "Paused", "department": "Information Technology", "industry": "Technology", "is_active": true}, {"country": "Nepal", "rate": 0.6709999999999999, "signup_date": "2023-05-07", "city": "Maple Ridge", "status": "Trial", "department": "Customer Support", "industry": "Retail", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.991, "signup_date": "2023-01-23", "city": "Maple Ridge", "status": "Trial", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Sint Maarten (Dutch part)", "rate": 0.092, "signup_date": "2023-12-17", "city": "Moyale", "status": "Paused", "department": "Marketing", "industry": "Retail", "is_active": false}, {"country": "Syrian Arab Republic", "rate": 0.8440000000000001, "signup_date": "2023-01-12", "city": "Moyale", "status": "Active", "department": "Marketing", "industry": "Retail", "is_active": true}, {"country": "Nepal", "rate": 0.565, "signup_date": "2023-01-27", "city": "Maple Ridge", "status": "Active", "department": "Marketing", "industry": "Technology", "is_active": false}, {"country": "Germany", "rate": 0.138, "signup_date": "2023-12-05", "city": "Maple Ridge", "status": "Active", "department": "Marketing", "industry": "Retail", "is_active": true}, {"country": "Germany", "rate": 0.823, "signup_date": "2023-12-06", "city": "Moyale", "status": "Active", "department": "Marketing", "industry": "Technology", "is_active": false}, {"country": "Germany", "rate": 0.629, "signup_date": "2023-11-18", "city": "Khandwa", "status": "Active", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Nepal", "rate": 0.855, "signup_date": "2023-06-10", "city": "Moyale", "status": "Trial", "department": "Customer Support", "industry": "Technology", "is_active": false}, {"country": "Germany", "rate": 0.462, "signup_date": "2023-01-15", "city": "Khandwa", "status": "Trial", "department": "Information Technology", "industry": "Retail", "is_active": false}, {"country": "Sint Maarten (Dutch part)", "rate": 0.28300000000000003, "signup_date": "2023-08-22", "city": "Maple Ridge", "status": "Trial", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Nepal", "rate": 0.941, "signup_date": "2023-06-08", "city": "Moyale", "status": "Active", "department": "Information Technology", "industry": "Technology", "is_active": false}, {"country": "Nepal", "rate": 0.087, "signup_date": "2023-11-09", "city": "Maple Ridge", "status": "Active", "department": "Information Technology", "industry": "Retail", "is_active": false}, {"country": "Germany", "rate": 0.039, "signup_date": "2023-06-10", "city": "Khandwa", "status": "Paused", "department": "Customer Support", "industry": "Technology", "is_active": false}, {"country": "Sint Maarten (Dutch part)", "rate": 0.562, "signup_date": "2023-10-11", "city": "Khandwa", "status": "Trial", "department": "Information Technology", "industry": "Technology", "is_active": true}, {"country": "Sint Maarten (Dutch part)", "rate": 0.255, "signup_date": "2023-11-05", "city": "Khandwa", "status": "Active", "department": "Marketing", "industry": "Retail", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.598, "signup_date": "2023-03-01", "city": "Maple Ridge", "status": "Trial", "department": "Marketing", "industry": "Retail", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.414, "signup_date": "2023-10-07", "city": "Khandwa", "status": "Trial", "department": "Information Technology", "industry": "Technology", "is_active": false}, {"country": "Nepal", "rate": 0.9009999999999999, "signup_date": "2023-03-18", "city": "Khandwa", "status": "Active", "department": "Information Technology", "industry": "Retail", "is_active": false}, {"country": "Sint Maarten (Dutch part)", "rate": 0.387, "signup_date": "2023-11-11", "city": "Khandwa", "status": "Paused", "department": "Information Technology", "industry": "Retail", "is_active": false}, {"country": "Nepal", "rate": 0.604, "signup_date": "2023-05-25", "city": "Maple Ridge", "status": "Active", "department": "Information Technology", "industry": "Retail", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.28600000000000003, "signup_date": "2023-03-09", "city": "Khandwa", "status": "Active", "department": "Information Technology", "industry": "Technology", "is_active": true}, {"country": "Syrian Arab Republic", "rate": 0.484, "signup_date": "2023-06-10", "city": "Khandwa", "status": "Active", "department": "Marketing", "industry": "Technology", "is_active": false}, {"country": "Nepal", "rate": 0.8640000000000001, "signup_date": "2023-11-10", "city": "Moyale", "status": "Paused", "department": "Marketing", "industry": "Technology", "is_active": false}, {"country": "Nepal", "rate": 0.006999999999999999, "signup_date": "2023-03-01", "city": "Moyale", "status": "Active", "department": "Marketing", "industry": "Retail", "is_active": false}], "dirty_cols": ["country", "rate", "signup_date", "city", "status", "department", "industry", "is_active", "unnamed"], "clean_cols": ["country", "rate", "signup_date", "city", "status", "department", "industry", "is_active"], "plan": {"dataset_summary": "40 rows × 9 columns. 8 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sy": "Syrian Arab Republic", "NEPAL": "Nepal", "syr": "Syrian Arab Republic", "SX": "Sint Maarten (Dutch part)", "NP": "Nepal", "Sint Maarten (Dutch Part)": "Sint Maarten (Dutch part)", "SY": "Syrian Arab Republic", "SXM": "Sint Maarten (Dutch part)", "sxm": "Sint Maarten (Dutch part)", "Nepaal": "Nepal", "Sy": "Syrian Arab Republic", "DEU": "Germany", "deu": "Germany", "NPL": "Nepal", "germany": "Germany", "syrian arab republic": "Syrian Arab Republic", "SYRIAN ARAB REPUBLIC": "Syrian Arab Republic", "Federal Demoocratic Republic of Nepal": "Nepal"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"khandaw": "Khandwa", "Khandna": "Khandwa", "maple ridge": "Maple Ridge", "Khandwwa": "Khandwa", "mapl ridge": "Maple Ridge", "MAPLE RIDGE": "Maple Ridge", "MOYALE": "Moyale", "Moale": "Moyale", "kdandwa": "Khandwa", "MAPLE RIDEG": "Maple Ridge", "MAPEL RIDGE": "Maple Ridge", "moyale": "Moyale", "Maple Rdge": "Maple Ridge", "khandwa": "Khandwa", "Kahndwa": "Khandwa", "Maplle Ridge": "Maple Ridge", "khanndwa": "Khandwa"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"free trial": "Trial", "on old": "Paused", "ACTIVE": "Active", "TIRAL": "Trial", "on hold": "Paused", "paused": "Paused", "pased": "Paused", "TRIAL": "Trial", "trial": "Trial", "active": "Active", "aczive": "Active"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "Mkt": "Marketing", "Cust Support": "Customer Support", "information technology": "Information Technology", "CS": "Customer Support", "cs": "Customer Support", "SUPPORT": "Customer Support", "Information Tecnology": "Information Technology", "MKTG": "Marketing", "MKT": "Marketing", "mkt": "Marketing", "marketing": "Marketing", "Inormation Technology": "Information Technology", "I.T.": "Information Technology", "i..": "Information Technology", "Mktg": "Marketing", "GROWTH": "Marketing", "I..T.": "Information Technology"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Software": "Technology", "tech": "Technology", "E-commerce": "Retail", "RETAIL & E-COMMERCE": "Retail", "RETAIL & E-COMEMRCE": "Retail", "E-Commerce": "Retail", "IT": "Technology", "RETAIL": "Retail", "it": "Technology", "tceh": "Technology", "e-commecre": "Retail", "TECH": "Technology", "Retail & E-commerce": "Retail", "TECCH": "Technology", "SOFTWARE": "Technology", "e-commerce": "Retail", "e-comberce": "Retail", "tehnology": "Technology"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"rate": "94.2%", "unit": " mL ", "name": " David Reyes", "email": "DFLONBF@TEST.ORG "}, {"rate": "39.7%", "unit": "Milliliter", "name": "Alice Fischer", "email": "fale@mail.com"}, {"rate": "36.8%", "unit": "milliliter ", "name": "Eve Khan", "email": "cnhdmkf@example.com  "}, {"rate": "79.7%", "unit": "ml", "name": "Frank Reyes", "email": " ndblo@example.com"}, {"rate": "46.0%", "unit": "MLS", "name": "Frank Petrov ", "email": "hfcipm@mail.com"}, {"rate": "34.8%", "unit": "   mls", "name": "Mona  Adams", "email": "bbkc@example.com"}, {"rate": "98.8%", "unit": "ml", "name": "Karl Reyes ", "email": "DNFIGBI@MAIL.COM"}, {"rate": "57.8%", "unit": " pounds", "name": "David Diaz", "email": "gdlf@test.org"}, {"rate": "82.1%", "unit": "mL", "name": "Sara Johnson", "email": "  MFLDBIH@CORP.IO"}, {"rate": "74.2%", "unit": "ml", "name": "Mona Ali", "email": "PBJNB@MAIL.COM"}, {"rate": "47.4%", "unit": " pounds ", "name": "Frank  Diaz", "email": "NCEP@EXAMPLE.COM"}, {"rate": "16.7%", "unit": "  mls", "name": "Judy Moore", "email": "  KLFN@TEST.ORG"}, {"rate": "92.5%", "unit": "LB", "name": "Carol Ali", "email": " nooanb@mail.com "}, {"rate": "97.2%", "unit": "pound", "name": "Frank Moore", "email": " ECMDAN@MAIL.COM "}, {"rate": "74.9%", "unit": " mls", "name": "Lena Johnson", "email": " CMBH@CORP.IO "}, {"rate": "59.7%", "unit": "milliliter", "name": "Frank Brandt", "email": " cilco@example.com "}, {"rate": "43.1%", "unit": "pounds", "name": "   Bob Brandt", "email": "cnio@corp.io"}, {"rate": "90.9%", "unit": "ml", "name": "Lena Wong", "email": " jijida@mail.com "}, {"rate": "56.7%", "unit": "milliliter", "name": "Frank Khan", "email": "ONNEH@TEST.ORG"}, {"rate": "34.8%", "unit": "   mls", "name": "Mona  Adams", "email": "bbkc@example.com"}], "clean": [{"rate": 0.9420000000000001, "unit": "mL", "name": "David Reyes", "email": "dflonbf@test.org"}, {"rate": 0.397, "unit": "mL", "name": "Alice Fischer", "email": "fale@mail.com"}, {"rate": 0.368, "unit": "mL", "name": "Eve Khan", "email": "cnhdmkf@example.com"}, {"rate": 0.797, "unit": "mL", "name": "Frank Reyes", "email": "ndblo@example.com"}, {"rate": 0.46, "unit": "mL", "name": "Frank Petrov", "email": "hfcipm@mail.com"}, {"rate": 0.348, "unit": "mL", "name": "Mona Adams", "email": "bbkc@example.com"}, {"rate": 0.988, "unit": "mL", "name": "Karl Reyes", "email": "dnfigbi@mail.com"}, {"rate": 0.578, "unit": "lb", "name": "David Diaz", "email": "gdlf@test.org"}, {"rate": 0.821, "unit": "mL", "name": "Sara Johnson", "email": "mfldbih@corp.io"}, {"rate": 0.742, "unit": "mL", "name": "Mona Ali", "email": "pbjnb@mail.com"}, {"rate": 0.474, "unit": "lb", "name": "Frank Diaz", "email": "ncep@example.com"}, {"rate": 0.16699999999999998, "unit": "mL", "name": "Judy Moore", "email": "klfn@test.org"}, {"rate": 0.925, "unit": "lb", "name": "Carol Ali", "email": "nooanb@mail.com"}, {"rate": 0.972, "unit": "lb", "name": "Frank Moore", "email": "ecmdan@mail.com"}, {"rate": 0.7490000000000001, "unit": "mL", "name": "Lena Johnson", "email": "cmbh@corp.io"}, {"rate": 0.597, "unit": "mL", "name": "Frank Brandt", "email": "cilco@example.com"}, {"rate": 0.431, "unit": "lb", "name": "Bob Brandt", "email": "cnio@corp.io"}, {"rate": 0.909, "unit": "mL", "name": "Lena Wong", "email": "jijida@mail.com"}, {"rate": 0.5670000000000001, "unit": "mL", "name": "Frank Khan", "email": "onneh@test.org"}], "dirty_cols": ["rate", "unit", "name", "email"], "clean_cols": ["rate", "unit", "name", "email"], "plan": {"dataset_summary": "20 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Milliliter": "mL", "milliliter": "mL", "ml": "mL", "MLS": "mL", "mls": "mL", "pounds": "lb", "LB": "lb", "pound": "lb"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"is_active": "false", "name": "Eve Park", "phone": "506-772-5806", "company": "Hooli   ", "col_x": ""}, {"is_active": "false", "name": "David Fischer", "phone": "(559)619-2748", "company": "Vehement", "col_x": ""}, {"is_active": "F", "name": "   Grace Smith", "phone": "5828080193", "company": "Acme Inc", "col_x": ""}, {"is_active": "T", "name": "Frank  Moore", "phone": "5006638049", "company": "Acme Inc", "col_x": ""}, {"is_active": "1", "name": "Omar Novak", "phone": "536.060.1800", "company": " Soylent Corp", "col_x": ""}, {"is_active": "T", "name": "Mona Novak", "phone": "507-099-7593", "company": "Umbrella   ", "col_x": ""}, {"is_active": "No", "name": "  Bob Fischer", "phone": "(547)551-3029", "company": " Globex ", "col_x": ""}, {"is_active": "F", "name": "Sara Ali", "phone": "541-705-2969", "company": "Soylent Corp", "col_x": ""}, {"is_active": "Y", "name": "Judy Johnson", "phone": "5846324910", "company": "Cyberdyne", "col_x": ""}, {"is_active": "Y", "name": "Ivan Park", "phone": "5543188010", "company": "Initech", "col_x": ""}, {"is_active": "true", "name": "Mona  Brandt", "phone": "582-592-3993", "company": "Stark Industries", "col_x": ""}, {"is_active": "No", "name": "Frank Khan", "phone": "578-013-4233", "company": "Initech", "col_x": ""}, {"is_active": "false", "name": "Eve Park", "phone": "506-772-5806", "company": "Hooli   ", "col_x": ""}], "clean": [{"is_active": false, "name": "Eve Park", "phone": "(506) 772-5806", "company": "Hooli"}, {"is_active": false, "name": "David Fischer", "phone": "(559) 619-2748", "company": "Vehement"}, {"is_active": false, "name": "Grace Smith", "phone": "(582) 808-0193", "company": "Acme Inc"}, {"is_active": true, "name": "Frank Moore", "phone": "(500) 663-8049", "company": "Acme Inc"}, {"is_active": true, "name": "Omar Novak", "phone": "(536) 060-1800", "company": "Soylent Corp"}, {"is_active": true, "name": "Mona Novak", "phone": "(507) 099-7593", "company": "Umbrella"}, {"is_active": false, "name": "Bob Fischer", "phone": "(547) 551-3029", "company": "Globex"}, {"is_active": false, "name": "Sara Ali", "phone": "(541) 705-2969", "company": "Soylent Corp"}, {"is_active": true, "name": "Judy Johnson", "phone": "(584) 632-4910", "company": "Cyberdyne"}, {"is_active": true, "name": "Ivan Park", "phone": "(554) 318-8010", "company": "Initech"}, {"is_active": true, "name": "Mona Brandt", "phone": "(582) 592-3993", "company": "Stark Industries"}, {"is_active": false, "name": "Frank Khan", "phone": "(578) 013-4233", "company": "Initech"}], "dirty_cols": ["is_active", "name", "phone", "company", "col_x"], "clean_cols": ["is_active", "name", "phone", "company"], "plan": {"dataset_summary": "13 rows × 5 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"signup_date": "45062", "rate": "63.2%", "industry": "Industrial", "unit": " LB ", "amount": "$5,569.90", "notes2": ""}, {"signup_date": "9/11/2023", "rate": "71.3%", "industry": "real estate", "unit": "l", "amount": "$3,672.63", "notes2": ""}, {"signup_date": "2023-10-09", "rate": "15.5%", "industry": "Health", "unit": "centimeters", "amount": "$112.87", "notes2": ""}, {"signup_date": "2023-02-25", "rate": "16.3%", "industry": "MEDICAL", "unit": "pounds", "amount": "2.732,89", "notes2": ""}, {"signup_date": "45171", "rate": "49.6%", "industry": "proptech  ", "unit": "centimeters", "amount": "$8,000.73", "notes2": ""}, {"signup_date": "5/24/2023", "rate": "32.4%", "industry": "Medical", "unit": " litrre", "amount": "5.250,01", "notes2": ""}, {"signup_date": "17 Mar 2023", "rate": "3.5%", "industry": "real estate", "unit": "liters", "amount": "$8,795.69", "notes2": ""}, {"signup_date": "2023-02-27", "rate": "40.8%", "industry": "Travel", "unit": "litrs", "amount": "4.121,32", "notes2": ""}, {"signup_date": "20 Apr 2023", "rate": "59.3%", "industry": "Industrial", "unit": "lb", "amount": "$5,697.56", "notes2": ""}, {"signup_date": "2023-10-11", "rate": "9.3%", "industry": "Manufacturing", "unit": "  L", "amount": "$4,354.37", "notes2": ""}, {"signup_date": "12/20/2023", "rate": "28.2%", "industry": "Manufacturing", "unit": "pound", "amount": "?", "notes2": ""}, {"signup_date": "4 May 2023", "rate": "94.8%", "industry": "  proptech", "unit": "Centimeter", "amount": "1.916,11", "notes2": ""}, {"signup_date": "45192", "rate": "47.2%", "industry": "PropTech", "unit": "LB", "amount": "7.497,83", "notes2": ""}, {"signup_date": "1/9/2023", "rate": "37.0%", "industry": "Mfg", "unit": "Cm", "amount": "$1,290.53", "notes2": ""}, {"signup_date": "12/16/2023", "rate": "98.4%", "industry": "Manufacturing", "unit": "Lb", "amount": "$2,883.05", "notes2": ""}, {"signup_date": "8/1/2023", "rate": "43.0%", "industry": "Health Care", "unit": "   L", "amount": "TBD", "notes2": ""}, {"signup_date": "2023-04-24", "rate": "97.4%", "industry": "TOURISM", "unit": " Lb ", "amount": "7.929,39", "notes2": ""}, {"signup_date": "2023-09-22", "rate": "64.7%", "industry": "Health", "unit": "CENTIMETERS", "amount": "null", "notes2": ""}, {"signup_date": "4 Feb 2023", "rate": "70.9%", "industry": " proptech ", "unit": "liter", "amount": "$1,961.54", "notes2": ""}, {"signup_date": "45011", "rate": "99.6%", "industry": "Mfg ", "unit": "CM", "amount": "TBD", "notes2": ""}, {"signup_date": "13 Feb 2023", "rate": "76.5%", "industry": "tourism", "unit": "  centimeters", "amount": "$7,806.27", "notes2": ""}, {"signup_date": "45243", "rate": "98.8%", "industry": "medical", "unit": "l", "amount": "$5,569.35", "notes2": ""}, {"signup_date": "2023-08-12", "rate": "10.3%", "industry": "REAL ESTATE", "unit": "lb", "amount": "8.777,91", "notes2": ""}, {"signup_date": "45190", "rate": "92.0%", "industry": "   Mfg", "unit": "Cm", "amount": "1.097,02", "notes2": ""}, {"signup_date": "18 May 2023", "rate": "59.5%", "industry": "Tourism", "unit": " liter ", "amount": "4.274,05", "notes2": ""}, {"signup_date": "9/1/2023", "rate": "24.8%", "industry": "REALESTATE", "unit": "l", "amount": "-100", "notes2": ""}, {"signup_date": "10 Apr 2023", "rate": "78.0%", "industry": "realestale", "unit": "Cm", "amount": "-", "notes2": ""}, {"signup_date": "2023-06-20", "rate": "86.7%", "industry": "Mfg", "unit": " pounds", "amount": "-", "notes2": ""}, {"signup_date": "4/16/2023", "rate": "25.7%", "industry": "Medical", "unit": "LITR", "amount": "--", "notes2": ""}, {"signup_date": "", "rate": "", "industry": "", "unit": "", "amount": "", "notes2": ""}], "clean": [{"signup_date": "2023-05-16", "rate": 0.632, "industry": "Manufacturing", "unit": "lb", "amount": 5569.9}, {"signup_date": "2023-09-11", "rate": 0.713, "industry": "Real Estate", "unit": "L", "amount": 3672.63}, {"signup_date": "2023-10-09", "rate": 0.155, "industry": "Healthcare", "unit": "cm", "amount": 112.87}, {"signup_date": "2023-02-25", "rate": 0.163, "industry": "Healthcare", "unit": "lb", "amount": 2732.89}, {"signup_date": "2023-09-02", "rate": 0.496, "industry": "Real Estate", "unit": "cm", "amount": 8000.73}, {"signup_date": "2023-05-24", "rate": 0.324, "industry": "Healthcare", "unit": "L", "amount": 5250.01}, {"signup_date": "2023-03-17", "rate": 0.035, "industry": "Real Estate", "unit": "L", "amount": 8795.69}, {"signup_date": "2023-02-27", "rate": 0.408, "industry": "Hospitality", "unit": "L", "amount": 4121.32}, {"signup_date": "2023-04-20", "rate": 0.593, "industry": "Manufacturing", "unit": "lb", "amount": 5697.56}, {"signup_date": "2023-10-11", "rate": 0.09300000000000001, "industry": "Manufacturing", "unit": "L", "amount": 4354.37}, {"signup_date": "2023-12-20", "rate": 0.282, "industry": "Manufacturing", "unit": "lb", "amount": NaN}, {"signup_date": "2023-05-04", "rate": 0.948, "industry": "Real Estate", "unit": "cm", "amount": 1916.11}, {"signup_date": "2023-09-23", "rate": 0.47200000000000003, "industry": "Real Estate", "unit": "lb", "amount": 7497.83}, {"signup_date": "2023-01-09", "rate": 0.37, "industry": "Manufacturing", "unit": "cm", "amount": 1290.53}, {"signup_date": "2023-12-16", "rate": 0.9840000000000001, "industry": "Manufacturing", "unit": "lb", "amount": 2883.05}, {"signup_date": "2023-08-01", "rate": 0.43, "industry": "Healthcare", "unit": "L", "amount": NaN}, {"signup_date": "2023-04-24", "rate": 0.9740000000000001, "industry": "Hospitality", "unit": "lb", "amount": 7929.39}, {"signup_date": "2023-09-22", "rate": 0.647, "industry": "Healthcare", "unit": "cm", "amount": NaN}, {"signup_date": "2023-02-04", "rate": 0.7090000000000001, "industry": "Real Estate", "unit": "L", "amount": 1961.54}, {"signup_date": "2023-03-26", "rate": 0.996, "industry": "Manufacturing", "unit": "cm", "amount": NaN}, {"signup_date": "2023-02-13", "rate": 0.765, "industry": "Hospitality", "unit": "cm", "amount": 7806.27}, {"signup_date": "2023-11-13", "rate": 0.988, "industry": "Healthcare", "unit": "L", "amount": 5569.35}, {"signup_date": "2023-08-12", "rate": 0.10300000000000001, "industry": "Real Estate", "unit": "lb", "amount": 8777.91}, {"signup_date": "2023-09-21", "rate": 0.92, "industry": "Manufacturing", "unit": "cm", "amount": 1097.02}, {"signup_date": "2023-05-18", "rate": 0.595, "industry": "Hospitality", "unit": "L", "amount": 4274.05}, {"signup_date": "2023-09-01", "rate": 0.248, "industry": "Real Estate", "unit": "L", "amount": -100.0}, {"signup_date": "2023-04-10", "rate": 0.78, "industry": "Real Estate", "unit": "cm", "amount": NaN}, {"signup_date": "2023-06-20", "rate": 0.867, "industry": "Manufacturing", "unit": "lb", "amount": NaN}, {"signup_date": "2023-04-16", "rate": 0.257, "industry": "Healthcare", "unit": "L", "amount": NaN}], "dirty_cols": ["signup_date", "rate", "industry", "unit", "amount", "notes2"], "clean_cols": ["signup_date", "rate", "industry", "unit", "amount"], "plan": {"dataset_summary": "30 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Industrial": "Manufacturing", "real estate": "Real Estate", "Health": "Healthcare", "MEDICAL": "Healthcare", "proptech": "Real Estate", "Medical": "Healthcare", "Travel": "Hospitality", "PropTech": "Real Estate", "Mfg": "Manufacturing", "Health Care": "Healthcare", "TOURISM": "Hospitality", "tourism": "Hospitality", "medical": "Healthcare", "REAL ESTATE": "Real Estate", "Tourism": "Hospitality", "REALESTATE": "Real Estate", "realestale": "Real Estate"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LB": "lb", "l": "L", "centimeters": "cm", "pounds": "lb", "litrre": "L", "liters": "L", "litrs": "L", "pound": "lb", "Centimeter": "cm", "Cm": "cm", "Lb": "lb", "CENTIMETERS": "cm", "liter": "L", "CM": "cm", "LITR": "L"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"country": "CCK", "industry": "Telecom   ", "rate": "69.6%", "name": "Lena Cruz", "notes2": ""}, {"country": "prt", "industry": "COMMS", "rate": "91.9%", "name": "Judy Brandt  ", "notes2": ""}, {"country": "wlf", "industry": " COMMS ", "rate": "55.2%", "name": "Sara Lee", "notes2": ""}, {"country": "portuguese republic  ", "industry": "Mfg", "rate": "78.2%", "name": "Mona  Khan", "notes2": ""}, {"country": "Wallis and Futuna", "industry": "Manufacturing", "rate": "30.0%", "name": "Mona Fischer   ", "notes2": ""}, {"country": "wlf", "industry": "   Industrial", "rate": "99.0%", "name": "Heidi Fischer", "notes2": ""}, {"country": "Sint Maarten (Dutch part)", "industry": "telco", "rate": "48.7%", "name": "  Sara Lee", "notes2": ""}, {"country": "CC   ", "industry": "Comms", "rate": "45.3%", "name": "Alice Reyes", "notes2": ""}, {"country": "WF", "industry": "educatino", "rate": "9.1%", "name": "Alice Moore", "notes2": ""}, {"country": "WF", "industry": "   Comms", "rate": "64.9%", "name": "Priya  Wong", "notes2": ""}, {"country": "portugueserepublic", "industry": "mfg", "rate": "64.5%", "name": "Grace  Smith", "notes2": ""}, {"country": "Cocos (Keeling) Islands  ", "industry": "Industrial", "rate": "9.4%", "name": "Karl Lee", "notes2": ""}, {"country": "PRT", "industry": "EdTech", "rate": "91.4%", "name": "Omar Johnson", "notes2": ""}, {"country": "wf", "industry": "Education", "rate": "84.2%", "name": "David Novak", "notes2": ""}, {"country": "PORTUGUESE REPUBLIC", "industry": "manufacturing", "rate": "64.8%", "name": "Bob  Adams", "notes2": ""}, {"country": "  sint maarten (dutch part)", "industry": "MPNUFACTURING", "rate": "87.5%", "name": "David Ali", "notes2": ""}, {"country": "CCK", "industry": "EDUCATION", "rate": "65.4%", "name": "Ivan Park", "notes2": ""}, {"country": "SXM", "industry": " TELCO ", "rate": "86.6%", "name": "Priya Diaz  ", "notes2": ""}, {"country": "PRT", "industry": "Telecom", "rate": "95.8%", "name": "Sara  Khan", "notes2": ""}, {"country": "cocos keeling isalnds", "industry": "Coms", "rate": "76.7%", "name": "  Eve Cruz", "notes2": ""}, {"country": "wlf", "industry": "Mfg", "rate": "35.1%", "name": "Ivan Smith", "notes2": ""}, {"country": "Portuguese Republic", "industry": "education", "rate": "25.1%", "name": "Judy  Ali", "notes2": ""}, {"country": "portugal", "industry": "mfg", "rate": "10.0%", "name": "Grace Fischer ", "notes2": ""}, {"country": "CCK  ", "industry": "Mfg   ", "rate": "17.5%", "name": "  Frank Petrov", "notes2": ""}, {"country": "CC", "industry": "Manufacturing", "rate": "23.9%", "name": "Grace Johnson", "notes2": ""}, {"country": "portugal", "industry": "mfg", "rate": "10.0%", "name": "Grace Fischer ", "notes2": ""}, {"country": "WF", "industry": "educatino", "rate": "9.1%", "name": "Alice Moore", "notes2": ""}, {"country": "", "industry": "", "rate": "", "name": "", "notes2": ""}], "clean": [{"country": "Cocos (Keeling) Islands", "industry": "Telecommunications", "rate": 0.696, "name": "Lena Cruz"}, {"country": "Portugal", "industry": "Telecommunications", "rate": 0.919, "name": "Judy Brandt"}, {"country": "Wallis and Futuna", "industry": "Telecommunications", "rate": 0.552, "name": "Sara Lee"}, {"country": "Portugal", "industry": "Manufacturing", "rate": 0.782, "name": "Mona Khan"}, {"country": "Wallis and Futuna", "industry": "Manufacturing", "rate": 0.3, "name": "Mona Fischer"}, {"country": "Wallis and Futuna", "industry": "Manufacturing", "rate": 0.99, "name": "Heidi Fischer"}, {"country": "Sint Maarten (Dutch part)", "industry": "Telecommunications", "rate": 0.48700000000000004, "name": "Sara Lee"}, {"country": "Cocos (Keeling) Islands", "industry": "Telecommunications", "rate": 0.45299999999999996, "name": "Alice Reyes"}, {"country": "Wallis and Futuna", "industry": "Education", "rate": 0.091, "name": "Alice Moore"}, {"country": "Wallis and Futuna", "industry": "Telecommunications", "rate": 0.649, "name": "Priya Wong"}, {"country": "Portugal", "industry": "Manufacturing", "rate": 0.645, "name": "Grace Smith"}, {"country": "Cocos (Keeling) Islands", "industry": "Manufacturing", "rate": 0.094, "name": "Karl Lee"}, {"country": "Portugal", "industry": "Education", "rate": 0.914, "name": "Omar Johnson"}, {"country": "Wallis and Futuna", "industry": "Education", "rate": 0.8420000000000001, "name": "David Novak"}, {"country": "Portugal", "industry": "Manufacturing", "rate": 0.648, "name": "Bob Adams"}, {"country": "Sint Maarten (Dutch part)", "industry": "Manufacturing", "rate": 0.875, "name": "David Ali"}, {"country": "Cocos (Keeling) Islands", "industry": "Education", "rate": 0.654, "name": "Ivan Park"}, {"country": "Sint Maarten (Dutch part)", "industry": "Telecommunications", "rate": 0.866, "name": "Priya Diaz"}, {"country": "Portugal", "industry": "Telecommunications", "rate": 0.958, "name": "Sara Khan"}, {"country": "Cocos (Keeling) Islands", "industry": "Telecommunications", "rate": 0.767, "name": "Eve Cruz"}, {"country": "Wallis and Futuna", "industry": "Manufacturing", "rate": 0.35100000000000003, "name": "Ivan Smith"}, {"country": "Portugal", "industry": "Education", "rate": 0.251, "name": "Judy Ali"}, {"country": "Portugal", "industry": "Manufacturing", "rate": 0.1, "name": "Grace Fischer"}, {"country": "Cocos (Keeling) Islands", "industry": "Manufacturing", "rate": 0.175, "name": "Frank Petrov"}, {"country": "Cocos (Keeling) Islands", "industry": "Manufacturing", "rate": 0.239, "name": "Grace Johnson"}], "dirty_cols": ["country", "industry", "rate", "name", "notes2"], "clean_cols": ["country", "industry", "rate", "name"], "plan": {"dataset_summary": "28 rows × 5 columns. 4 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CCK": "Cocos (Keeling) Islands", "prt": "Portugal", "wlf": "Wallis and Futuna", "portuguese republic": "Portugal", "CC": "Cocos (Keeling) Islands", "WF": "Wallis and Futuna", "portugueserepublic": "Portugal", "PRT": "Portugal", "wf": "Wallis and Futuna", "PORTUGUESE REPUBLIC": "Portugal", "sint maarten (dutch part)": "Sint Maarten (Dutch part)", "SXM": "Sint Maarten (Dutch part)", "cocos keeling isalnds": "Cocos (Keeling) Islands", "Portuguese Republic": "Portugal", "portugal": "Portugal"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Telecom": "Telecommunications", "COMMS": "Telecommunications", "Mfg": "Manufacturing", "Industrial": "Manufacturing", "telco": "Telecommunications", "Comms": "Telecommunications", "educatino": "Education", "mfg": "Manufacturing", "EdTech": "Education", "manufacturing": "Manufacturing", "MPNUFACTURING": "Manufacturing", "EDUCATION": "Education", "TELCO": "Telecommunications", "Coms": "Telecommunications", "education": "Education"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"job_title": "Chief Technology Officer", "status": "OVEERDUE", "currency": "NAKFA", "company": " Globex ", "signup_date": "45030", "state": "KY", "col_x": ""}, {"job_title": "Sales Rep", "status": " paid", "currency": "ERN", "company": " Initech", "signup_date": "44942", "state": "Idaho", "col_x": ""}, {"job_title": "Chief Technology Officer   ", "status": "unpaid", "currency": "  DJF", "company": "Wonka Co", "signup_date": "13 Mar 2023", "state": "kentucky", "col_x": ""}, {"job_title": "engineering manager", "status": "Paid", "currency": "Djibouti Franc", "company": " Hooli", "signup_date": "23 Aug 2023", "state": "kentucky", "col_x": ""}, {"job_title": " cto ", "status": "due", "currency": "djf", "company": "Hooli   ", "signup_date": "12 Sep 2023", "state": "KY", "col_x": ""}, {"job_title": "Director", "status": "   overdue", "currency": "djf", "company": "Hooli", "signup_date": "7 Feb 2023", "state": "ID  ", "col_x": ""}, {"job_title": "Sales Rep", "status": "Pending", "currency": "DJF", "company": "Stark Industries", "signup_date": "9 Feb 2023", "state": "ky", "col_x": ""}, {"job_title": "Rep", "status": "late", "currency": "ern ", "company": "   Wonka Co", "signup_date": "2023-08-20", "state": "VT", "col_x": ""}, {"job_title": "eng manager", "status": "pending", "currency": "Nakfa", "company": " Soylent Corp", "signup_date": "2023-03-25", "state": "KENTUCKY", "col_x": ""}, {"job_title": "Engineering Mgr", "status": "late", "currency": "NAKFA", "company": "   Stark Industries", "signup_date": "2023-10-08", "state": "Kentucky", "col_x": ""}, {"job_title": "Eng Mgr", "status": " due", "currency": "djf", "company": " Umbrella ", "signup_date": "25 Feb 2023", "state": "VT", "col_x": ""}, {"job_title": " Dir ", "status": "due", "currency": "DJF", "company": "Initech", "signup_date": "10/27/2023", "state": "Kentucky", "col_x": ""}, {"job_title": "sales  representative", "status": " pending ", "currency": "NAKFA", "company": "Acme Inc   ", "signup_date": "2023-12-18", "state": " VT ", "col_x": ""}, {"job_title": "dir", "status": "Late", "currency": "ERN", "company": "Wonka  Co", "signup_date": "45191", "state": "VT", "col_x": ""}, {"job_title": "Dr.", "status": "pending", "currency": "djibouti  franc", "company": "  Initech", "signup_date": "45085", "state": "ky", "col_x": ""}, {"job_title": "Rep", "status": "   PAID", "currency": "Ern", "company": "Cyberdyne", "signup_date": "11/27/2023", "state": "Idaho", "col_x": ""}, {"job_title": "  Engineering Mgr", "status": "pening", "currency": "ERN", "company": "Vehement  ", "signup_date": "1/17/2023", "state": "Idaho", "col_x": ""}, {"job_title": "engineering  mgr", "status": "Pending", "currency": "DJF  ", "company": "   Vehement", "signup_date": "5/7/2023", "state": "ID", "col_x": ""}, {"job_title": "C.T.O.  ", "status": "PAID", "currency": "djibouti franc", "company": "Globex", "signup_date": "9 Feb 2023", "state": "Vfrmont", "col_x": ""}, {"job_title": "chief technology officer", "status": "lte", "currency": "ERN", "company": "Globex", "signup_date": "2023-03-05", "state": "VT", "col_x": ""}, {"job_title": "  Director", "status": "Pending", "currency": "ERN", "company": "Globex", "signup_date": "2023-10-28", "state": "Idaho", "col_x": ""}, {"job_title": "Engineering Manager", "status": "overdue", "currency": "NAKFA", "company": "Acme Inc", "signup_date": "7/11/2023", "state": " Idaho ", "col_x": ""}, {"job_title": "sales representative", "status": "late", "currency": "Djibouti Franc", "company": " Vehement ", "signup_date": "45013", "state": " id ", "col_x": ""}, {"job_title": "Engineering Manager", "status": "overdue", "currency": "NAKFA", "company": "Acme Inc", "signup_date": "7/11/2023", "state": " Idaho ", "col_x": ""}], "clean": [{"job_title": "Chief Technology Officer", "status": "Overdue", "currency": "ERN", "company": "Globex", "signup_date": "2023-04-14", "state": "Kentucky"}, {"job_title": "Sales Representative", "status": "Paid", "currency": "ERN", "company": "Initech", "signup_date": "2023-01-16", "state": "Idaho"}, {"job_title": "Chief Technology Officer", "status": "Pending", "currency": "DJF", "company": "Wonka Co", "signup_date": "2023-03-13", "state": "Kentucky"}, {"job_title": "Engineering Manager", "status": "Paid", "currency": "DJF", "company": "Hooli", "signup_date": "2023-08-23", "state": "Kentucky"}, {"job_title": "Chief Technology Officer", "status": "Pending", "currency": "DJF", "company": "Hooli", "signup_date": "2023-09-12", "state": "Kentucky"}, {"job_title": "Director", "status": "Overdue", "currency": "DJF", "company": "Hooli", "signup_date": "2023-02-07", "state": "Idaho"}, {"job_title": "Sales Representative", "status": "Pending", "currency": "DJF", "company": "Stark Industries", "signup_date": "2023-02-09", "state": "Kentucky"}, {"job_title": "Sales Representative", "status": "Overdue", "currency": "ERN", "company": "Wonka Co", "signup_date": "2023-08-20", "state": "Vermont"}, {"job_title": "Engineering Manager", "status": "Pending", "currency": "ERN", "company": "Soylent Corp", "signup_date": "2023-03-25", "state": "Kentucky"}, {"job_title": "Engineering Manager", "status": "Overdue", "currency": "ERN", "company": "Stark Industries", "signup_date": "2023-10-08", "state": "Kentucky"}, {"job_title": "Engineering Manager", "status": "Pending", "currency": "DJF", "company": "Umbrella", "signup_date": "2023-02-25", "state": "Vermont"}, {"job_title": "Director", "status": "Pending", "currency": "DJF", "company": "Initech", "signup_date": "2023-10-27", "state": "Kentucky"}, {"job_title": "Sales Representative", "status": "Pending", "currency": "ERN", "company": "Acme Inc", "signup_date": "2023-12-18", "state": "Vermont"}, {"job_title": "Director", "status": "Overdue", "currency": "ERN", "company": "Wonka Co", "signup_date": "2023-09-22", "state": "Vermont"}, {"job_title": "Director", "status": "Pending", "currency": "DJF", "company": "Initech", "signup_date": "2023-06-08", "state": "Kentucky"}, {"job_title": "Sales Representative", "status": "Paid", "currency": "ERN", "company": "Cyberdyne", "signup_date": "2023-11-27", "state": "Idaho"}, {"job_title": "Engineering Manager", "status": "Pending", "currency": "ERN", "company": "Vehement", "signup_date": "2023-01-17", "state": "Idaho"}, {"job_title": "Engineering Manager", "status": "Pending", "currency": "DJF", "company": "Vehement", "signup_date": "2023-05-07", "state": "Idaho"}, {"job_title": "Chief Technology Officer", "status": "Paid", "currency": "DJF", "company": "Globex", "signup_date": "2023-02-09", "state": "Vermont"}, {"job_title": "Chief Technology Officer", "status": "Overdue", "currency": "ERN", "company": "Globex", "signup_date": "2023-03-05", "state": "Vermont"}, {"job_title": "Director", "status": "Pending", "currency": "ERN", "company": "Globex", "signup_date": "2023-10-28", "state": "Idaho"}, {"job_title": "Engineering Manager", "status": "Overdue", "currency": "ERN", "company": "Acme Inc", "signup_date": "2023-07-11", "state": "Idaho"}, {"job_title": "Sales Representative", "status": "Overdue", "currency": "DJF", "company": "Vehement", "signup_date": "2023-03-28", "state": "Idaho"}], "dirty_cols": ["job_title", "status", "currency", "company", "signup_date", "state", "col_x"], "clean_cols": ["job_title", "status", "currency", "company", "signup_date", "state"], "plan": {"dataset_summary": "24 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Sales Rep": "Sales Representative", "engineering manager": "Engineering Manager", "cto": "Chief Technology Officer", "Rep": "Sales Representative", "eng manager": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "Eng Mgr": "Engineering Manager", "Dir": "Director", "sales representative": "Sales Representative", "dir": "Director", "Dr.": "Director", "engineering mgr": "Engineering Manager", "C.T.O.": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OVEERDUE": "Overdue", "paid": "Paid", "unpaid": "Pending", "due": "Pending", "overdue": "Overdue", "late": "Overdue", "pending": "Pending", "Late": "Overdue", "PAID": "Paid", "pening": "Pending", "lte": "Overdue"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NAKFA": "ERN", "Djibouti Franc": "DJF", "djf": "DJF", "ern": "ERN", "Nakfa": "ERN", "djibouti franc": "DJF", "Ern": "ERN"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KY": "Kentucky", "kentucky": "Kentucky", "ID": "Idaho", "ky": "Kentucky", "VT": "Vermont", "KENTUCKY": "Kentucky", "Vfrmont": "Vermont", "id": "Idaho"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "ADMIN ASSISTANT  ", "unit": "centimeter", "status": "Overdue", "name": "Ivan Wong "}, {"job_title": "sr engineer", "unit": "   Ml", "status": "pendisg", "name": "Mona Moore"}, {"job_title": "senior eng", "unit": "centimeters", "status": "   overdue", "name": "Omar  Cruz"}, {"job_title": "admin assistant   ", "unit": "centimeter", "status": "   PIAD", "name": "Eve Diaz"}, {"job_title": "   Administrative Assistant", "unit": "mls", "status": "unpaid", "name": "David  Novak"}, {"job_title": "Admin Asst", "unit": "milliliter", "status": "Overdue", "name": "   Lena Novak"}, {"job_title": "Sr. Engineer", "unit": "Centimeters", "status": "unpaid", "name": "Karl Lee"}, {"job_title": "sr engineer", "unit": "km", "status": "Due", "name": "Heidi Ali"}, {"job_title": "SR ENGINNEER", "unit": "centimeters", "status": " PENDING ", "name": "Omar Park   "}, {"job_title": "SENIOR ENGINEER", "unit": "Cm", "status": "overdue   ", "name": "David  Wong"}, {"job_title": "Snr Engineer", "unit": "mls", "status": "Overdue", "name": "Grace Park   "}, {"job_title": "Administrative  Assistant", "unit": "kg", "status": "Pending", "name": "  Heidi Johnson"}, {"job_title": "Senior Engineer", "unit": "  ml", "status": "pending ", "name": "Judy Fischer"}, {"job_title": " admin", "unit": "KG", "status": "PENDING ", "name": "Sara Cruz  "}, {"job_title": "senior eng", "unit": "km", "status": "late", "name": "Karl Johnson"}, {"job_title": "SR ENGINNEER", "unit": "centimeters", "status": " PENDING ", "name": "Omar Park   "}], "clean": [{"job_title": "Administrative Assistant", "unit": "cm", "status": "Overdue", "name": "Ivan Wong"}, {"job_title": "Senior Engineer", "unit": "mL", "status": "Pending", "name": "Mona Moore"}, {"job_title": "Senior Engineer", "unit": "cm", "status": "Overdue", "name": "Omar Cruz"}, {"job_title": "Administrative Assistant", "unit": "cm", "status": "Paid", "name": "Eve Diaz"}, {"job_title": "Administrative Assistant", "unit": "mL", "status": "Pending", "name": "David Novak"}, {"job_title": "Administrative Assistant", "unit": "mL", "status": "Overdue", "name": "Lena Novak"}, {"job_title": "Senior Engineer", "unit": "cm", "status": "Pending", "name": "Karl Lee"}, {"job_title": "Senior Engineer", "unit": "km", "status": "Pending", "name": "Heidi Ali"}, {"job_title": "Senior Engineer", "unit": "cm", "status": "Pending", "name": "Omar Park"}, {"job_title": "Senior Engineer", "unit": "cm", "status": "Overdue", "name": "David Wong"}, {"job_title": "Senior Engineer", "unit": "mL", "status": "Overdue", "name": "Grace Park"}, {"job_title": "Administrative Assistant", "unit": "kg", "status": "Pending", "name": "Heidi Johnson"}, {"job_title": "Senior Engineer", "unit": "mL", "status": "Pending", "name": "Judy Fischer"}, {"job_title": "Administrative Assistant", "unit": "kg", "status": "Pending", "name": "Sara Cruz"}, {"job_title": "Senior Engineer", "unit": "km", "status": "Overdue", "name": "Karl Johnson"}], "dirty_cols": ["job_title", "unit", "status", "name"], "clean_cols": ["job_title", "unit", "status", "name"], "plan": {"dataset_summary": "16 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ADMIN ASSISTANT": "Administrative Assistant", "sr engineer": "Senior Engineer", "senior eng": "Senior Engineer", "admin assistant": "Administrative Assistant", "Admin Asst": "Administrative Assistant", "Sr. Engineer": "Senior Engineer", "SR ENGINNEER": "Senior Engineer", "SENIOR ENGINEER": "Senior Engineer", "Snr Engineer": "Senior Engineer", "admin": "Administrative Assistant"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"centimeter": "cm", "Ml": "mL", "centimeters": "cm", "mls": "mL", "milliliter": "mL", "Centimeters": "cm", "Cm": "cm", "ml": "mL", "KG": "kg"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pendisg": "Pending", "overdue": "Overdue", "PIAD": "Paid", "unpaid": "Pending", "Due": "Pending", "PENDING": "Pending", "pending": "Pending", "late": "Overdue"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"amount": "$3,655.22", "department": "accounting", "status": "Active", "currency": "HUF", "job_title": "rep", "signup_date": "2023-06-27", "extra": ""}, {"amount": "N/A", "department": "Fin", "status": "CANCELLED", "currency": "Forint", "job_title": "Sales Representatvie", "signup_date": "8/22/2023", "extra": ""}, {"amount": "$8,368.39", "department": "Marketing", "status": "   ACTIVE", "currency": "Dominican Peso", "job_title": "ENG MGR", "signup_date": "44965", "extra": ""}, {"amount": "8.726,75", "department": "accounting", "status": "   canceled", "currency": "  forint", "job_title": "Dir", "signup_date": "45116", "extra": ""}, {"amount": "$2,836.96", "department": "ENG.", "status": "Active", "currency": "   Huf", "job_title": "  SalesRep", "signup_date": "2023-11-08", "extra": ""}, {"amount": "$3,610.55", "department": "  Fin", "status": "   active", "currency": "Huf", "job_title": "dir. ", "signup_date": "2023-03-11", "extra": ""}, {"amount": "7.133,66", "department": "   Accounting", "status": "on hold", "currency": "HUF", "job_title": "senior engineer", "signup_date": "1 Dec 2023", "extra": ""}, {"amount": "4.082,90", "department": "Eng.", "status": "active", "currency": "dop", "job_title": "engineering  mgr", "signup_date": "2023-07-28", "extra": ""}, {"amount": "7.356,37", "department": " R&D ", "status": "paused", "currency": "dop", "job_title": "dir", "signup_date": "44997", "extra": ""}, {"amount": "8.769,62", "department": "   Accounring", "status": "churned", "currency": "dop", "job_title": "direcor", "signup_date": "45156", "extra": ""}, {"amount": "5.013,72", "department": "MKTG", "status": "churned", "currency": "Gourde", "job_title": "Sr Engineer", "signup_date": "2023-01-20", "extra": ""}, {"amount": "null", "department": "Mkt", "status": "   ative", "currency": "huf", "job_title": "Rep", "signup_date": "2023-01-18", "extra": ""}, {"amount": "7.806,03", "department": " Dev ", "status": "Churned", "currency": "bdt", "job_title": "dir.", "signup_date": "26 Sep 2023", "extra": ""}, {"amount": "4.818,29", "department": "R&D", "status": "PAUSED", "currency": "huf", "job_title": "eng mgr", "signup_date": "2023-12-24", "extra": ""}, {"amount": "TBD", "department": "   R&D", "status": "active", "currency": "Goudre", "job_title": "DIR.", "signup_date": "2023-11-12", "extra": ""}, {"amount": "1.166,71", "department": "FIN", "status": "on hold", "currency": "GOURDE", "job_title": "Director", "signup_date": "4 Jan 2023", "extra": ""}, {"amount": "3.763,19", "department": "Fin", "status": " paused ", "currency": "HTG", "job_title": "director", "signup_date": "45159", "extra": ""}, {"amount": "--", "department": "r&d", "status": "Active", "currency": "   GOURDE", "job_title": "Sales Rep", "signup_date": "5/6/2023", "extra": ""}, {"amount": "2.992,48", "department": "dev", "status": "canceled", "currency": "DOP", "job_title": "  SALES REP", "signup_date": "2023-02-20", "extra": ""}, {"amount": "$8,955.42", "department": "Eng", "status": "churned", "currency": "HUF", "job_title": "Rep", "signup_date": "10/22/2023", "extra": ""}, {"amount": "1.263,74", "department": "Finance", "status": "churned", "currency": " DOP", "job_title": "Snr Engineer", "signup_date": "9/3/2023", "extra": ""}, {"amount": "$1,673.98", "department": "marketing  ", "status": "churned", "currency": "Taka", "job_title": "   senior engineer", "signup_date": "45273", "extra": ""}, {"amount": "na", "department": "Finance", "status": "CANCELLED", "currency": "Taka", "job_title": "Sales  Rep", "signup_date": "2023-08-27", "extra": ""}, {"amount": "$2,283.04", "department": "Accounting", "status": "Actvie", "currency": "Forint", "job_title": "engineering manager", "signup_date": "2023-05-03", "extra": ""}, {"amount": "$8,562.74", "department": "Engineering", "status": "churned", "currency": "tnka", "job_title": "SALES REP   ", "signup_date": "45171", "extra": ""}, {"amount": "$1,388.66", "department": "En.", "status": " cancelled ", "currency": "dop", "job_title": "sales representative", "signup_date": "2023-04-16", "extra": ""}, {"amount": "2.023,51", "department": "R&D", "status": "cancelled", "currency": "Taka", "job_title": "Sales Representative", "signup_date": "8/27/2023", "extra": ""}, {"amount": "$2,772.33", "department": "Fin", "status": "   PASED", "currency": "taka", "job_title": "Senior Engineer", "signup_date": "11 May 2023", "extra": ""}, {"amount": "$5,168.54", "department": "growth", "status": "   CHURNED", "currency": "Dop ", "job_title": "  DIR", "signup_date": "2023-06-19", "extra": ""}, {"amount": "5.872,06", "department": " Finance", "status": "active", "currency": "HUF", "job_title": "dir", "signup_date": "45142", "extra": ""}, {"amount": "$5,535.21", "department": "Finance", "status": "paused", "currency": "Dominican Peso  ", "job_title": "eng  managr", "signup_date": "2023-06-19", "extra": ""}, {"amount": "4.577,19", "department": "Fin", "status": "CHURNED", "currency": "Taka", "job_title": "senior engineer   ", "signup_date": "45215", "extra": ""}, {"amount": "3.270,15", "department": "R&D", "status": "paused   ", "currency": " Gourde ", "job_title": "sales  rep", "signup_date": "9/24/2023", "extra": ""}, {"amount": "2.316,05", "department": "  Mareting", "status": "churned", "currency": "Dominicn Peso", "job_title": "Eng Mr", "signup_date": "2023-04-13", "extra": ""}, {"amount": "$7,269.04", "department": "mkt", "status": "ACTIVE", "currency": "BDT", "job_title": "Snr Engineer", "signup_date": "19 May 2023", "extra": ""}, {"amount": "6.096,23", "department": " ACCOUNTING", "status": "active", "currency": "Taka", "job_title": "Engineering Mgr", "signup_date": "11 Dec 2023", "extra": ""}, {"amount": "2.408,31", "department": "Mkt", "status": " Churned ", "currency": "gorude   ", "job_title": "Eng Mgr", "signup_date": "8/13/2023", "extra": ""}, {"amount": "?", "department": "Growth", "status": "Chutned", "currency": "DOMINICAN PESO", "job_title": "  senior eng", "signup_date": "45145", "extra": ""}, {"amount": "$6,416.54", "department": "GROWTH", "status": " PAUSED ", "currency": "Taka", "job_title": "ENG MGR", "signup_date": "2023-10-17", "extra": ""}, {"amount": "1.588,27", "department": "Eng", "status": "cancleled", "currency": "Fxrint", "job_title": "Sales Representative", "signup_date": "2023-01-21", "extra": ""}, {"amount": "$1,671.69", "department": "   Mkt", "status": "ACTIVE", "currency": "BDT", "job_title": "snr engineer", "signup_date": "2023-12-11", "extra": ""}, {"amount": "$5,865.24", "department": "  Finance", "status": "  Churned", "currency": "taka", "job_title": "Engbneering Manager", "signup_date": "45246", "extra": ""}, {"amount": "$8,311.12", "department": "Finane", "status": "active", "currency": "taka", "job_title": "Snr Engineer  ", "signup_date": "7/12/2023", "extra": ""}, {"amount": "na", "department": "dev", "status": "on hold", "currency": "taka", "job_title": "ENG MGR", "signup_date": "5/20/2023", "extra": ""}, {"amount": "$8,669.40", "department": "r&d", "status": " ACTIVE ", "currency": "DOMINICAN PESO", "job_title": "engineering mgr", "signup_date": "11 Jun 2023", "extra": ""}, {"amount": "$4,759.00", "department": "  growth", "status": "Active", "currency": "HTG", "job_title": "Salees Representative", "signup_date": "45246", "extra": ""}, {"amount": "6.633,32", "department": "Eng.", "status": "churned", "currency": "dominican peso", "job_title": "sr. enineer", "signup_date": "2023-09-05", "extra": ""}, {"amount": "7.361,39", "department": "DEV", "status": "canceled", "currency": " forint", "job_title": "snr engineer", "signup_date": "2/6/2023", "extra": ""}, {"amount": "N/A", "department": "marketign  ", "status": "Active", "currency": " dominican peso", "job_title": "   Senior Engineer", "signup_date": "44967", "extra": ""}, {"amount": "$2,181.76", "department": "dev", "status": "cancelled", "currency": "gourde", "job_title": "Snz Engineer", "signup_date": "6/1/2023", "extra": ""}, {"amount": "8.095,93", "department": " ENG", "status": "Canclled", "currency": "HTG", "job_title": "Eng Manager", "signup_date": "10/10/2023", "extra": ""}, {"amount": "8.140,02", "department": "Fin", "status": "Cancelled", "currency": "BDT", "job_title": "dir.   ", "signup_date": "2023-09-03", "extra": ""}, {"amount": "$8,647.19", "department": "Finance", "status": "CHUREND", "currency": "Takka", "job_title": "dir", "signup_date": "44988", "extra": ""}, {"amount": "3.217,46", "department": " Eng ", "status": "CHUDNED", "currency": "HTG", "job_title": "Engineering Manager  ", "signup_date": "44975", "extra": ""}, {"amount": "6.233,02", "department": "Marketing", "status": "Paused", "currency": "HTG", "job_title": "sr engineer", "signup_date": "45022", "extra": ""}, {"amount": "6.021,24", "department": "Engineering", "status": "cancelled", "currency": "DOP", "job_title": "Senior Engineer ", "signup_date": "45245", "extra": ""}, {"amount": "$5,815.76", "department": " MKT ", "status": "cancelled", "currency": "DOP", "job_title": "ENG MGR", "signup_date": "26 Jul 2023", "extra": ""}, {"amount": "7.159,63", "department": "R&D", "status": "churned", "currency": "Dominican Peso", "job_title": " Director ", "signup_date": "45203", "extra": ""}, {"amount": "?", "department": "Growth", "status": "Chutned", "currency": "DOMINICAN PESO", "job_title": "  senior eng", "signup_date": "45145", "extra": ""}, {"amount": "TBD", "department": "   R&D", "status": "active", "currency": "Goudre", "job_title": "DIR.", "signup_date": "2023-11-12", "extra": ""}], "clean": [{"amount": 3655.22, "department": "Finance", "status": "Active", "currency": "HUF", "job_title": "Sales Representative", "signup_date": "2023-06-27"}, {"amount": NaN, "department": "Finance", "status": "Churned", "currency": "HUF", "job_title": "Sales Representative", "signup_date": "2023-08-22"}, {"amount": 8368.39, "department": "Marketing", "status": "Active", "currency": "DOP", "job_title": "Engineering Manager", "signup_date": "2023-02-08"}, {"amount": 8726.75, "department": "Finance", "status": "Churned", "currency": "HUF", "job_title": "Director", "signup_date": "2023-07-09"}, {"amount": 2836.96, "department": "Engineering", "status": "Active", "currency": "HUF", "job_title": "Sales Representative", "signup_date": "2023-11-08"}, {"amount": 3610.55, "department": "Finance", "status": "Active", "currency": "HUF", "job_title": "Director", "signup_date": "2023-03-11"}, {"amount": 7133.66, "department": "Finance", "status": "Paused", "currency": "HUF", "job_title": "Senior Engineer", "signup_date": "2023-12-01"}, {"amount": 4082.9, "department": "Engineering", "status": "Active", "currency": "DOP", "job_title": "Engineering Manager", "signup_date": "2023-07-28"}, {"amount": 7356.37, "department": "Engineering", "status": "Paused", "currency": "DOP", "job_title": "Director", "signup_date": "2023-03-12"}, {"amount": 8769.62, "department": "Finance", "status": "Churned", "currency": "DOP", "job_title": "Director", "signup_date": "2023-08-18"}, {"amount": 5013.72, "department": "Marketing", "status": "Churned", "currency": "HTG", "job_title": "Senior Engineer", "signup_date": "2023-01-20"}, {"amount": NaN, "department": "Marketing", "status": "Active", "currency": "HUF", "job_title": "Sales Representative", "signup_date": "2023-01-18"}, {"amount": 7806.03, "department": "Engineering", "status": "Churned", "currency": "BDT", "job_title": "Director", "signup_date": "2023-09-26"}, {"amount": 4818.29, "department": "Engineering", "status": "Paused", "currency": "HUF", "job_title": "Engineering Manager", "signup_date": "2023-12-24"}, {"amount": NaN, "department": "Engineering", "status": "Active", "currency": "HTG", "job_title": "Director", "signup_date": "2023-11-12"}, {"amount": 1166.71, "department": "Finance", "status": "Paused", "currency": "HTG", "job_title": "Director", "signup_date": "2023-01-04"}, {"amount": 3763.19, "department": "Finance", "status": "Paused", "currency": "HTG", "job_title": "Director", "signup_date": "2023-08-21"}, {"amount": NaN, "department": "Engineering", "status": "Active", "currency": "HTG", "job_title": "Sales Representative", "signup_date": "2023-05-06"}, {"amount": 2992.48, "department": "Engineering", "status": "Churned", "currency": "DOP", "job_title": "Sales Representative", "signup_date": "2023-02-20"}, {"amount": 8955.42, "department": "Engineering", "status": "Churned", "currency": "HUF", "job_title": "Sales Representative", "signup_date": "2023-10-22"}, {"amount": 1263.74, "department": "Finance", "status": "Churned", "currency": "DOP", "job_title": "Senior Engineer", "signup_date": "2023-09-03"}, {"amount": 1673.98, "department": "Marketing", "status": "Churned", "currency": "BDT", "job_title": "Senior Engineer", "signup_date": "2023-12-13"}, {"amount": NaN, "department": "Finance", "status": "Churned", "currency": "BDT", "job_title": "Sales Representative", "signup_date": "2023-08-27"}, {"amount": 2283.04, "department": "Finance", "status": "Active", "currency": "HUF", "job_title": "Engineering Manager", "signup_date": "2023-05-03"}, {"amount": 8562.74, "department": "Engineering", "status": "Churned", "currency": "BDT", "job_title": "Sales Representative", "signup_date": "2023-09-02"}, {"amount": 1388.66, "department": "Engineering", "status": "Churned", "currency": "DOP", "job_title": "Sales Representative", "signup_date": "2023-04-16"}, {"amount": 2023.51, "department": "Engineering", "status": "Churned", "currency": "BDT", "job_title": "Sales Representative", "signup_date": "2023-08-27"}, {"amount": 2772.33, "department": "Finance", "status": "Paused", "currency": "BDT", "job_title": "Senior Engineer", "signup_date": "2023-05-11"}, {"amount": 5168.54, "department": "Marketing", "status": "Churned", "currency": "DOP", "job_title": "Director", "signup_date": "2023-06-19"}, {"amount": 5872.06, "department": "Finance", "status": "Active", "currency": "HUF", "job_title": "Director", "signup_date": "2023-08-04"}, {"amount": 5535.21, "department": "Finance", "status": "Paused", "currency": "DOP", "job_title": "Engineering Manager", "signup_date": "2023-06-19"}, {"amount": 4577.19, "department": "Finance", "status": "Churned", "currency": "BDT", "job_title": "Senior Engineer", "signup_date": "2023-10-16"}, {"amount": 3270.15, "department": "Engineering", "status": "Paused", "currency": "HTG", "job_title": "Sales Representative", "signup_date": "2023-09-24"}, {"amount": 2316.05, "department": "Marketing", "status": "Churned", "currency": "DOP", "job_title": "Engineering Manager", "signup_date": "2023-04-13"}, {"amount": 7269.04, "department": "Marketing", "status": "Active", "currency": "BDT", "job_title": "Senior Engineer", "signup_date": "2023-05-19"}, {"amount": 6096.23, "department": "Finance", "status": "Active", "currency": "BDT", "job_title": "Engineering Manager", "signup_date": "2023-12-11"}, {"amount": 2408.31, "department": "Marketing", "status": "Churned", "currency": "HTG", "job_title": "Engineering Manager", "signup_date": "2023-08-13"}, {"amount": NaN, "department": "Marketing", "status": "Churned", "currency": "DOP", "job_title": "Senior Engineer", "signup_date": "2023-08-07"}, {"amount": 6416.54, "department": "Marketing", "status": "Paused", "currency": "BDT", "job_title": "Engineering Manager", "signup_date": "2023-10-17"}, {"amount": 1588.27, "department": "Engineering", "status": "Churned", "currency": "HUF", "job_title": "Sales Representative", "signup_date": "2023-01-21"}, {"amount": 1671.69, "department": "Marketing", "status": "Active", "currency": "BDT", "job_title": "Senior Engineer", "signup_date": "2023-12-11"}, {"amount": 5865.24, "department": "Finance", "status": "Churned", "currency": "BDT", "job_title": "Engineering Manager", "signup_date": "2023-11-16"}, {"amount": 8311.12, "department": "Finance", "status": "Active", "currency": "BDT", "job_title": "Senior Engineer", "signup_date": "2023-07-12"}, {"amount": NaN, "department": "Engineering", "status": "Paused", "currency": "BDT", "job_title": "Engineering Manager", "signup_date": "2023-05-20"}, {"amount": 8669.4, "department": "Engineering", "status": "Active", "currency": "DOP", "job_title": "Engineering Manager", "signup_date": "2023-06-11"}, {"amount": 4759.0, "department": "Marketing", "status": "Active", "currency": "HTG", "job_title": "Sales Representative", "signup_date": "2023-11-16"}, {"amount": 6633.32, "department": "Engineering", "status": "Churned", "currency": "DOP", "job_title": "Senior Engineer", "signup_date": "2023-09-05"}, {"amount": 7361.39, "department": "Engineering", "status": "Churned", "currency": "HUF", "job_title": "Senior Engineer", "signup_date": "2023-02-06"}, {"amount": NaN, "department": "Marketing", "status": "Active", "currency": "DOP", "job_title": "Senior Engineer", "signup_date": "2023-02-10"}, {"amount": 2181.76, "department": "Engineering", "status": "Churned", "currency": "HTG", "job_title": "Senior Engineer", "signup_date": "2023-06-01"}, {"amount": 8095.93, "department": "Engineering", "status": "Churned", "currency": "HTG", "job_title": "Engineering Manager", "signup_date": "2023-10-10"}, {"amount": 8140.02, "department": "Finance", "status": "Churned", "currency": "BDT", "job_title": "Director", "signup_date": "2023-09-03"}, {"amount": 8647.19, "department": "Finance", "status": "Churned", "currency": "BDT", "job_title": "Director", "signup_date": "2023-03-03"}, {"amount": 3217.46, "department": "Engineering", "status": "Churned", "currency": "HTG", "job_title": "Engineering Manager", "signup_date": "2023-02-18"}, {"amount": 6233.02, "department": "Marketing", "status": "Paused", "currency": "HTG", "job_title": "Senior Engineer", "signup_date": "2023-04-06"}, {"amount": 6021.24, "department": "Engineering", "status": "Churned", "currency": "DOP", "job_title": "Senior Engineer", "signup_date": "2023-11-15"}, {"amount": 5815.76, "department": "Marketing", "status": "Churned", "currency": "DOP", "job_title": "Engineering Manager", "signup_date": "2023-07-26"}, {"amount": 7159.63, "department": "Engineering", "status": "Churned", "currency": "DOP", "job_title": "Director", "signup_date": "2023-10-04"}], "dirty_cols": ["amount", "department", "status", "currency", "job_title", "signup_date", "extra"], "clean_cols": ["amount", "department", "status", "currency", "job_title", "signup_date"], "plan": {"dataset_summary": "60 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"accounting": "Finance", "Fin": "Finance", "ENG.": "Engineering", "Accounting": "Finance", "Eng.": "Engineering", "R&D": "Engineering", "Accounring": "Finance", "MKTG": "Marketing", "Mkt": "Marketing", "Dev": "Engineering", "FIN": "Finance", "r&d": "Engineering", "dev": "Engineering", "Eng": "Engineering", "marketing": "Marketing", "En.": "Engineering", "growth": "Marketing", "Mareting": "Marketing", "mkt": "Marketing", "ACCOUNTING": "Finance", "Growth": "Marketing", "GROWTH": "Marketing", "Finane": "Finance", "DEV": "Engineering", "marketign": "Marketing", "ENG": "Engineering", "MKT": "Marketing"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CANCELLED": "Churned", "ACTIVE": "Active", "canceled": "Churned", "active": "Active", "on hold": "Paused", "paused": "Paused", "churned": "Churned", "ative": "Active", "PAUSED": "Paused", "Actvie": "Active", "cancelled": "Churned", "PASED": "Paused", "CHURNED": "Churned", "Chutned": "Churned", "cancleled": "Churned", "Canclled": "Churned", "Cancelled": "Churned", "CHUREND": "Churned", "CHUDNED": "Churned"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Forint": "HUF", "Dominican Peso": "DOP", "forint": "HUF", "Huf": "HUF", "dop": "DOP", "Gourde": "HTG", "huf": "HUF", "bdt": "BDT", "Goudre": "HTG", "GOURDE": "HTG", "Taka": "BDT", "tnka": "BDT", "taka": "BDT", "Dop": "DOP", "Dominicn Peso": "DOP", "gorude": "HTG", "DOMINICAN PESO": "DOP", "Fxrint": "HUF", "dominican peso": "DOP", "gourde": "HTG", "Takka": "BDT"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"rep": "Sales Representative", "Sales Representatvie": "Sales Representative", "ENG MGR": "Engineering Manager", "Dir": "Director", "SalesRep": "Sales Representative", "dir.": "Director", "senior engineer": "Senior Engineer", "engineering mgr": "Engineering Manager", "dir": "Director", "direcor": "Director", "Sr Engineer": "Senior Engineer", "Rep": "Sales Representative", "eng mgr": "Engineering Manager", "DIR.": "Director", "director": "Director", "Sales Rep": "Sales Representative", "SALES REP": "Sales Representative", "Snr Engineer": "Senior Engineer", "engineering manager": "Engineering Manager", "sales representative": "Sales Representative", "DIR": "Director", "eng managr": "Engineering Manager", "sales rep": "Sales Representative", "Eng Mr": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "Eng Mgr": "Engineering Manager", "senior eng": "Senior Engineer", "snr engineer": "Senior Engineer", "Engbneering Manager": "Engineering Manager", "Salees Representative": "Sales Representative", "sr. enineer": "Senior Engineer", "Snz Engineer": "Senior Engineer", "Eng Manager": "Engineering Manager", "sr engineer": "Senior Engineer"}, "rationale": "Unified 34 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"currency": " cou", "amount": "$2,076.37", "industry": "Tech", "email": "   ACJGB@TEST.ORG", "phone": "(522)378-1204"}, {"currency": "htg", "amount": "9999999", "industry": "Retail", "email": " feekcn@mail.com ", "phone": "589-353-0483"}, {"currency": "COU", "amount": "766,04", "industry": "Teh", "email": "idfped@test.org", "phone": "584-835-9190"}, {"currency": "Unidad de Valor Real  ", "amount": "na", "industry": "  ecommerce", "email": "   bdcjk@test.org", "phone": "(558)940-8394"}, {"currency": "BURUNDI FRANC", "amount": "7.234,52", "industry": "retail & e-commerce", "email": "baaccnj@mail.com", "phone": "574-537-7220"}, {"currency": "Gourde", "amount": "6.136,63", "industry": "e-commerce", "email": "ppbbngl@mail.com", "phone": "(562)883-4115"}, {"currency": "COU ", "amount": "105,63", "industry": "Technology   ", "email": "  ghofg@mail.com", "phone": "(571)200-5359"}, {"currency": "Bueundi  Franc", "amount": "$7,857.12", "industry": "Technology", "email": "OMFK@CORP.IO", "phone": "544-992-2129"}, {"currency": "Costa Rican Colon", "amount": "$672.87", "industry": "Tehc", "email": "dpck@corp.io", "phone": "584.285.7126"}, {"currency": " bif ", "amount": "4.482,09", "industry": " TECH ", "email": "fmdlp@example.com", "phone": "5014259575"}, {"currency": "Gourde", "amount": "$7,825.27", "industry": "ecommerce", "email": " NAOCDB@TEST.ORG", "phone": "(554)703-0425"}, {"currency": "Cou", "amount": "N/A", "industry": "   Tech", "email": "BGGKN@MAIL.COM", "phone": "582.649.1165"}, {"currency": "Htg", "amount": "3.012,73", "industry": "  tech", "email": " impnc@example.com ", "phone": "(525)403-1820"}, {"currency": " bif", "amount": "$4,351.67", "industry": "tech", "email": "klfi@example.com", "phone": "(503)751-7604"}, {"currency": "Crc", "amount": "3.442,36", "industry": "technoology", "email": "DOOOCLI@CORP.IO", "phone": "526.641.9474"}, {"currency": "Gourde", "amount": "?", "industry": "Retail & E-commerce", "email": " BJIO@CORP.IO ", "phone": "592.804.2115"}, {"currency": "CRC", "amount": "$1,768.50", "industry": "retil", "email": "pmpjp@test.org", "phone": "(521)195-6469"}, {"currency": "goourde", "amount": "$3,382.91", "industry": "it", "email": "oiho@example.com", "phone": "550-543-6038"}, {"currency": "htg", "amount": "$3,791.58", "industry": "retal", "email": "  KDBI@EXAMPLE.COM", "phone": "539.695.3999"}, {"currency": "burundi franc", "amount": "3.217,73", "industry": "retail", "email": "kalibl@mail.com   ", "phone": "590.456.4290"}, {"currency": "Gourde", "amount": "N/A", "industry": "Tech", "email": "FPHIP@TEST.ORG", "phone": "(513)430-7700"}, {"currency": "Gourde", "amount": "6.425,18", "industry": "retail", "email": "IIKIONJ@EXAMPLE.COM", "phone": "(591)028-5394"}, {"currency": "Gourde", "amount": "6.136,63", "industry": "e-commerce", "email": "ppbbngl@mail.com", "phone": "(562)883-4115"}, {"currency": "burundi franc", "amount": "3.217,73", "industry": "retail", "email": "kalibl@mail.com   ", "phone": "590.456.4290"}, {"currency": "", "amount": "", "industry": "", "email": "", "phone": ""}], "clean": [{"currency": "COU", "amount": 2076.37, "industry": "Technology", "email": "acjgb@test.org", "phone": "(522) 378-1204"}, {"currency": "HTG", "amount": 9999999.0, "industry": "Retail", "email": "feekcn@mail.com", "phone": "(589) 353-0483"}, {"currency": "COU", "amount": 766.04, "industry": "Technology", "email": "idfped@test.org", "phone": "(584) 835-9190"}, {"currency": "COU", "amount": NaN, "industry": "Retail", "email": "bdcjk@test.org", "phone": "(558) 940-8394"}, {"currency": "BIF", "amount": 7234.52, "industry": "Retail", "email": "baaccnj@mail.com", "phone": "(574) 537-7220"}, {"currency": "HTG", "amount": 6136.63, "industry": "Retail", "email": "ppbbngl@mail.com", "phone": "(562) 883-4115"}, {"currency": "COU", "amount": 105.63, "industry": "Technology", "email": "ghofg@mail.com", "phone": "(571) 200-5359"}, {"currency": "BIF", "amount": 7857.12, "industry": "Technology", "email": "omfk@corp.io", "phone": "(544) 992-2129"}, {"currency": "CRC", "amount": 672.87, "industry": "Technology", "email": "dpck@corp.io", "phone": "(584) 285-7126"}, {"currency": "BIF", "amount": 4482.09, "industry": "Technology", "email": "fmdlp@example.com", "phone": "(501) 425-9575"}, {"currency": "HTG", "amount": 7825.27, "industry": "Retail", "email": "naocdb@test.org", "phone": "(554) 703-0425"}, {"currency": "COU", "amount": NaN, "industry": "Technology", "email": "bggkn@mail.com", "phone": "(582) 649-1165"}, {"currency": "HTG", "amount": 3012.73, "industry": "Technology", "email": "impnc@example.com", "phone": "(525) 403-1820"}, {"currency": "BIF", "amount": 4351.67, "industry": "Technology", "email": "klfi@example.com", "phone": "(503) 751-7604"}, {"currency": "CRC", "amount": 3442.36, "industry": "Technology", "email": "dooocli@corp.io", "phone": "(526) 641-9474"}, {"currency": "HTG", "amount": NaN, "industry": "Retail", "email": "bjio@corp.io", "phone": "(592) 804-2115"}, {"currency": "CRC", "amount": 1768.5, "industry": "Retail", "email": "pmpjp@test.org", "phone": "(521) 195-6469"}, {"currency": "HTG", "amount": 3382.91, "industry": "Technology", "email": "oiho@example.com", "phone": "(550) 543-6038"}, {"currency": "HTG", "amount": 3791.58, "industry": "Retail", "email": "kdbi@example.com", "phone": "(539) 695-3999"}, {"currency": "BIF", "amount": 3217.73, "industry": "Retail", "email": "kalibl@mail.com", "phone": "(590) 456-4290"}, {"currency": "HTG", "amount": NaN, "industry": "Technology", "email": "fphip@test.org", "phone": "(513) 430-7700"}, {"currency": "HTG", "amount": 6425.18, "industry": "Retail", "email": "iikionj@example.com", "phone": "(591) 028-5394"}], "dirty_cols": ["currency", "amount", "industry", "email", "phone"], "clean_cols": ["currency", "amount", "industry", "email", "phone"], "plan": {"dataset_summary": "25 rows × 5 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cou": "COU", "htg": "HTG", "Unidad de Valor Real": "COU", "BURUNDI FRANC": "BIF", "Gourde": "HTG", "Bueundi Franc": "BIF", "Costa Rican Colon": "CRC", "bif": "BIF", "Cou": "COU", "Htg": "HTG", "Crc": "CRC", "goourde": "HTG", "burundi franc": "BIF"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Tech": "Technology", "Teh": "Technology", "ecommerce": "Retail", "retail & e-commerce": "Retail", "e-commerce": "Retail", "Tehc": "Technology", "TECH": "Technology", "tech": "Technology", "technoology": "Technology", "Retail & E-commerce": "Retail", "retil": "Retail", "it": "Technology", "retal": "Retail", "retail": "Retail"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"email": "molgkp@test.org", "is_active": "Y", "phone": "520.521.8527", "country": "republic of uganda", "state": "Massachusetts", "company": "Vehement", "amount": "$92.49", "rate": "29.1%", "col_x": ""}, {"email": " ohhfp@example.com ", "is_active": "TRUE", "phone": "5542130378", "country": "GH", "state": " CA ", "company": "Soylent Corp", "amount": "5.053,03", "rate": "36.0%", "col_x": ""}, {"email": "   ichm@mail.com", "is_active": "Y", "phone": "542-610-6535", "country": "HMD", "state": "MP", "company": "  Cyberdyne", "amount": "$8,330.31", "rate": "34.6%", "col_x": ""}, {"email": "CJNMMJK@TEST.ORG", "is_active": "F", "phone": "(599)709-3365", "country": "DEMOCRATIC REPUBLIC OF SAO TOME AND PRINCIPE", "state": "NY", "company": " Hooli", "amount": "$685.03", "rate": "5.4%", "col_x": ""}, {"email": " HPKCKK@EXAMPLE.COM ", "is_active": "false", "phone": "564.864.3951", "country": "ST", "state": "ma", "company": "Hooli", "amount": "782,07", "rate": "67.5%", "col_x": ""}, {"email": "LKEJFHH@CORP.IO", "is_active": "TRUE", "phone": "5568614693", "country": " HMD ", "state": "new york", "company": "Vehement   ", "amount": "?", "rate": "72.6%", "col_x": ""}, {"email": "GIGNCMK@CORP.IO", "is_active": "FALSE", "phone": "5245657048", "country": "  Ug", "state": "Massacchusetts", "company": "Vehement ", "amount": "$4,616.26", "rate": "15.3%", "col_x": ""}, {"email": " anligee@test.org ", "is_active": "1", "phone": "(569)447-5187", "country": "STP", "state": " massachusetts ", "company": "Initech", "amount": "6.728,16", "rate": "69.0%", "col_x": ""}, {"email": " DBGOO@MAIL.COM ", "is_active": "Y", "phone": "(513)011-2619", "country": "hm", "state": "MP", "company": "Umbrella   ", "amount": "$2,577.72", "rate": "26.5%", "col_x": ""}, {"email": "KLCMB@EXAMPLE.COM   ", "is_active": "N", "phone": "5082514431", "country": "Heard Island and McDonald Islands", "state": "NORTHERN MARIANA ISLANDS", "company": "Hooli   ", "amount": "4.600,13", "rate": "2.2%", "col_x": ""}, {"email": "BNKIHFP@MAIL.COM", "is_active": "T", "phone": "527-927-9639", "country": "lr", "state": "ny", "company": "Stark Industries", "amount": "$8,492.60", "rate": "17.2%", "col_x": ""}, {"email": "HGDDHD@MAIL.COM", "is_active": "No", "phone": "560-427-9996", "country": "lbr", "state": "california", "company": "Acme Inc", "amount": "3.650,85", "rate": "98.4%", "col_x": ""}, {"email": " KCFFEK@MAIL.COM ", "is_active": "N", "phone": "5990001838", "country": "UG", "state": " MP ", "company": "Hooli   ", "amount": "$2,176.66", "rate": "12.2%", "col_x": ""}, {"email": "GLKCKGB@TEST.ORG", "is_active": "1", "phone": "559-930-8283", "country": "Ghana", "state": "MA", "company": "Cyberdyne   ", "amount": "$4,089.60", "rate": "73.2%", "col_x": ""}, {"email": "BFBM@CORP.IO   ", "is_active": "true", "phone": "539.481.2546", "country": "republic of uganda  ", "state": "Northern Mariana Islands", "company": " Hooli", "amount": "6.593,79", "rate": "42.3%", "col_x": ""}, {"email": "GEOIHGL@EXAMPLE.COM", "is_active": "N", "phone": "5352207278", "country": "Republic of Ghana", "state": " CALIFORNIA ", "company": "Initech", "amount": "8.409,44", "rate": "96.6%", "col_x": ""}, {"email": "iebe@example.com", "is_active": "FALSE", "phone": "5243323330", "country": "heard island and mcdonald islands", "state": "norshern mariana  islands", "company": "Wonka  Co", "amount": "1.778,18", "rate": "7.4%", "col_x": ""}, {"email": "GDKNPNG@CORP.IO", "is_active": "false", "phone": "(504)367-2675", "country": "HM", "state": "  Massachusetts", "company": " Hooli", "amount": "$8,376.65", "rate": "82.4%", "col_x": ""}, {"email": "dpei@example.com", "is_active": "T", "phone": "(575)389-3605", "country": "HMD", "state": "   CA", "company": "   Cyberdyne", "amount": "$396.59", "rate": "70.6%", "col_x": ""}, {"email": "jilbaf@mail.com", "is_active": "No", "phone": "(575)429-4899", "country": "Republic of Liberia", "state": "California", "company": "Acme  Inc", "amount": "8.267,14", "rate": "24.3%", "col_x": ""}, {"email": " hkammp@corp.io ", "is_active": "TRUE", "phone": "5794948713", "country": "Uganda", "state": "CA", "company": "Globex   ", "amount": "$4,868.24", "rate": "10.0%", "col_x": ""}, {"email": "   gfobkc@test.org", "is_active": "1", "phone": "5656718801", "country": "ghana", "state": "Massvchusetts", "company": "Stark  Industries", "amount": "-", "rate": "53.0%", "col_x": ""}, {"email": "HDHNHKI@TEST.ORG", "is_active": "N", "phone": "560-354-3479", "country": "Ug", "state": "Mhssachusetts", "company": " Vehement ", "amount": "2.952,52", "rate": "45.0%", "col_x": ""}, {"email": "  HMMBBCL@CORP.IO", "is_active": "FALSE", "phone": "562.880.3289", "country": "Republic of Liberia", "state": "CA", "company": "Initech ", "amount": "$8,652.14", "rate": "75.3%", "col_x": ""}, {"email": " CLKK@TEST.ORG", "is_active": "Yes", "phone": "5847369369", "country": "  lr", "state": "CA", "company": "Globex", "amount": "$53.10", "rate": "82.4%", "col_x": ""}, {"email": "hnfihln@test.org ", "is_active": "T", "phone": "553-027-3347", "country": " GHA", "state": "NORTHERN MARIANA ISLANDS", "company": "Stark  Industries", "amount": "1.409,26", "rate": "48.6%", "col_x": ""}, {"email": "DIBL@CORP.IO", "is_active": "TRUE", "phone": "500-571-8469", "country": "GH", "state": "NEW YORK", "company": "Initech", "amount": "-", "rate": "85.9%", "col_x": ""}, {"email": "LNIOB@TEST.ORG", "is_active": "F", "phone": "568.445.9292", "country": "lbr", "state": "NY", "company": "Initech", "amount": "$8,885.70", "rate": "4.3%", "col_x": ""}, {"email": "jgdlgom@test.org", "is_active": "T", "phone": "5141925046", "country": "Republi cOf Ghana", "state": "northern mariana islands", "company": "   Vehement", "amount": "$506.21", "rate": "20.4%", "col_x": ""}, {"email": "blkpme@example.com", "is_active": "Yes", "phone": "535.909.0330", "country": "Liberia", "state": "CALIFORNIA", "company": "Soylent  Corp", "amount": "1.377,40", "rate": "31.8%", "col_x": ""}, {"email": "lfand@example.com", "is_active": "true", "phone": "5103346814", "country": "uga", "state": "ny", "company": "  Soylent Corp", "amount": "$2,009.51", "rate": "61.5%", "col_x": ""}, {"email": "  jagj@example.com", "is_active": "TRUE", "phone": "577-637-4516", "country": " lbr ", "state": "ca", "company": "Wonka Co", "amount": "$4,663.38", "rate": "67.5%", "col_x": ""}, {"email": "FDOMI@TEST.ORG", "is_active": "F", "phone": "575.463.9589", "country": "GH", "state": "Neww York", "company": "Globex", "amount": "$8,476.74", "rate": "57.7%", "col_x": ""}, {"email": " NFNJJIG@EXAMPLE.COM ", "is_active": "FALSE", "phone": "595.423.1859", "country": "hmd", "state": "California", "company": "Stark Industries  ", "amount": "6.426,90", "rate": "1.7%", "col_x": ""}, {"email": "MCJBI@TEST.ORG ", "is_active": "Yes", "phone": "5432069477", "country": "HM", "state": " CA", "company": "Hooli", "amount": "$5,597.60", "rate": "27.8%", "col_x": ""}, {"email": " AOFM@TEST.ORG ", "is_active": "1", "phone": "(544)557-3012", "country": "UGA", "state": "MP", "company": "Vehement ", "amount": "$4,135.91", "rate": "81.0%", "col_x": ""}, {"email": "GNAKEG@CORP.IO", "is_active": "F", "phone": "(547)852-0177", "country": "republic of liberia", "state": "ma ", "company": "Wonka Co", "amount": "$8,314.96", "rate": "80.6%", "col_x": ""}, {"email": "  LACEHC@TEST.ORG", "is_active": "0", "phone": "(558)788-5063", "country": "heard island and mcdonald islands   ", "state": "MA", "company": "Vehement", "amount": "null", "rate": "47.9%", "col_x": ""}, {"email": "GNJMANJ@EXAMPLE.COM", "is_active": "FALSE", "phone": "(520)875-8877", "country": "Republic  of Liberia", "state": "massachusetts", "company": "Hooli", "amount": "8.274,44", "rate": "19.3%", "col_x": ""}, {"email": "clfd@example.com", "is_active": "F", "phone": "5566482118", "country": "Liberia", "state": "massachusetts", "company": "Vehement", "amount": "$7,296.27", "rate": "1.3%", "col_x": ""}, {"email": "FLPO@CORP.IO", "is_active": "FALSE", "phone": "5371283547", "country": "GH", "state": "California", "company": "   Initech", "amount": "$501.70", "rate": "84.8%", "col_x": ""}, {"email": " BEILEC@EXAMPLE.COM ", "is_active": "FALSE", "phone": "544-817-5090", "country": "gha   ", "state": "massachusetts", "company": "Wonka Co", "amount": "6.641,10", "rate": "66.9%", "col_x": ""}, {"email": "   bcocpa@corp.io", "is_active": "0", "phone": "566-864-1547", "country": "hmd", "state": "CA", "company": " Wonka Co", "amount": "2.583,59", "rate": "71.2%", "col_x": ""}, {"email": "ecfgepm@example.com", "is_active": "FALSE", "phone": "5498048467", "country": "HM   ", "state": " Califorfia", "company": "Wonka Co", "amount": "6.232,93", "rate": "22.5%", "col_x": ""}, {"email": "dkbbpc@corp.io", "is_active": "FALSE", "phone": "541-181-8244", "country": "HEARD ISLAND AND MCDONALD ISLANDS", "state": "NY", "company": "Vehement", "amount": "$3,357.31", "rate": "53.5%", "col_x": ""}, {"email": "KFMBNHL@EXAMPLE.COM ", "is_active": "0", "phone": "527.452.1035", "country": "GH", "state": "New York", "company": "Hooli", "amount": "4.445,74", "rate": "23.3%", "col_x": ""}, {"email": " ffog@corp.io ", "is_active": "true", "phone": "5162707492", "country": "STP", "state": "ny", "company": " Wonka Co", "amount": "8.443,83", "rate": "55.8%", "col_x": ""}, {"email": "   LCKNF@EXAMPLE.COM", "is_active": "true", "phone": "(530)977-2574", "country": "republic of uganda", "state": "california", "company": "Stark Industries ", "amount": "$8,548.91", "rate": "73.9%", "col_x": ""}, {"email": "EAFOBNI@EXAMPLE.COM", "is_active": "FALSE", "phone": "577.923.4426", "country": "democratic republic of sao tome and principe", "state": "New York", "company": "Soylent Corp", "amount": "7.273,73", "rate": "22.1%", "col_x": ""}, {"email": "HBOGM@EXAMPLE.COM", "is_active": "true", "phone": "587.466.1777", "country": "Sao Tome and Principe", "state": "NORTHERN MARIANA ISLANDS", "company": "Stark  Industries", "amount": "7.048,26", "rate": "35.8%", "col_x": ""}, {"email": "KPHNF@EXAMPLE.COM", "is_active": "true", "phone": "5292820520", "country": "Heard Island And Mcdonald Islands", "state": "MP", "company": "Hooli ", "amount": "$7,991.57", "rate": "23.4%", "col_x": ""}, {"email": "lggdjp@test.org", "is_active": "1", "phone": "5960339184", "country": "  Liberia", "state": "California", "company": "Globex", "amount": "2.180,71", "rate": "66.7%", "col_x": ""}, {"email": "   LCKNF@EXAMPLE.COM", "is_active": "true", "phone": "(530)977-2574", "country": "republic of uganda", "state": "california", "company": "Stark Industries ", "amount": "$8,548.91", "rate": "73.9%", "col_x": ""}], "clean": [{"email": "molgkp@test.org", "is_active": true, "phone": "(520) 521-8527", "country": "Uganda", "state": "Massachusetts", "company": "Vehement", "amount": 92.49, "rate": 0.29100000000000004}, {"email": "ohhfp@example.com", "is_active": true, "phone": "(554) 213-0378", "country": "Ghana", "state": "California", "company": "Soylent Corp", "amount": 5053.03, "rate": 0.36}, {"email": "ichm@mail.com", "is_active": true, "phone": "(542) 610-6535", "country": "Heard Island and McDonald Islands", "state": "Northern Mariana Islands", "company": "Cyberdyne", "amount": 8330.31, "rate": 0.34600000000000003}, {"email": "cjnmmjk@test.org", "is_active": false, "phone": "(599) 709-3365", "country": "Sao Tome and Principe", "state": "New York", "company": "Hooli", "amount": 685.03, "rate": 0.054000000000000006}, {"email": "hpkckk@example.com", "is_active": false, "phone": "(564) 864-3951", "country": "Sao Tome and Principe", "state": "Massachusetts", "company": "Hooli", "amount": 782.07, "rate": 0.675}, {"email": "lkejfhh@corp.io", "is_active": true, "phone": "(556) 861-4693", "country": "Heard Island and McDonald Islands", "state": "New York", "company": "Vehement", "amount": NaN, "rate": 0.726}, {"email": "gigncmk@corp.io", "is_active": false, "phone": "(524) 565-7048", "country": "Uganda", "state": "Massachusetts", "company": "Vehement", "amount": 4616.26, "rate": 0.153}, {"email": "anligee@test.org", "is_active": true, "phone": "(569) 447-5187", "country": "Sao Tome and Principe", "state": "Massachusetts", "company": "Initech", "amount": 6728.16, "rate": 0.69}, {"email": "dbgoo@mail.com", "is_active": true, "phone": "(513) 011-2619", "country": "Heard Island and McDonald Islands", "state": "Northern Mariana Islands", "company": "Umbrella", "amount": 2577.72, "rate": 0.265}, {"email": "klcmb@example.com", "is_active": false, "phone": "(508) 251-4431", "country": "Heard Island and McDonald Islands", "state": "Northern Mariana Islands", "company": "Hooli", "amount": 4600.13, "rate": 0.022000000000000002}, {"email": "bnkihfp@mail.com", "is_active": true, "phone": "(527) 927-9639", "country": "Liberia", "state": "New York", "company": "Stark Industries", "amount": 8492.6, "rate": 0.172}, {"email": "hgddhd@mail.com", "is_active": false, "phone": "(560) 427-9996", "country": "Liberia", "state": "California", "company": "Acme Inc", "amount": 3650.85, "rate": 0.9840000000000001}, {"email": "kcffek@mail.com", "is_active": false, "phone": "(599) 000-1838", "country": "Uganda", "state": "Northern Mariana Islands", "company": "Hooli", "amount": 2176.66, "rate": 0.122}, {"email": "glkckgb@test.org", "is_active": true, "phone": "(559) 930-8283", "country": "Ghana", "state": "Massachusetts", "company": "Cyberdyne", "amount": 4089.6, "rate": 0.732}, {"email": "bfbm@corp.io", "is_active": true, "phone": "(539) 481-2546", "country": "Uganda", "state": "Northern Mariana Islands", "company": "Hooli", "amount": 6593.79, "rate": 0.423}, {"email": "geoihgl@example.com", "is_active": false, "phone": "(535) 220-7278", "country": "Ghana", "state": "California", "company": "Initech", "amount": 8409.44, "rate": 0.966}, {"email": "iebe@example.com", "is_active": false, "phone": "(524) 332-3330", "country": "Heard Island and McDonald Islands", "state": "Northern Mariana Islands", "company": "Wonka Co", "amount": 1778.18, "rate": 0.07400000000000001}, {"email": "gdknpng@corp.io", "is_active": false, "phone": "(504) 367-2675", "country": "Heard Island and McDonald Islands", "state": "Massachusetts", "company": "Hooli", "amount": 8376.65, "rate": 0.8240000000000001}, {"email": "dpei@example.com", "is_active": true, "phone": "(575) 389-3605", "country": "Heard Island and McDonald Islands", "state": "California", "company": "Cyberdyne", "amount": 396.59, "rate": 0.706}, {"email": "jilbaf@mail.com", "is_active": false, "phone": "(575) 429-4899", "country": "Liberia", "state": "California", "company": "Acme Inc", "amount": 8267.14, "rate": 0.243}, {"email": "hkammp@corp.io", "is_active": true, "phone": "(579) 494-8713", "country": "Uganda", "state": "California", "company": "Globex", "amount": 4868.24, "rate": 0.1}, {"email": "gfobkc@test.org", "is_active": true, "phone": "(565) 671-8801", "country": "Ghana", "state": "Massachusetts", "company": "Stark Industries", "amount": NaN, "rate": 0.53}, {"email": "hdhnhki@test.org", "is_active": false, "phone": "(560) 354-3479", "country": "Uganda", "state": "Massachusetts", "company": "Vehement", "amount": 2952.52, "rate": 0.45}, {"email": "hmmbbcl@corp.io", "is_active": false, "phone": "(562) 880-3289", "country": "Liberia", "state": "California", "company": "Initech", "amount": 8652.14, "rate": 0.753}, {"email": "clkk@test.org", "is_active": true, "phone": "(584) 736-9369", "country": "Liberia", "state": "California", "company": "Globex", "amount": 53.1, "rate": 0.8240000000000001}, {"email": "hnfihln@test.org", "is_active": true, "phone": "(553) 027-3347", "country": "Ghana", "state": "Northern Mariana Islands", "company": "Stark Industries", "amount": 1409.26, "rate": 0.486}, {"email": "dibl@corp.io", "is_active": true, "phone": "(500) 571-8469", "country": "Ghana", "state": "New York", "company": "Initech", "amount": NaN, "rate": 0.8590000000000001}, {"email": "lniob@test.org", "is_active": false, "phone": "(568) 445-9292", "country": "Liberia", "state": "New York", "company": "Initech", "amount": 8885.7, "rate": 0.043}, {"email": "jgdlgom@test.org", "is_active": true, "phone": "(514) 192-5046", "country": "Ghana", "state": "Northern Mariana Islands", "company": "Vehement", "amount": 506.21, "rate": 0.204}, {"email": "blkpme@example.com", "is_active": true, "phone": "(535) 909-0330", "country": "Liberia", "state": "California", "company": "Soylent Corp", "amount": 1377.4, "rate": 0.318}, {"email": "lfand@example.com", "is_active": true, "phone": "(510) 334-6814", "country": "Uganda", "state": "New York", "company": "Soylent Corp", "amount": 2009.51, "rate": 0.615}, {"email": "jagj@example.com", "is_active": true, "phone": "(577) 637-4516", "country": "Liberia", "state": "California", "company": "Wonka Co", "amount": 4663.38, "rate": 0.675}, {"email": "fdomi@test.org", "is_active": false, "phone": "(575) 463-9589", "country": "Ghana", "state": "New York", "company": "Globex", "amount": 8476.74, "rate": 0.5770000000000001}, {"email": "nfnjjig@example.com", "is_active": false, "phone": "(595) 423-1859", "country": "Heard Island and McDonald Islands", "state": "California", "company": "Stark Industries", "amount": 6426.9, "rate": 0.017}, {"email": "mcjbi@test.org", "is_active": true, "phone": "(543) 206-9477", "country": "Heard Island and McDonald Islands", "state": "California", "company": "Hooli", "amount": 5597.6, "rate": 0.278}, {"email": "aofm@test.org", "is_active": true, "phone": "(544) 557-3012", "country": "Uganda", "state": "Northern Mariana Islands", "company": "Vehement", "amount": 4135.91, "rate": 0.81}, {"email": "gnakeg@corp.io", "is_active": false, "phone": "(547) 852-0177", "country": "Liberia", "state": "Massachusetts", "company": "Wonka Co", "amount": 8314.96, "rate": 0.8059999999999999}, {"email": "lacehc@test.org", "is_active": false, "phone": "(558) 788-5063", "country": "Heard Island and McDonald Islands", "state": "Massachusetts", "company": "Vehement", "amount": NaN, "rate": 0.479}, {"email": "gnjmanj@example.com", "is_active": false, "phone": "(520) 875-8877", "country": "Liberia", "state": "Massachusetts", "company": "Hooli", "amount": 8274.44, "rate": 0.193}, {"email": "clfd@example.com", "is_active": false, "phone": "(556) 648-2118", "country": "Liberia", "state": "Massachusetts", "company": "Vehement", "amount": 7296.27, "rate": 0.013000000000000001}, {"email": "flpo@corp.io", "is_active": false, "phone": "(537) 128-3547", "country": "Ghana", "state": "California", "company": "Initech", "amount": 501.7, "rate": 0.848}, {"email": "beilec@example.com", "is_active": false, "phone": "(544) 817-5090", "country": "Ghana", "state": "Massachusetts", "company": "Wonka Co", "amount": 6641.1, "rate": 0.669}, {"email": "bcocpa@corp.io", "is_active": false, "phone": "(566) 864-1547", "country": "Heard Island and McDonald Islands", "state": "California", "company": "Wonka Co", "amount": 2583.59, "rate": 0.7120000000000001}, {"email": "ecfgepm@example.com", "is_active": false, "phone": "(549) 804-8467", "country": "Heard Island and McDonald Islands", "state": "California", "company": "Wonka Co", "amount": 6232.93, "rate": 0.225}, {"email": "dkbbpc@corp.io", "is_active": false, "phone": "(541) 181-8244", "country": "Heard Island and McDonald Islands", "state": "New York", "company": "Vehement", "amount": 3357.31, "rate": 0.535}, {"email": "kfmbnhl@example.com", "is_active": false, "phone": "(527) 452-1035", "country": "Ghana", "state": "New York", "company": "Hooli", "amount": 4445.74, "rate": 0.233}, {"email": "ffog@corp.io", "is_active": true, "phone": "(516) 270-7492", "country": "Sao Tome and Principe", "state": "New York", "company": "Wonka Co", "amount": 8443.83, "rate": 0.5579999999999999}, {"email": "lcknf@example.com", "is_active": true, "phone": "(530) 977-2574", "country": "Uganda", "state": "California", "company": "Stark Industries", "amount": 8548.91, "rate": 0.7390000000000001}, {"email": "eafobni@example.com", "is_active": false, "phone": "(577) 923-4426", "country": "Sao Tome and Principe", "state": "New York", "company": "Soylent Corp", "amount": 7273.73, "rate": 0.221}, {"email": "hbogm@example.com", "is_active": true, "phone": "(587) 466-1777", "country": "Sao Tome and Principe", "state": "Northern Mariana Islands", "company": "Stark Industries", "amount": 7048.26, "rate": 0.358}, {"email": "kphnf@example.com", "is_active": true, "phone": "(529) 282-0520", "country": "Heard Island and McDonald Islands", "state": "Northern Mariana Islands", "company": "Hooli", "amount": 7991.57, "rate": 0.23399999999999999}, {"email": "lggdjp@test.org", "is_active": true, "phone": "(596) 033-9184", "country": "Liberia", "state": "California", "company": "Globex", "amount": 2180.71, "rate": 0.667}], "dirty_cols": ["email", "is_active", "phone", "country", "state", "company", "amount", "rate", "col_x"], "clean_cols": ["email", "is_active", "phone", "country", "state", "company", "amount", "rate"], "plan": {"dataset_summary": "53 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"republic of uganda": "Uganda", "GH": "Ghana", "HMD": "Heard Island and McDonald Islands", "DEMOCRATIC REPUBLIC OF SAO TOME AND PRINCIPE": "Sao Tome and Principe", "ST": "Sao Tome and Principe", "Ug": "Uganda", "STP": "Sao Tome and Principe", "hm": "Heard Island and McDonald Islands", "lr": "Liberia", "lbr": "Liberia", "UG": "Uganda", "Republic of Ghana": "Ghana", "heard island and mcdonald islands": "Heard Island and McDonald Islands", "HM": "Heard Island and McDonald Islands", "Republic of Liberia": "Liberia", "ghana": "Ghana", "GHA": "Ghana", "Republi cOf Ghana": "Ghana", "uga": "Uganda", "hmd": "Heard Island and McDonald Islands", "UGA": "Uganda", "republic of liberia": "Liberia", "gha": "Ghana", "HEARD ISLAND AND MCDONALD ISLANDS": "Heard Island and McDonald Islands", "democratic republic of sao tome and principe": "Sao Tome and Principe", "Heard Island And Mcdonald Islands": "Heard Island and McDonald Islands"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CA": "California", "MP": "Northern Mariana Islands", "NY": "New York", "ma": "Massachusetts", "new york": "New York", "Massacchusetts": "Massachusetts", "massachusetts": "Massachusetts", "NORTHERN MARIANA ISLANDS": "Northern Mariana Islands", "ny": "New York", "california": "California", "MA": "Massachusetts", "CALIFORNIA": "California", "norshern mariana islands": "Northern Mariana Islands", "Massvchusetts": "Massachusetts", "Mhssachusetts": "Massachusetts", "NEW YORK": "New York", "northern mariana islands": "Northern Mariana Islands", "ca": "California", "Neww York": "New York", "Califorfia": "California"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"status": "High", "country": "FRO", "name": "Eve Adams", "industry": "Property", "city": "Tanggu", "col_x": ""}, {"status": "HIGH", "country": "   Idn", "name": "Sara Cruz", "industry": " retail ", "city": " tanggu ", "col_x": ""}, {"status": "  low", "country": "FO", "name": "Priya Adams", "industry": "real zstate", "city": "Danjiangkou", "col_x": ""}, {"status": "   Med", "country": "TV", "name": "  Priya Adams", "industry": "Industrial", "city": "Danjiangkou", "col_x": ""}, {"status": "H ", "country": "FO  ", "name": "David Park", "industry": "Real Esttte", "city": " Danjiangkou ", "col_x": ""}, {"status": "medoum", "country": "Åland Islands", "name": "Priya Ali", "industry": "Retail", "city": "DANJIANGKOU", "col_x": ""}, {"status": "P3", "country": "FO", "name": "Grace Moore", "industry": "Industrial", "city": "Tanggu", "col_x": ""}, {"status": "high", "country": "ID", "name": "Grace Johnson", "industry": "inbustrial", "city": "Tanggu   ", "col_x": ""}, {"status": "Hiigh", "country": "tuv", "name": "   Frank Adams", "industry": "property", "city": "Tanggu", "col_x": ""}, {"status": "L", "country": "Idn", "name": "Karl Cruz ", "industry": "REALESTATE", "city": "Tanggu", "col_x": ""}, {"status": "med", "country": "fo", "name": "  Priya Lee", "industry": " Industrial ", "city": " tanggu ", "col_x": ""}, {"status": "p3", "country": "Tuvalu", "name": "Carol Cruz ", "industry": "industrial", "city": "danjiangkou", "col_x": ""}, {"status": "med", "country": "TV", "name": "   Frank Lee", "industry": "property", "city": "TANGGU  ", "col_x": ""}, {"status": "low", "country": "   FO", "name": "Omar  Brandt", "industry": "industrial", "city": "tanggu", "col_x": ""}, {"status": "medium", "country": "TUV", "name": "Grace  Johnson", "industry": "e-commerce", "city": "tanggu", "col_x": ""}, {"status": "P1", "country": "faroe islands  ", "name": "Ivan Petrov   ", "industry": "Real Estate", "city": "Tanggu", "col_x": ""}, {"status": "medium", "country": "FRO", "name": "Bob Lee", "industry": " PropTech ", "city": "Danjiangksu", "col_x": ""}, {"status": "l", "country": "åland islaods", "name": "Eve  Ali", "industry": "retail", "city": " tanggu ", "col_x": ""}, {"status": "h", "country": " ALA ", "name": " Mona Moore", "industry": "industrial", "city": "tanggu", "col_x": ""}, {"status": "High", "country": "TV", "name": "  Bob Adams", "industry": "   mfg", "city": " Tanggu", "col_x": ""}, {"status": "H", "country": "tv", "name": "   Lena Reyes", "industry": "mfg", "city": "Tanggu", "col_x": ""}, {"status": "L", "country": "Indonesia", "name": "Mona  Brandt", "industry": " mfg ", "city": "Danjiangkou", "col_x": ""}, {"status": "Low", "country": "Faroe Islands  ", "name": " Ivan Smith", "industry": " retail", "city": "TANGGU", "col_x": ""}, {"status": "p1", "country": "TUUVALU", "name": " Ivan Smith", "industry": "Retail  & E-commece", "city": "Danjiangkou", "col_x": ""}, {"status": "med", "country": "AX", "name": "Ivan  Diaz", "industry": "RealEstate", "city": "tanggu", "col_x": ""}, {"status": "HIGH", "country": "Ala", "name": "Sara Wong", "industry": "  Retail", "city": "tanggu", "col_x": ""}, {"status": "p1", "country": "Id  ", "name": "Lena Lee  ", "industry": "E-COMMERCE", "city": " Tanggu ", "col_x": ""}, {"status": "M", "country": "FRO", "name": "Judy  Johnson", "industry": "industrial", "city": "TANGGU", "col_x": ""}, {"status": "high", "country": "tuv", "name": "Eve  Petrov", "industry": "real estate", "city": "   TNGGU", "col_x": ""}, {"status": "H", "country": "idn", "name": "Ivan Brandt  ", "industry": "proptech", "city": "Danjiangkou", "col_x": ""}, {"status": "LOW", "country": "fo", "name": "Alice  Johnson", "industry": "RETAIL", "city": "tanggu", "col_x": ""}, {"status": "Medium", "country": "TUV", "name": "Frank Lee", "industry": "  E-commerce", "city": "Danjiangkou", "col_x": ""}, {"status": " Medium", "country": "land Islands", "name": "Grace Smith", "industry": "Retail", "city": "tanggu  ", "col_x": ""}], "clean": [{"status": "High", "country": "Faroe Islands", "name": "Eve Adams", "industry": "Real Estate", "city": "Tanggu"}, {"status": "High", "country": "Indonesia", "name": "Sara Cruz", "industry": "Retail", "city": "Tanggu"}, {"status": "Low", "country": "Faroe Islands", "name": "Priya Adams", "industry": "Real Estate", "city": "Danjiangkou"}, {"status": "Medium", "country": "Tuvalu", "name": "Priya Adams", "industry": "Manufacturing", "city": "Danjiangkou"}, {"status": "High", "country": "Faroe Islands", "name": "David Park", "industry": "Real Estate", "city": "Danjiangkou"}, {"status": "Medium", "country": "Åland Islands", "name": "Priya Ali", "industry": "Retail", "city": "Danjiangkou"}, {"status": "Low", "country": "Faroe Islands", "name": "Grace Moore", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "High", "country": "Indonesia", "name": "Grace Johnson", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "High", "country": "Tuvalu", "name": "Frank Adams", "industry": "Real Estate", "city": "Tanggu"}, {"status": "Low", "country": "Indonesia", "name": "Karl Cruz", "industry": "Real Estate", "city": "Tanggu"}, {"status": "Medium", "country": "Faroe Islands", "name": "Priya Lee", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "Low", "country": "Tuvalu", "name": "Carol Cruz", "industry": "Manufacturing", "city": "Danjiangkou"}, {"status": "Medium", "country": "Tuvalu", "name": "Frank Lee", "industry": "Real Estate", "city": "Tanggu"}, {"status": "Low", "country": "Faroe Islands", "name": "Omar Brandt", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "Medium", "country": "Tuvalu", "name": "Grace Johnson", "industry": "Retail", "city": "Tanggu"}, {"status": "High", "country": "Faroe Islands", "name": "Ivan Petrov", "industry": "Real Estate", "city": "Tanggu"}, {"status": "Medium", "country": "Faroe Islands", "name": "Bob Lee", "industry": "Real Estate", "city": "Danjiangkou"}, {"status": "Low", "country": "Åland Islands", "name": "Eve Ali", "industry": "Retail", "city": "Tanggu"}, {"status": "High", "country": "Åland Islands", "name": "Mona Moore", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "High", "country": "Tuvalu", "name": "Bob Adams", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "High", "country": "Tuvalu", "name": "Lena Reyes", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "Low", "country": "Indonesia", "name": "Mona Brandt", "industry": "Manufacturing", "city": "Danjiangkou"}, {"status": "Low", "country": "Faroe Islands", "name": "Ivan Smith", "industry": "Retail", "city": "Tanggu"}, {"status": "High", "country": "Tuvalu", "name": "Ivan Smith", "industry": "Retail", "city": "Danjiangkou"}, {"status": "Medium", "country": "Åland Islands", "name": "Ivan Diaz", "industry": "Real Estate", "city": "Tanggu"}, {"status": "High", "country": "Åland Islands", "name": "Sara Wong", "industry": "Retail", "city": "Tanggu"}, {"status": "High", "country": "Indonesia", "name": "Lena Lee", "industry": "Retail", "city": "Tanggu"}, {"status": "Medium", "country": "Faroe Islands", "name": "Judy Johnson", "industry": "Manufacturing", "city": "Tanggu"}, {"status": "High", "country": "Tuvalu", "name": "Eve Petrov", "industry": "Real Estate", "city": "Tanggu"}, {"status": "High", "country": "Indonesia", "name": "Ivan Brandt", "industry": "Real Estate", "city": "Danjiangkou"}, {"status": "Low", "country": "Faroe Islands", "name": "Alice Johnson", "industry": "Retail", "city": "Tanggu"}, {"status": "Medium", "country": "Tuvalu", "name": "Frank Lee", "industry": "Retail", "city": "Danjiangkou"}, {"status": "Medium", "country": "Åland Islands", "name": "Grace Smith", "industry": "Retail", "city": "Tanggu"}], "dirty_cols": ["status", "country", "name", "industry", "city", "col_x"], "clean_cols": ["status", "country", "name", "industry", "city"], "plan": {"dataset_summary": "33 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HIGH": "High", "low": "Low", "Med": "Medium", "H": "High", "medoum": "Medium", "P3": "Low", "high": "High", "Hiigh": "High", "L": "Low", "med": "Medium", "p3": "Low", "medium": "Medium", "P1": "High", "l": "Low", "h": "High", "p1": "High", "M": "Medium", "LOW": "Low"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"FRO": "Faroe Islands", "Idn": "Indonesia", "FO": "Faroe Islands", "TV": "Tuvalu", "ID": "Indonesia", "tuv": "Tuvalu", "fo": "Faroe Islands", "TUV": "Tuvalu", "faroe islands": "Faroe Islands", "åland islaods": "Åland Islands", "ALA": "Åland Islands", "tv": "Tuvalu", "TUUVALU": "Tuvalu", "AX": "Åland Islands", "Ala": "Åland Islands", "Id": "Indonesia", "idn": "Indonesia", "land Islands": "Åland Islands"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Property": "Real Estate", "retail": "Retail", "real zstate": "Real Estate", "Industrial": "Manufacturing", "Real Esttte": "Real Estate", "inbustrial": "Manufacturing", "property": "Real Estate", "REALESTATE": "Real Estate", "industrial": "Manufacturing", "e-commerce": "Retail", "PropTech": "Real Estate", "mfg": "Manufacturing", "Retail & E-commece": "Retail", "RealEstate": "Real Estate", "E-COMMERCE": "Retail", "real estate": "Real Estate", "proptech": "Real Estate", "RETAIL": "Retail", "E-commerce": "Retail"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"tanggu": "Tanggu", "DANJIANGKOU": "Danjiangkou", "danjiangkou": "Danjiangkou", "TANGGU": "Tanggu", "Danjiangksu": "Danjiangkou", "TNGGU": "Tanggu"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "Berlin  ", "rate": "2.1%", "currency": "Danish Krone", "status": "PAID", "email": "pfcnjgp@example.com  ", "col_x": ""}, {"city": "N'zeto", "rate": "85.7%", "currency": "DALASI", "status": "PAID", "email": "jpocpnh@test.org", "col_x": ""}, {"city": "N'ZETO", "rate": "63.1%", "currency": "Egp", "status": "PAID", "email": "nghg@test.org", "col_x": ""}, {"city": " berlin", "rate": "0.4%", "currency": "CUP", "status": "Overdue", "email": " cfpbc@test.org ", "col_x": ""}, {"city": "berlin", "rate": "63.8%", "currency": "DKK", "status": "Overdue", "email": "pgfhh@example.com", "col_x": ""}, {"city": "Nzeto", "rate": "99.7%", "currency": "EGP", "status": " late", "email": "gdhl@mail.com", "col_x": ""}, {"city": "BERLIN", "rate": "64.6%", "currency": " EGP ", "status": "OVERDUE", "email": "hkop@example.com", "col_x": ""}, {"city": "n'seto", "rate": "97.9%", "currency": " GMD ", "status": "late", "email": "   BFAE@MAIL.COM", "col_x": ""}, {"city": " N'zeto", "rate": "99.8%", "currency": "dkk", "status": "Paid", "email": " kcndm@mail.com", "col_x": ""}, {"city": "Nezyobnaya", "rate": "12.5%", "currency": "Egyptian Pound", "status": "paid", "email": "   kldmalc@test.org", "col_x": ""}, {"city": "   Berlin", "rate": "50.2%", "currency": "EGP", "status": " Paid", "email": "bpgcp@mail.com", "col_x": ""}, {"city": " Berlin", "rate": "88.9%", "currency": "gmd", "status": "Paid", "email": " ENFAN@MAIL.COM ", "col_x": ""}, {"city": "Nezlobnaya", "rate": "40.4%", "currency": "GMD", "status": "   oevrdue", "email": "GLAMN@MAIL.COM", "col_x": ""}, {"city": "Berliin", "rate": "90.3%", "currency": "Cuban Peso", "status": "Late", "email": "fjhh@mail.com", "col_x": ""}, {"city": " N'ZETO ", "rate": "29.8%", "currency": "danish krone", "status": "paid", "email": " IKLIFGI@CORP.IO", "col_x": ""}, {"city": "N'zeto", "rate": "70.3%", "currency": "CUP   ", "status": "   overdue", "email": "DLHOJ@TEST.ORG", "col_x": ""}, {"city": "nezlobnaya", "rate": "54.1%", "currency": "Cuban  Peso", "status": "paid", "email": "   MIBD@CORP.IO", "col_x": ""}, {"city": " N'zeto ", "rate": "59.5%", "currency": "DKK", "status": "PAID", "email": " hnjoae@test.org", "col_x": ""}, {"city": "   NEZLOBNAYA", "rate": "84.0%", "currency": "EGP", "status": "OVERDUE", "email": "JPLCOED@EXAMPLE.COM", "col_x": ""}, {"city": " Nezlobnaya", "rate": "97.1%", "currency": "Egp", "status": "late", "email": "bkfdk@mail.com", "col_x": ""}, {"city": "N'zeto", "rate": "43.3%", "currency": "CUP", "status": "OVERDUUE", "email": " DOHKF@CORP.IO", "col_x": ""}, {"city": "N'zeto", "rate": "95.2%", "currency": "Egyptian  Pound", "status": "PAID", "email": "bkle@corp.io", "col_x": ""}, {"city": "   berlin", "rate": "86.9%", "currency": "DANISH KRNOE   ", "status": "paid", "email": "ncbd@test.org", "col_x": ""}, {"city": "Nezlobnaya", "rate": "30.3%", "currency": "DKK", "status": "OVERDUE", "email": "MJJIPO@CORP.IO ", "col_x": ""}, {"city": "N'Zeto", "rate": "50.9%", "currency": "cup", "status": "PAID", "email": " OHADH@CORP.IO", "col_x": ""}, {"city": " N'ZETO ", "rate": "56.8%", "currency": " CUP ", "status": "overdue", "email": "MHCA@TEST.ORG", "col_x": ""}, {"city": "   NEZLOBNAYA", "rate": "84.0%", "currency": "EGP", "status": "OVERDUE", "email": "JPLCOED@EXAMPLE.COM", "col_x": ""}, {"city": "", "rate": "", "currency": "", "status": "", "email": "", "col_x": ""}], "clean": [{"city": "Berlin", "rate": 0.021, "currency": "DKK", "status": "Paid", "email": "pfcnjgp@example.com"}, {"city": "N'zeto", "rate": 0.857, "currency": "GMD", "status": "Paid", "email": "jpocpnh@test.org"}, {"city": "N'zeto", "rate": 0.631, "currency": "EGP", "status": "Paid", "email": "nghg@test.org"}, {"city": "Berlin", "rate": 0.004, "currency": "CUP", "status": "Overdue", "email": "cfpbc@test.org"}, {"city": "Berlin", "rate": 0.638, "currency": "DKK", "status": "Overdue", "email": "pgfhh@example.com"}, {"city": "N'zeto", "rate": 0.997, "currency": "EGP", "status": "Overdue", "email": "gdhl@mail.com"}, {"city": "Berlin", "rate": 0.6459999999999999, "currency": "EGP", "status": "Overdue", "email": "hkop@example.com"}, {"city": "N'zeto", "rate": 0.9790000000000001, "currency": "GMD", "status": "Overdue", "email": "bfae@mail.com"}, {"city": "N'zeto", "rate": 0.998, "currency": "DKK", "status": "Paid", "email": "kcndm@mail.com"}, {"city": "Nezlobnaya", "rate": 0.125, "currency": "EGP", "status": "Paid", "email": "kldmalc@test.org"}, {"city": "Berlin", "rate": 0.502, "currency": "EGP", "status": "Paid", "email": "bpgcp@mail.com"}, {"city": "Berlin", "rate": 0.889, "currency": "GMD", "status": "Paid", "email": "enfan@mail.com"}, {"city": "Nezlobnaya", "rate": 0.40399999999999997, "currency": "GMD", "status": "Overdue", "email": "glamn@mail.com"}, {"city": "Berlin", "rate": 0.903, "currency": "CUP", "status": "Overdue", "email": "fjhh@mail.com"}, {"city": "N'zeto", "rate": 0.298, "currency": "DKK", "status": "Paid", "email": "iklifgi@corp.io"}, {"city": "N'zeto", "rate": 0.703, "currency": "CUP", "status": "Overdue", "email": "dlhoj@test.org"}, {"city": "Nezlobnaya", "rate": 0.541, "currency": "CUP", "status": "Paid", "email": "mibd@corp.io"}, {"city": "N'zeto", "rate": 0.595, "currency": "DKK", "status": "Paid", "email": "hnjoae@test.org"}, {"city": "Nezlobnaya", "rate": 0.84, "currency": "EGP", "status": "Overdue", "email": "jplcoed@example.com"}, {"city": "Nezlobnaya", "rate": 0.971, "currency": "EGP", "status": "Overdue", "email": "bkfdk@mail.com"}, {"city": "N'zeto", "rate": 0.433, "currency": "CUP", "status": "Overdue", "email": "dohkf@corp.io"}, {"city": "N'zeto", "rate": 0.9520000000000001, "currency": "EGP", "status": "Paid", "email": "bkle@corp.io"}, {"city": "Berlin", "rate": 0.8690000000000001, "currency": "DKK", "status": "Paid", "email": "ncbd@test.org"}, {"city": "Nezlobnaya", "rate": 0.303, "currency": "DKK", "status": "Overdue", "email": "mjjipo@corp.io"}, {"city": "N'zeto", "rate": 0.509, "currency": "CUP", "status": "Paid", "email": "ohadh@corp.io"}, {"city": "N'zeto", "rate": 0.568, "currency": "CUP", "status": "Overdue", "email": "mhca@test.org"}], "dirty_cols": ["city", "rate", "currency", "status", "email", "col_x"], "clean_cols": ["city", "rate", "currency", "status", "email"], "plan": {"dataset_summary": "28 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"N'ZETO": "N'zeto", "berlin": "Berlin", "Nzeto": "N'zeto", "BERLIN": "Berlin", "n'seto": "N'zeto", "Nezyobnaya": "Nezlobnaya", "Berliin": "Berlin", "nezlobnaya": "Nezlobnaya", "NEZLOBNAYA": "Nezlobnaya", "N'Zeto": "N'zeto"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Danish Krone": "DKK", "DALASI": "GMD", "Egp": "EGP", "dkk": "DKK", "Egyptian Pound": "EGP", "gmd": "GMD", "Cuban Peso": "CUP", "danish krone": "DKK", "DANISH KRNOE": "DKK", "cup": "CUP"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PAID": "Paid", "late": "Overdue", "OVERDUE": "Overdue", "paid": "Paid", "oevrdue": "Overdue", "Late": "Overdue", "overdue": "Overdue", "OVERDUUE": "Overdue"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"phone": "536.744.0665", "amount": "7.828,66", "email": "jiniap@example.com", "company": "Acme Inc", "status": "OVERDUE", "unit": " in", "unnamed": ""}, {"phone": "5912075023", "amount": "None", "email": "obbhj@test.org", "company": "Cyberdyne", "status": "unpaid", "unit": "Centimeters", "unnamed": ""}, {"phone": "543.917.2997", "amount": "1.944,68", "email": " MDLCF@TEST.ORG", "company": " Soylent Corp", "status": " LATE ", "unit": "km", "unnamed": ""}, {"phone": "580-612-3992", "amount": "N/A", "email": "   cbpi@test.org", "company": "Umbrella", "status": "Due   ", "unit": "ML", "unnamed": ""}, {"phone": "543.323.8415", "amount": "4.346,81", "email": "   JHAFC@TEST.ORG", "company": "Hooli  ", "status": "late", "unit": " ml ", "unnamed": ""}, {"phone": "585-970-5151", "amount": "5.838,31", "email": " chnjhij@test.org", "company": "Acme Inc ", "status": "unpaid  ", "unit": " Cm ", "unnamed": ""}, {"phone": "563.220.5219", "amount": "$4,279.69", "email": "  gdhhppj@example.com", "company": " Globex ", "status": " unpaid ", "unit": "   mL", "unnamed": ""}, {"phone": "(572)663-1977", "amount": "6.348,94", "email": "mmgognm@example.com", "company": "Globex", "status": "Pending", "unit": "inches", "unnamed": ""}, {"phone": "5601008267", "amount": "9999999", "email": " GCMLG@CORP.IO ", "company": "Cyberdyne", "status": "pending", "unit": "KM", "unnamed": ""}, {"phone": "(561)024-5205", "amount": "4.844,76", "email": "  LAJNKLN@MAIL.COM", "company": "Stark Industries", "status": "overdue", "unit": "ML", "unnamed": ""}, {"phone": "554.231.5663", "amount": "2.959,91", "email": " oelpkan@test.org", "company": "Umbrella", "status": "due", "unit": "milliliter", "unnamed": ""}, {"phone": "586-017-5541", "amount": "6.023,38", "email": "EHNI@EXAMPLE.COM", "company": "  Stark Industries", "status": "Pending", "unit": "milliliter", "unnamed": ""}, {"phone": "546-775-3065", "amount": "4.450,49", "email": "   johdo@corp.io", "company": "  Acme Inc", "status": " Overdue ", "unit": "km", "unnamed": ""}, {"phone": "(568)088-1074", "amount": "727,57", "email": "aodkh@example.com", "company": "Wonka Co ", "status": "Pending", "unit": "kilometer", "unnamed": ""}, {"phone": "511.570.4086", "amount": "8.959,67", "email": "DNJL@TEST.ORG", "company": "Vehement", "status": "Pknding", "unit": "centimetesr", "unnamed": ""}, {"phone": "563.220.5219", "amount": "$4,279.69", "email": "  gdhhppj@example.com", "company": " Globex ", "status": " unpaid ", "unit": "   mL", "unnamed": ""}, {"phone": "(568)088-1074", "amount": "727,57", "email": "aodkh@example.com", "company": "Wonka Co ", "status": "Pending", "unit": "kilometer", "unnamed": ""}], "clean": [{"phone": "(536) 744-0665", "amount": 7828.66, "email": "jiniap@example.com", "company": "Acme Inc", "status": "Overdue", "unit": "in"}, {"phone": "(591) 207-5023", "amount": NaN, "email": "obbhj@test.org", "company": "Cyberdyne", "status": "Pending", "unit": "cm"}, {"phone": "(543) 917-2997", "amount": 1944.68, "email": "mdlcf@test.org", "company": "Soylent Corp", "status": "Overdue", "unit": "km"}, {"phone": "(580) 612-3992", "amount": NaN, "email": "cbpi@test.org", "company": "Umbrella", "status": "Pending", "unit": "mL"}, {"phone": "(543) 323-8415", "amount": 4346.81, "email": "jhafc@test.org", "company": "Hooli", "status": "Overdue", "unit": "mL"}, {"phone": "(585) 970-5151", "amount": 5838.31, "email": "chnjhij@test.org", "company": "Acme Inc", "status": "Pending", "unit": "cm"}, {"phone": "(563) 220-5219", "amount": 4279.69, "email": "gdhhppj@example.com", "company": "Globex", "status": "Pending", "unit": "mL"}, {"phone": "(572) 663-1977", "amount": 6348.94, "email": "mmgognm@example.com", "company": "Globex", "status": "Pending", "unit": "in"}, {"phone": "(560) 100-8267", "amount": 9999999.0, "email": "gcmlg@corp.io", "company": "Cyberdyne", "status": "Pending", "unit": "km"}, {"phone": "(561) 024-5205", "amount": 4844.76, "email": "lajnkln@mail.com", "company": "Stark Industries", "status": "Overdue", "unit": "mL"}, {"phone": "(554) 231-5663", "amount": 2959.91, "email": "oelpkan@test.org", "company": "Umbrella", "status": "Pending", "unit": "mL"}, {"phone": "(586) 017-5541", "amount": 6023.38, "email": "ehni@example.com", "company": "Stark Industries", "status": "Pending", "unit": "mL"}, {"phone": "(546) 775-3065", "amount": 4450.49, "email": "johdo@corp.io", "company": "Acme Inc", "status": "Overdue", "unit": "km"}, {"phone": "(568) 088-1074", "amount": 727.57, "email": "aodkh@example.com", "company": "Wonka Co", "status": "Pending", "unit": "km"}, {"phone": "(511) 570-4086", "amount": 8959.67, "email": "dnjl@test.org", "company": "Vehement", "status": "Pending", "unit": "cm"}], "dirty_cols": ["phone", "amount", "email", "company", "status", "unit", "unnamed"], "clean_cols": ["phone", "amount", "email", "company", "status", "unit"], "plan": {"dataset_summary": "17 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OVERDUE": "Overdue", "unpaid": "Pending", "LATE": "Overdue", "Due": "Pending", "late": "Overdue", "pending": "Pending", "overdue": "Overdue", "due": "Pending", "Pknding": "Pending"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Centimeters": "cm", "ML": "mL", "ml": "mL", "Cm": "cm", "inches": "in", "KM": "km", "milliliter": "mL", "kilometer": "km", "centimetesr": "cm"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"amount": "$5,826.21", "company": "Initech", "unit": "kilograms", "status": "Churned"}, {"amount": "2.245,64", "company": "Umbrella ", "unit": "KG", "status": "active"}, {"amount": "7.354,13", "company": " Umbrella", "unit": "kg", "status": "ACTIVE"}, {"amount": "-100", "company": "Cyberdyne", "unit": "Kg", "status": "ACTIVE"}, {"amount": "8.996,12", "company": "   Umbrella", "unit": "Lb", "status": "Canceled"}, {"amount": "2.939,31", "company": "Hooli", "unit": "pound", "status": "cancelled"}, {"amount": "$2,732.52", "company": "Soylent Corp", "unit": "Kg", "status": "ACTIVE"}, {"amount": "6.787,97", "company": "Soylent Corp", "unit": "kg", "status": "   cancelled"}, {"amount": "4.022,95", "company": "Acme Inc", "unit": "Cm", "status": "  CHURNED"}, {"amount": "3.754,04", "company": " Cyberdyne", "unit": "centimeters", "status": "atcive"}, {"amount": "$8,652.25", "company": "Acme Inc ", "unit": " Lb ", "status": "Churned"}, {"amount": "2.939,31", "company": "Hooli", "unit": "pound", "status": "cancelled"}, {"amount": "", "company": "", "unit": "", "status": ""}], "clean": [{"amount": 5826.21, "company": "Initech", "unit": "kg", "status": "Churned"}, {"amount": 2245.64, "company": "Umbrella", "unit": "kg", "status": "Active"}, {"amount": 7354.13, "company": "Umbrella", "unit": "kg", "status": "Active"}, {"amount": -100.0, "company": "Cyberdyne", "unit": "kg", "status": "Active"}, {"amount": 8996.12, "company": "Umbrella", "unit": "lb", "status": "Churned"}, {"amount": 2939.31, "company": "Hooli", "unit": "lb", "status": "Churned"}, {"amount": 2732.52, "company": "Soylent Corp", "unit": "kg", "status": "Active"}, {"amount": 6787.97, "company": "Soylent Corp", "unit": "kg", "status": "Churned"}, {"amount": 4022.95, "company": "Acme Inc", "unit": "cm", "status": "Churned"}, {"amount": 3754.04, "company": "Cyberdyne", "unit": "cm", "status": "Active"}, {"amount": 8652.25, "company": "Acme Inc", "unit": "lb", "status": "Churned"}], "dirty_cols": ["amount", "company", "unit", "status"], "clean_cols": ["amount", "company", "unit", "status"], "plan": {"dataset_summary": "13 rows × 4 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilograms": "kg", "KG": "kg", "Kg": "kg", "Lb": "lb", "pound": "lb", "Cm": "cm", "centimeters": "cm"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"active": "Active", "ACTIVE": "Active", "Canceled": "Churned", "cancelled": "Churned", "CHURNED": "Churned", "atcive": "Active"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"phone": "514-747-3304", "signup_date": "18 Nov 2023", "country": "BF"}, {"phone": "594-004-4443", "signup_date": "8/10/2023", "country": "British Indian Ocean Territory"}, {"phone": "598.692.6447", "signup_date": "45130", "country": " io"}, {"phone": "5555485727", "signup_date": "2 Feb 2023", "country": " Cambodia "}, {"phone": "553.910.1102", "signup_date": "8 Nov 2023", "country": "So"}, {"phone": "502.431.2838", "signup_date": "1/25/2023", "country": "   Bfa"}, {"phone": "5105330613", "signup_date": "7/3/2023", "country": "  Federal Republic of Somalia"}, {"phone": "(578)869-0378", "signup_date": "11/10/2023", "country": "SO"}, {"phone": "5535709100", "signup_date": "5 Nov 2023", "country": "kingdom off cambodia"}, {"phone": "(586)733-5264", "signup_date": "2023-07-08", "country": "British Indian Ocean Territory"}, {"phone": "547-437-3213", "signup_date": "10 Dec 2023", "country": "Kingdom of Cambodia"}, {"phone": "5877113935", "signup_date": "2023-12-24", "country": "IO"}, {"phone": "510.871.2666", "signup_date": "45002", "country": " bf"}, {"phone": "566.907.4549", "signup_date": "45074", "country": "  io"}, {"phone": "(557)395-4757", "signup_date": "1/17/2023", "country": "burkina faso"}, {"phone": "541-941-0435", "signup_date": "2023-01-05", "country": "BFA"}, {"phone": "(589)082-0106", "signup_date": "7 Apr 2023", "country": "som"}, {"phone": "5008848165", "signup_date": "4/7/2023", "country": "Somalia"}, {"phone": "(504)000-2660", "signup_date": "14 Feb 2023", "country": "KHM"}, {"phone": "566.907.4549", "signup_date": "45074", "country": "  io"}], "clean": [{"phone": "(514) 747-3304", "signup_date": "2023-11-18", "country": "Burkina Faso"}, {"phone": "(594) 004-4443", "signup_date": "2023-08-10", "country": "British Indian Ocean Territory"}, {"phone": "(598) 692-6447", "signup_date": "2023-07-23", "country": "British Indian Ocean Territory"}, {"phone": "(555) 548-5727", "signup_date": "2023-02-02", "country": "Cambodia"}, {"phone": "(553) 910-1102", "signup_date": "2023-11-08", "country": "Somalia"}, {"phone": "(502) 431-2838", "signup_date": "2023-01-25", "country": "Burkina Faso"}, {"phone": "(510) 533-0613", "signup_date": "2023-07-03", "country": "Somalia"}, {"phone": "(578) 869-0378", "signup_date": "2023-11-10", "country": "Somalia"}, {"phone": "(553) 570-9100", "signup_date": "2023-11-05", "country": "Cambodia"}, {"phone": "(586) 733-5264", "signup_date": "2023-07-08", "country": "British Indian Ocean Territory"}, {"phone": "(547) 437-3213", "signup_date": "2023-12-10", "country": "Cambodia"}, {"phone": "(587) 711-3935", "signup_date": "2023-12-24", "country": "British Indian Ocean Territory"}, {"phone": "(510) 871-2666", "signup_date": "2023-03-17", "country": "Burkina Faso"}, {"phone": "(566) 907-4549", "signup_date": "2023-05-28", "country": "British Indian Ocean Territory"}, {"phone": "(557) 395-4757", "signup_date": "2023-01-17", "country": "Burkina Faso"}, {"phone": "(541) 941-0435", "signup_date": "2023-01-05", "country": "Burkina Faso"}, {"phone": "(589) 082-0106", "signup_date": "2023-04-07", "country": "Somalia"}, {"phone": "(500) 884-8165", "signup_date": "2023-04-07", "country": "Somalia"}, {"phone": "(504) 000-2660", "signup_date": "2023-02-14", "country": "Cambodia"}], "dirty_cols": ["phone", "signup_date", "country"], "clean_cols": ["phone", "signup_date", "country"], "plan": {"dataset_summary": "20 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BF": "Burkina Faso", "io": "British Indian Ocean Territory", "So": "Somalia", "Bfa": "Burkina Faso", "Federal Republic of Somalia": "Somalia", "SO": "Somalia", "kingdom off cambodia": "Cambodia", "Kingdom of Cambodia": "Cambodia", "IO": "British Indian Ocean Territory", "bf": "Burkina Faso", "burkina faso": "Burkina Faso", "BFA": "Burkina Faso", "som": "Somalia", "KHM": "Cambodia"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "517-170-2270", "status": "new", "state": "  Maine", "currency": "LARI", "company": "Wonka Co  ", "department": "biz dev", "email": "  OGGBABK@TEST.ORG", "unit": "kg", "unnamed": ""}, {"phone": "5539146015", "status": "WIP", "state": "IOWA", "currency": "wir franc", "company": "Acme Inc", "department": "Operations", "email": " jlfdpg@corp.io ", "unit": " kilogram ", "unnamed": ""}, {"phone": "501.478.6332", "status": "Closed Won", "state": " Iowa ", "currency": "GEL", "company": "   Globex", "department": "ops ", "email": " gieepgi@mail.com", "unit": "KG", "unnamed": ""}, {"phone": "(587)926-8805", "status": "Won", "state": "IA", "currency": "WIR Euro", "company": " Soylent Corp", "department": "  Prod", "email": "GGAEC@CORP.IO", "unit": "KG", "unnamed": ""}, {"phone": "(534)358-0379", "status": "Lost", "state": "Ia", "currency": "chw", "company": "Stark  Industries", "department": "OPS", "email": "poob@corp.io", "unit": "pounds", "unnamed": ""}, {"phone": "(595)567-7346", "status": " Wip ", "state": "ME", "currency": " CHE", "company": "   Acme Inc", "department": "Ops", "email": "FIBH@MAIL.COM", "unit": "  Lb", "unnamed": ""}, {"phone": "575.623.7126", "status": "lost", "state": " Iowa ", "currency": "GEL", "company": "Hooli", "department": "OPS", "email": "djnnih@mail.com", "unit": "   Lb", "unnamed": ""}, {"phone": "532-834-0427", "status": "won", "state": "Maine ", "currency": "gel", "company": " Umbrella", "department": "Prod", "email": "fkkcmj@corp.io", "unit": " kg ", "unnamed": ""}, {"phone": "(546)580-6977", "status": "WON", "state": "IA", "currency": "WIR Franc  ", "company": "Stark Industries", "department": "Ops", "email": "DGKKN@EXAMPLE.COM", "unit": "Lb  ", "unnamed": ""}, {"phone": "5222448001", "status": "  WON", "state": "IA", "currency": "CHE", "company": "Umbrella ", "department": "Biz Dev", "email": "BOKGLGF@TEST.ORG", "unit": "lb", "unnamed": ""}, {"phone": "572.087.5452", "status": "Closed Wno", "state": "Maine", "currency": "WIR Euro", "company": "Cyberdyne   ", "department": "Biz  Dev", "email": "dnlefo@corp.io", "unit": "KG", "unnamed": ""}, {"phone": "579-104-4148", "status": "lost", "state": "Maine", "currency": "gel", "company": "  Globex", "department": "customer support", "email": "MFMMGLF@CORP.IO", "unit": "lb", "unnamed": ""}, {"phone": "5046277439", "status": "WON", "state": "Maine", "currency": "wir euro", "company": "Hooli  ", "department": "PRODUCT", "email": "NECB@CORP.IO", "unit": "kilogram   ", "unnamed": ""}, {"phone": "513.364.2707", "status": "closed olst", "state": "IA", "currency": " CHW ", "company": "  Stark Industries", "department": "sales", "email": "mmhgj@test.org   ", "unit": " lbs", "unnamed": ""}, {"phone": "5190059042", "status": "in progress", "state": "Maine", "currency": " WIR Franc", "company": "Vehement ", "department": "Proyuct", "email": " ccjdecc@test.org ", "unit": "LBS", "unnamed": ""}, {"phone": "571-439-7783", "status": "  won", "state": "Maine", "currency": "CHE", "company": "Umbrella", "department": "PRODUCT", "email": "  ndldn@mail.com", "unit": "KILOGRAMS", "unnamed": ""}, {"phone": "5565162857", "status": "won", "state": " Mane", "currency": "WIR Euo", "company": "Acme Inc", "department": "pm", "email": "pgbncp@corp.io", "unit": "krlogram", "unnamed": ""}, {"phone": "5465497569", "status": "new", "state": "ia", "currency": "Wir Franc", "company": "Stark Industries", "department": "Sales", "email": " indfdb@test.org", "unit": "Lb", "unnamed": ""}, {"phone": "512-355-7955", "status": " onoging ", "state": "ME", "currency": "laari", "company": "  Umbrella", "department": " Sales ", "email": "kbanm@mail.com", "unit": "pound", "unnamed": ""}, {"phone": "(561)644-7093", "status": "closed won   ", "state": " ME", "currency": "wir franc", "company": "Acme Inc", "department": "support", "email": "   gjagljc@corp.io", "unit": "pound", "unnamed": ""}, {"phone": "(505)947-6859", "status": "closed-won", "state": "ME", "currency": "CHE", "company": "Wonka  Co", "department": "support", "email": "IOFND@CORP.IO", "unit": "kilogram  ", "unnamed": ""}, {"phone": "5371434866", "status": "NEW", "state": "IA", "currency": "che", "company": "Stark  Industries", "department": "biz dev", "email": "PIOKKCL@CORP.IO", "unit": "kg", "unnamed": ""}, {"phone": "571-912-4391", "status": "lst", "state": " ME", "currency": "gel", "company": "Umbrella", "department": "   Sales", "email": "didfoop@mail.com", "unit": "kg", "unnamed": ""}, {"phone": "548-399-6659", "status": "Lust", "state": "Iowa", "currency": "Lari", "company": "Initech", "department": "OPS", "email": "HCDNDHG@EXAMPLE.COM", "unit": "Kg", "unnamed": ""}, {"phone": "(552)039-6202", "status": "Closed Won", "state": "IA", "currency": "Lari", "company": "Vehement", "department": "Opesations", "email": "OGOJAN@CORP.IO", "unit": "  kg", "unnamed": ""}, {"phone": "584.951.1571", "status": "Lost", "state": "ME", "currency": "WIR Franc", "company": "Initech", "department": "Prod", "email": "fgpj@test.org", "unit": "Lb", "unnamed": ""}, {"phone": "5015111620", "status": "New", "state": "   ia", "currency": "GEL", "company": "Acme  Inc", "department": "Cust Spport  ", "email": "  JJPKMIF@TEST.ORG", "unit": "Lb", "unnamed": ""}, {"phone": "544-246-3461", "status": "LOST", "state": "maine", "currency": "GEL", "company": " Globex ", "department": "Operations", "email": "lhcglbm@example.com  ", "unit": "kg", "unnamed": ""}, {"phone": "5513216009", "status": "open", "state": "me", "currency": "CHW", "company": "Initech", "department": "Operations", "email": "GFFPNN@EXAMPLE.COM", "unit": "pound", "unnamed": ""}, {"phone": "567.786.3570", "status": "in  progress", "state": "  IA", "currency": "WIR FRANC", "company": "Vehement", "department": "CUST SUPPORT", "email": "MBBG@EXAMPLE.COM", "unit": "LBS", "unnamed": ""}, {"phone": "513.930.2847", "status": "WON", "state": "ia", "currency": "Lari", "company": "  Acme Inc", "department": "biz dev", "email": "pepe@example.com ", "unit": "pound", "unnamed": ""}, {"phone": "(519)776-2675", "status": "lost", "state": "Maine", "currency": "   CHE", "company": "Vehement", "department": "Sales", "email": "mnah@test.org", "unit": "kiloggram", "unnamed": ""}, {"phone": "(563)557-7428", "status": "Closed Won", "state": "IA", "currency": "wir euro", "company": "Wonka Co  ", "department": "OPERATIONS ", "email": "ljboc@mail.com", "unit": "punds", "unnamed": ""}, {"phone": "(516)601-2033", "status": " Won ", "state": "maine", "currency": "chw", "company": " Cyberdyne ", "department": "prod", "email": "   ECBAA@EXAMPLE.COM", "unit": "lbs  ", "unnamed": ""}, {"phone": "505-592-9314", "status": " won ", "state": "ME", "currency": "WIR rFanc ", "company": "   Umbrella", "department": "operations", "email": "NEFB@CORP.IO", "unit": "kiilogram", "unnamed": ""}, {"phone": "(542)774-6775", "status": " ongoing", "state": "Iowa", "currency": "WJR Euro", "company": "Wonka Co", "department": "Biz Dev", "email": "afomopo@mail.com", "unit": "kilograms", "unnamed": ""}, {"phone": "519.192.8303", "status": "new", "state": "Maine", "currency": "WIR FRANC", "company": "Acme Inc", "department": "prod", "email": "LBJFM@MAIL.COM", "unit": " kg ", "unnamed": ""}, {"phone": "(567)480-4774", "status": "Closed Lost", "state": "ia", "currency": "CHW", "company": "Soylent Corp", "department": "Customer Support", "email": "  DBGIC@MAIL.COM", "unit": "  lb", "unnamed": ""}, {"phone": "5662997697", "status": "In Progress", "state": "me", "currency": "WIR Euro", "company": "Hooli", "department": "product", "email": "JMLH@CORP.IO", "unit": "klograms", "unnamed": ""}, {"phone": "541.779.6461", "status": "CLOESD LOST", "state": "Iowa", "currency": "gel", "company": "   Initech", "department": "Prod", "email": " DPIAEN@MAIL.COM", "unit": "POUND", "unnamed": ""}, {"phone": "510-465-9147", "status": "LOST", "state": "iowa ", "currency": "Lari", "company": "Wonka Co", "department": "Product", "email": "ekadmi@example.com", "unit": " pound ", "unnamed": ""}, {"phone": "5337268916", "status": " in-progress", "state": "ME", "currency": "LARI  ", "company": "Umbrella", "department": "SALAS", "email": " pofpb@test.org", "unit": "kilogram", "unnamed": ""}, {"phone": "566.273.5458", "status": "in progress", "state": "  Iowa", "currency": "wir euro", "company": "  Umbrella", "department": "pm", "email": "ECOFE@MAIL.COM", "unit": "pound", "unnamed": ""}, {"phone": "589.022.2842", "status": "in-progress", "state": "Iowa  ", "currency": " lari", "company": "   Globex", "department": "Sales", "email": "eabohib@example.com", "unit": "lb", "unnamed": ""}, {"phone": "5095968063", "status": "open", "state": "  IA", "currency": "CHW  ", "company": "   Hooli", "department": "Customer Support", "email": "eajnha@mail.com  ", "unit": "Poound", "unnamed": ""}, {"phone": "550.840.9837", "status": "  Closed Lost", "state": "ME", "currency": "Lari", "company": "   Acme Inc", "department": "PM", "email": "BDILIOC@MAIL.COM", "unit": "kilogram", "unnamed": ""}, {"phone": "5294390721", "status": "  open", "state": "Maine  ", "currency": "LAIR", "company": "Vehement", "department": "pm", "email": " GEEFJAJ@CORP.IO", "unit": "KG ", "unnamed": ""}, {"phone": "(535)361-5282", "status": "ongonig", "state": "Iowa", "currency": "che   ", "company": "Umbrella", "department": "Customer Support", "email": "aibilb@test.org", "unit": " lbs", "unnamed": ""}, {"phone": "526-122-1701", "status": "in progress", "state": "Maine  ", "currency": "wir franc", "company": "Acme  Inc", "department": "prroduct", "email": "KDJHI@MAIL.COM", "unit": "lbs", "unnamed": ""}, {"phone": "525-409-0223", "status": "  closed-won", "state": "ia", "currency": "chw", "company": "Wonka Co", "department": "   Product", "email": "bdnp@test.org", "unit": "Lb", "unnamed": ""}, {"phone": "590.998.0887", "status": "Won", "state": "Iowa", "currency": "WIR Franc", "company": " Umbrella", "department": "Cust Support", "email": "PJBKJG@EXAMPLE.COM", "unit": "pounds", "unnamed": ""}, {"phone": "582-376-3600", "status": "IN-PROGRESS", "state": " Iowa", "currency": "CHW", "company": " Globex ", "department": "BIZ DEV", "email": "KHAKH@MAIL.COM ", "unit": " kilograms", "unnamed": ""}, {"phone": "5329022133", "status": "ongoing", "state": "Maine", "currency": "WIR  EURO", "company": "  Soylent Corp", "department": "Ops", "email": "EFJFIAG@EXAMPLE.COM", "unit": "POUNDS  ", "unnamed": ""}, {"phone": "5630333071", "status": "lost", "state": "Maine", "currency": "che", "company": "  Initech", "department": "Product", "email": "PCNF@TEST.ORG   ", "unit": "LB", "unnamed": ""}, {"phone": "569-564-5784", "status": "in rpogress", "state": "IA", "currency": "Wir Euro", "company": "Vehement   ", "department": "pm", "email": "PAKB@MAIL.COM", "unit": "kg", "unnamed": ""}, {"phone": "5671206768", "status": "Ongoing", "state": "IA", "currency": "WIR Euro", "company": "Globex  ", "department": "Prod", "email": " iobe@test.org ", "unit": " kg ", "unnamed": ""}, {"phone": "(552)039-6202", "status": "Closed Won", "state": "IA", "currency": "Lari", "company": "Vehement", "department": "Opesations", "email": "OGOJAN@CORP.IO", "unit": "  kg", "unnamed": ""}], "clean": [{"phone": "(517) 170-2270", "status": "Open", "state": "Maine", "currency": "GEL", "company": "Wonka Co", "department": "Sales", "email": "oggbabk@test.org", "unit": "kg"}, {"phone": "(553) 914-6015", "status": "In Progress", "state": "Iowa", "currency": "CHW", "company": "Acme Inc", "department": "Operations", "email": "jlfdpg@corp.io", "unit": "kg"}, {"phone": "(501) 478-6332", "status": "Won", "state": "Iowa", "currency": "GEL", "company": "Globex", "department": "Operations", "email": "gieepgi@mail.com", "unit": "kg"}, {"phone": "(587) 926-8805", "status": "Won", "state": "Iowa", "currency": "CHE", "company": "Soylent Corp", "department": "Product", "email": "ggaec@corp.io", "unit": "kg"}, {"phone": "(534) 358-0379", "status": "Lost", "state": "Iowa", "currency": "CHW", "company": "Stark Industries", "department": "Operations", "email": "poob@corp.io", "unit": "lb"}, {"phone": "(595) 567-7346", "status": "In Progress", "state": "Maine", "currency": "CHE", "company": "Acme Inc", "department": "Operations", "email": "fibh@mail.com", "unit": "lb"}, {"phone": "(575) 623-7126", "status": "Lost", "state": "Iowa", "currency": "GEL", "company": "Hooli", "department": "Operations", "email": "djnnih@mail.com", "unit": "lb"}, {"phone": "(532) 834-0427", "status": "Won", "state": "Maine", "currency": "GEL", "company": "Umbrella", "department": "Product", "email": "fkkcmj@corp.io", "unit": "kg"}, {"phone": "(546) 580-6977", "status": "Won", "state": "Iowa", "currency": "CHW", "company": "Stark Industries", "department": "Operations", "email": "dgkkn@example.com", "unit": "lb"}, {"phone": "(522) 244-8001", "status": "Won", "state": "Iowa", "currency": "CHE", "company": "Umbrella", "department": "Sales", "email": "bokglgf@test.org", "unit": "lb"}, {"phone": "(572) 087-5452", "status": "Won", "state": "Maine", "currency": "CHE", "company": "Cyberdyne", "department": "Sales", "email": "dnlefo@corp.io", "unit": "kg"}, {"phone": "(579) 104-4148", "status": "Lost", "state": "Maine", "currency": "GEL", "company": "Globex", "department": "Customer Support", "email": "mfmmglf@corp.io", "unit": "lb"}, {"phone": "(504) 627-7439", "status": "Won", "state": "Maine", "currency": "CHE", "company": "Hooli", "department": "Product", "email": "necb@corp.io", "unit": "kg"}, {"phone": "(513) 364-2707", "status": "Lost", "state": "Iowa", "currency": "CHW", "company": "Stark Industries", "department": "Sales", "email": "mmhgj@test.org", "unit": "lb"}, {"phone": "(519) 005-9042", "status": "In Progress", "state": "Maine", "currency": "CHW", "company": "Vehement", "department": "Product", "email": "ccjdecc@test.org", "unit": "lb"}, {"phone": "(571) 439-7783", "status": "Won", "state": "Maine", "currency": "CHE", "company": "Umbrella", "department": "Product", "email": "ndldn@mail.com", "unit": "kg"}, {"phone": "(556) 516-2857", "status": "Won", "state": "Maine", "currency": "CHE", "company": "Acme Inc", "department": "Product", "email": "pgbncp@corp.io", "unit": "kg"}, {"phone": "(546) 549-7569", "status": "Open", "state": "Iowa", "currency": "CHW", "company": "Stark Industries", "department": "Sales", "email": "indfdb@test.org", "unit": "lb"}, {"phone": "(512) 355-7955", "status": "In Progress", "state": "Maine", "currency": "GEL", "company": "Umbrella", "department": "Sales", "email": "kbanm@mail.com", "unit": "lb"}, {"phone": "(561) 644-7093", "status": "Won", "state": "Maine", "currency": "CHW", "company": "Acme Inc", "department": "Customer Support", "email": "gjagljc@corp.io", "unit": "lb"}, {"phone": "(505) 947-6859", "status": "Won", "state": "Maine", "currency": "CHE", "company": "Wonka Co", "department": "Customer Support", "email": "iofnd@corp.io", "unit": "kg"}, {"phone": "(537) 143-4866", "status": "Open", "state": "Iowa", "currency": "CHE", "company": "Stark Industries", "department": "Sales", "email": "piokkcl@corp.io", "unit": "kg"}, {"phone": "(571) 912-4391", "status": "Lost", "state": "Maine", "currency": "GEL", "company": "Umbrella", "department": "Sales", "email": "didfoop@mail.com", "unit": "kg"}, {"phone": "(548) 399-6659", "status": "Lost", "state": "Iowa", "currency": "GEL", "company": "Initech", "department": "Operations", "email": "hcdndhg@example.com", "unit": "kg"}, {"phone": "(552) 039-6202", "status": "Won", "state": "Iowa", "currency": "GEL", "company": "Vehement", "department": "Operations", "email": "ogojan@corp.io", "unit": "kg"}, {"phone": "(584) 951-1571", "status": "Lost", "state": "Maine", "currency": "CHW", "company": "Initech", "department": "Product", "email": "fgpj@test.org", "unit": "lb"}, {"phone": "(501) 511-1620", "status": "Open", "state": "Iowa", "currency": "GEL", "company": "Acme Inc", "department": "Customer Support", "email": "jjpkmif@test.org", "unit": "lb"}, {"phone": "(544) 246-3461", "status": "Lost", "state": "Maine", "currency": "GEL", "company": "Globex", "department": "Operations", "email": "lhcglbm@example.com", "unit": "kg"}, {"phone": "(551) 321-6009", "status": "Open", "state": "Maine", "currency": "CHW", "company": "Initech", "department": "Operations", "email": "gffpnn@example.com", "unit": "lb"}, {"phone": "(567) 786-3570", "status": "In Progress", "state": "Iowa", "currency": "CHW", "company": "Vehement", "department": "Customer Support", "email": "mbbg@example.com", "unit": "lb"}, {"phone": "(513) 930-2847", "status": "Won", "state": "Iowa", "currency": "GEL", "company": "Acme Inc", "department": "Sales", "email": "pepe@example.com", "unit": "lb"}, {"phone": "(519) 776-2675", "status": "Lost", "state": "Maine", "currency": "CHE", "company": "Vehement", "department": "Sales", "email": "mnah@test.org", "unit": "kg"}, {"phone": "(563) 557-7428", "status": "Won", "state": "Iowa", "currency": "CHE", "company": "Wonka Co", "department": "Operations", "email": "ljboc@mail.com", "unit": "lb"}, {"phone": "(516) 601-2033", "status": "Won", "state": "Maine", "currency": "CHW", "company": "Cyberdyne", "department": "Product", "email": "ecbaa@example.com", "unit": "lb"}, {"phone": "(505) 592-9314", "status": "Won", "state": "Maine", "currency": "CHW", "company": "Umbrella", "department": "Operations", "email": "nefb@corp.io", "unit": "kg"}, {"phone": "(542) 774-6775", "status": "In Progress", "state": "Iowa", "currency": "CHE", "company": "Wonka Co", "department": "Sales", "email": "afomopo@mail.com", "unit": "kg"}, {"phone": "(519) 192-8303", "status": "Open", "state": "Maine", "currency": "CHW", "company": "Acme Inc", "department": "Product", "email": "lbjfm@mail.com", "unit": "kg"}, {"phone": "(567) 480-4774", "status": "Lost", "state": "Iowa", "currency": "CHW", "company": "Soylent Corp", "department": "Customer Support", "email": "dbgic@mail.com", "unit": "lb"}, {"phone": "(566) 299-7697", "status": "In Progress", "state": "Maine", "currency": "CHE", "company": "Hooli", "department": "Product", "email": "jmlh@corp.io", "unit": "kg"}, {"phone": "(541) 779-6461", "status": "Lost", "state": "Iowa", "currency": "GEL", "company": "Initech", "department": "Product", "email": "dpiaen@mail.com", "unit": "lb"}, {"phone": "(510) 465-9147", "status": "Lost", "state": "Iowa", "currency": "GEL", "company": "Wonka Co", "department": "Product", "email": "ekadmi@example.com", "unit": "lb"}, {"phone": "(533) 726-8916", "status": "In Progress", "state": "Maine", "currency": "GEL", "company": "Umbrella", "department": "Sales", "email": "pofpb@test.org", "unit": "kg"}, {"phone": "(566) 273-5458", "status": "In Progress", "state": "Iowa", "currency": "CHE", "company": "Umbrella", "department": "Product", "email": "ecofe@mail.com", "unit": "lb"}, {"phone": "(589) 022-2842", "status": "In Progress", "state": "Iowa", "currency": "GEL", "company": "Globex", "department": "Sales", "email": "eabohib@example.com", "unit": "lb"}, {"phone": "(509) 596-8063", "status": "Open", "state": "Iowa", "currency": "CHW", "company": "Hooli", "department": "Customer Support", "email": "eajnha@mail.com", "unit": "lb"}, {"phone": "(550) 840-9837", "status": "Lost", "state": "Maine", "currency": "GEL", "company": "Acme Inc", "department": "Product", "email": "bdilioc@mail.com", "unit": "kg"}, {"phone": "(529) 439-0721", "status": "Open", "state": "Maine", "currency": "GEL", "company": "Vehement", "department": "Product", "email": "geefjaj@corp.io", "unit": "kg"}, {"phone": "(535) 361-5282", "status": "In Progress", "state": "Iowa", "currency": "CHE", "company": "Umbrella", "department": "Customer Support", "email": "aibilb@test.org", "unit": "lb"}, {"phone": "(526) 122-1701", "status": "In Progress", "state": "Maine", "currency": "CHW", "company": "Acme Inc", "department": "Product", "email": "kdjhi@mail.com", "unit": "lb"}, {"phone": "(525) 409-0223", "status": "Won", "state": "Iowa", "currency": "CHW", "company": "Wonka Co", "department": "Product", "email": "bdnp@test.org", "unit": "lb"}, {"phone": "(590) 998-0887", "status": "Won", "state": "Iowa", "currency": "CHW", "company": "Umbrella", "department": "Customer Support", "email": "pjbkjg@example.com", "unit": "lb"}, {"phone": "(582) 376-3600", "status": "In Progress", "state": "Iowa", "currency": "CHW", "company": "Globex", "department": "Sales", "email": "khakh@mail.com", "unit": "kg"}, {"phone": "(532) 902-2133", "status": "In Progress", "state": "Maine", "currency": "CHE", "company": "Soylent Corp", "department": "Operations", "email": "efjfiag@example.com", "unit": "lb"}, {"phone": "(563) 033-3071", "status": "Lost", "state": "Maine", "currency": "CHE", "company": "Initech", "department": "Product", "email": "pcnf@test.org", "unit": "lb"}, {"phone": "(569) 564-5784", "status": "In Progress", "state": "Iowa", "currency": "CHE", "company": "Vehement", "department": "Product", "email": "pakb@mail.com", "unit": "kg"}, {"phone": "(567) 120-6768", "status": "In Progress", "state": "Iowa", "currency": "CHE", "company": "Globex", "department": "Product", "email": "iobe@test.org", "unit": "kg"}], "dirty_cols": ["phone", "status", "state", "currency", "company", "department", "email", "unit", "unnamed"], "clean_cols": ["phone", "status", "state", "currency", "company", "department", "email", "unit"], "plan": {"dataset_summary": "57 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"new": "Open", "WIP": "In Progress", "Closed Won": "Won", "Wip": "In Progress", "lost": "Lost", "won": "Won", "WON": "Won", "Closed Wno": "Won", "closed olst": "Lost", "in progress": "In Progress", "onoging": "In Progress", "closed won": "Won", "closed-won": "Won", "NEW": "Open", "lst": "Lost", "Lust": "Lost", "New": "Open", "LOST": "Lost", "open": "Open", "ongoing": "In Progress", "Closed Lost": "Lost", "CLOESD LOST": "Lost", "in-progress": "In Progress", "ongonig": "In Progress", "IN-PROGRESS": "In Progress", "in rpogress": "In Progress", "Ongoing": "In Progress"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IOWA": "Iowa", "IA": "Iowa", "Ia": "Iowa", "ME": "Maine", "Mane": "Maine", "ia": "Iowa", "maine": "Maine", "me": "Maine", "iowa": "Iowa"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LARI": "GEL", "wir franc": "CHW", "WIR Euro": "CHE", "chw": "CHW", "gel": "GEL", "WIR Franc": "CHW", "wir euro": "CHE", "WIR Euo": "CHE", "Wir Franc": "CHW", "laari": "GEL", "che": "CHE", "Lari": "GEL", "WIR FRANC": "CHW", "WIR rFanc": "CHW", "WJR Euro": "CHE", "lari": "GEL", "LAIR": "GEL", "WIR EURO": "CHE", "Wir Euro": "CHE"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"biz dev": "Sales", "ops": "Operations", "Prod": "Product", "OPS": "Operations", "Ops": "Operations", "Biz Dev": "Sales", "customer support": "Customer Support", "PRODUCT": "Product", "sales": "Sales", "Proyuct": "Product", "pm": "Product", "support": "Customer Support", "Opesations": "Operations", "Cust Spport": "Customer Support", "CUST SUPPORT": "Customer Support", "OPERATIONS": "Operations", "prod": "Product", "operations": "Operations", "product": "Product", "SALAS": "Sales", "PM": "Product", "prroduct": "Product", "Cust Support": "Customer Support", "BIZ DEV": "Sales"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilogram": "kg", "KG": "kg", "pounds": "lb", "Lb": "lb", "lbs": "lb", "LBS": "lb", "KILOGRAMS": "kg", "krlogram": "kg", "pound": "lb", "Kg": "kg", "kiloggram": "kg", "punds": "lb", "kiilogram": "kg", "kilograms": "kg", "klograms": "kg", "POUND": "lb", "Poound": "lb", "POUNDS": "lb", "LB": "lb"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"unit": "   ML", "job_title": " V.P.", "amount": "--", "company": "Umbrella", "name": "Judy Adams", "city": " Mugina ", "is_active": "0", "notes2": ""}, {"unit": "milliliter", "job_title": "Vice President", "amount": "6.024,96", "company": "Soylent  Corp", "name": "Omar Fischer  ", "city": "San Fran", "is_active": "1", "notes2": ""}, {"unit": "Cetnimeters", "job_title": "V.P.  ", "amount": "$3,192.34", "company": "Umbrella", "name": "Frank Cruz  ", "city": "Phoenix", "is_active": "T", "notes2": ""}, {"unit": "cm", "job_title": "vice president", "amount": "$6,852.42", "company": "Soylent Corp", "name": "David Novak", "city": "Phoenix", "is_active": "false", "notes2": ""}, {"unit": "centimeter", "job_title": "prod  manager", "amount": "8.851,32", "company": "Vehement", "name": "Priya Khan  ", "city": "Sf", "is_active": "T", "notes2": ""}, {"unit": "centimeter", "job_title": "  Vice President", "amount": "$8,504.54", "company": "  Wonka Co", "name": "Carol Petrov", "city": "Mugina", "is_active": "1", "notes2": ""}, {"unit": "MILLILITER", "job_title": "Prod Manager", "amount": "334,49", "company": "Vehement", "name": "Frank Park", "city": "PHOMNIX", "is_active": "No", "notes2": ""}, {"unit": "milliliter", "job_title": "Product Manager", "amount": "3.760,35", "company": "Stark Industries  ", "name": "Eve  Petrov", "city": "mugina", "is_active": "false", "notes2": ""}, {"unit": " centimeters ", "job_title": "Product Manager", "amount": "5.118,98", "company": "Cyberdyne", "name": "Priya Brandt", "city": "San  Francisco", "is_active": "0", "notes2": ""}, {"unit": "milliliter", "job_title": "Product Manager", "amount": "$8,990.21", "company": "Wonka  Co", "name": "Omar Smith", "city": "Mugina", "is_active": "F", "notes2": ""}, {"unit": "centimeters", "job_title": "Vice Pres", "amount": "8.633,98", "company": "Wonka  Co", "name": "Frank Adams", "city": "sf", "is_active": "Yes", "notes2": ""}, {"unit": "mls", "job_title": "Prod Mgr", "amount": "N/A", "company": "Initech", "name": "Lena Khan ", "city": "san fran", "is_active": "1", "notes2": ""}, {"unit": "mls", "job_title": " V..P. ", "amount": "5.725,50", "company": "Hooli", "name": "Frank Johnson", "city": "  San Francisco", "is_active": "TRUE", "notes2": ""}, {"unit": "  CM", "job_title": "VP", "amount": "$3,266.99", "company": "Globex", "name": "Bob Smith ", "city": "Mugina", "is_active": "false", "notes2": ""}, {"unit": "ml", "job_title": "PM", "amount": "$5,576.66", "company": "  Stark Industries", "name": "David Johnson", "city": "mugina", "is_active": "true", "notes2": ""}, {"unit": " Cm ", "job_title": "Prod Manager", "amount": "1.695,69", "company": " Cyberdyne ", "name": "Karl Lee", "city": "Mugipa", "is_active": "true", "notes2": ""}, {"unit": " milliliter ", "job_title": "   VICE PRESIDENT", "amount": "1.351,17", "company": "Umbrella", "name": "Carol Fischer", "city": " phoenix ", "is_active": "true", "notes2": ""}, {"unit": " Cm", "job_title": "v.p.", "amount": "$7,065.19", "company": "Hooli ", "name": "Heidi Novak", "city": "PHOENIX", "is_active": "FALSE", "notes2": ""}, {"unit": "mls", "job_title": "Prod Manager", "amount": "4.427,41", "company": " Wonka Co", "name": "Karl Diaz", "city": "san fran", "is_active": "true", "notes2": ""}, {"unit": "Cm", "job_title": "prod manager", "amount": "$8,411.55", "company": "Hooli", "name": "Carol  Smith", "city": "Phoenix", "is_active": "true", "notes2": ""}, {"unit": "Ml", "job_title": " Vice President", "amount": "#N/A", "company": "   Soylent Corp", "name": "Bob Johnson", "city": " MUGINA", "is_active": "N", "notes2": ""}, {"unit": "  centimeters", "job_title": "Prod Manager", "amount": "$1,626.85", "company": "Stark Industries", "name": "Alice Wong", "city": " San Fran", "is_active": "1", "notes2": ""}, {"unit": "CENTIMETER", "job_title": "PM", "amount": "1.656,80", "company": "Cyberdyne", "name": "   Ivan Novak", "city": "   phoenix", "is_active": "Y", "notes2": ""}, {"unit": "MLS", "job_title": "VP ", "amount": "$3,360.81", "company": "Hooli", "name": "Ivan Fischer", "city": "phhenix", "is_active": "No", "notes2": ""}, {"unit": "Milliliter", "job_title": "PM", "amount": "$3,472.24", "company": "Cyberdyne", "name": "  Heidi Khan", "city": "SF", "is_active": "true", "notes2": ""}, {"unit": "centimeters", "job_title": "PM", "amount": "$7,657.08", "company": "Hooli", "name": "Sara Johnson", "city": "SAN FRAN", "is_active": "FALSE", "notes2": ""}, {"unit": "ML", "job_title": "PROD MGR", "amount": "7.116,97", "company": " Umbrella ", "name": "Priya Fischer", "city": "  San Francisco", "is_active": "false", "notes2": ""}, {"unit": "ml", "job_title": "vp", "amount": "$7,954.51", "company": "Soylent  Corp", "name": "Heidi Khan", "city": "San Francisco", "is_active": "true", "notes2": ""}, {"unit": "Cm", "job_title": "Prod Mgr", "amount": "TBD", "company": " Hooli ", "name": "Heidi Johnson", "city": "Mugina ", "is_active": "T", "notes2": ""}, {"unit": "ml", "job_title": "vp", "amount": "4.995,46", "company": "Cyberdyne", "name": "Heidi Wong", "city": " SF ", "is_active": "false", "notes2": ""}, {"unit": "mls", "job_title": "Prod  Manager", "amount": "5.916,47", "company": " Globex ", "name": "Alice Fischer", "city": "MUGINA", "is_active": "FALSE", "notes2": ""}, {"unit": "millilitter", "job_title": "Vice President", "amount": "N/A", "company": "Umbrella", "name": "Sara Brandt", "city": " San Francisco", "is_active": "1", "notes2": ""}, {"unit": "cm", "job_title": "V.P.", "amount": "$3,640.92", "company": "   Cyberdyne", "name": "Eve Johnson", "city": "sf", "is_active": "1", "notes2": ""}, {"unit": "Cm", "job_title": "v.p.   ", "amount": "$427.02", "company": "Vehement", "name": "Alice Brandt", "city": "sf", "is_active": "0", "notes2": ""}, {"unit": "centimeters", "job_title": "V.A.", "amount": "null", "company": "  Initech", "name": "  Bob Fischer", "city": "  mugina", "is_active": "Y", "notes2": ""}, {"unit": "cm", "job_title": "  PM", "amount": "$8,163.49", "company": "Hooli", "name": "David Johnson", "city": "san francisco", "is_active": "FALSE", "notes2": ""}, {"unit": "milliliter", "job_title": "Product Manager", "amount": "?", "company": "Initech", "name": "Bob  Reyes", "city": "Mugina", "is_active": "0", "notes2": ""}, {"unit": "centimeters", "job_title": "pm", "amount": "7.693,03", "company": "Hooli", "name": "Bob Novak", "city": "sf", "is_active": "Yes", "notes2": ""}, {"unit": " centimeter ", "job_title": "Pm", "amount": "$1,020.53", "company": "Wonka Co", "name": "Priya Petrov ", "city": "MUGINA", "is_active": "T", "notes2": ""}, {"unit": "MLS", "job_title": "Vice  President", "amount": "$6,846.46", "company": "   Globex", "name": "Frank Diaz   ", "city": "poenix", "is_active": "1", "notes2": ""}, {"unit": "MILLILITER", "job_title": "pm", "amount": "1.197,91", "company": "Hooli   ", "name": "  Eve Wong", "city": "PHOENIX", "is_active": "Y", "notes2": ""}, {"unit": "millilietr", "job_title": "V.P.", "amount": "$499.08", "company": " Cyberdyne", "name": "Bob Cruz", "city": "  mugina", "is_active": "1", "notes2": ""}, {"unit": "mL", "job_title": "Vice  Prjs", "amount": "5.390,49", "company": "Stark Industries", "name": "Eve  Petrov", "city": "Phoenix", "is_active": "N", "notes2": ""}, {"unit": "mL", "job_title": "Prod Mgr", "amount": "3.918,58", "company": "   Initech", "name": "Sara Petrov", "city": "SF", "is_active": "F", "notes2": ""}, {"unit": "Ml", "job_title": " product manager", "amount": "7.217,74", "company": "Vehement", "name": "Eve Fischer   ", "city": "Mugina", "is_active": "Y", "notes2": ""}, {"unit": "cm", "job_title": "   Prod Manager", "amount": "$2,764.68", "company": " Initech", "name": "Lena Khan", "city": "sae fran", "is_active": "N", "notes2": ""}, {"unit": "ml", "job_title": "Prod Manager", "amount": "4.170,60", "company": "Initech", "name": "   David Adams", "city": "san fran", "is_active": "1", "notes2": ""}, {"unit": "centimeters", "job_title": "Vice Pres", "amount": "1.497,49", "company": " Initech", "name": "Judy  Park", "city": "Phoenix", "is_active": "FALSE", "notes2": ""}, {"unit": "CEENTIMETER", "job_title": "prod mgr   ", "amount": "$4,303.99", "company": " Initech ", "name": "Frank Park", "city": "SF", "is_active": "TRUE", "notes2": ""}, {"unit": "ml", "job_title": "prod mggr", "amount": "6.308,75", "company": " Cyberdyne", "name": "Ivan Smith", "city": "phoenix", "is_active": "TRUE", "notes2": ""}, {"unit": "ml", "job_title": "pm", "amount": "$4,656.36", "company": "Hooli", "name": "Frank Smith   ", "city": "Sf", "is_active": "Yes", "notes2": ""}], "clean": [{"unit": "mL", "job_title": "Vice President", "amount": NaN, "company": "Umbrella", "name": "Judy Adams", "city": "Mugina", "is_active": false}, {"unit": "mL", "job_title": "Vice President", "amount": 6024.96, "company": "Soylent Corp", "name": "Omar Fischer", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 3192.34, "company": "Umbrella", "name": "Frank Cruz", "city": "Phoenix", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 6852.42, "company": "Soylent Corp", "name": "David Novak", "city": "Phoenix", "is_active": false}, {"unit": "cm", "job_title": "Product Manager", "amount": 8851.32, "company": "Vehement", "name": "Priya Khan", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 8504.54, "company": "Wonka Co", "name": "Carol Petrov", "city": "Mugina", "is_active": true}, {"unit": "mL", "job_title": "Product Manager", "amount": 334.49, "company": "Vehement", "name": "Frank Park", "city": "Phoenix", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 3760.35, "company": "Stark Industries", "name": "Eve Petrov", "city": "Mugina", "is_active": false}, {"unit": "cm", "job_title": "Product Manager", "amount": 5118.98, "company": "Cyberdyne", "name": "Priya Brandt", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 8990.21, "company": "Wonka Co", "name": "Omar Smith", "city": "Mugina", "is_active": false}, {"unit": "cm", "job_title": "Vice President", "amount": 8633.98, "company": "Wonka Co", "name": "Frank Adams", "city": "San Francisco", "is_active": true}, {"unit": "mL", "job_title": "Product Manager", "amount": NaN, "company": "Initech", "name": "Lena Khan", "city": "San Francisco", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 5725.5, "company": "Hooli", "name": "Frank Johnson", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 3266.99, "company": "Globex", "name": "Bob Smith", "city": "Mugina", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 5576.66, "company": "Stark Industries", "name": "David Johnson", "city": "Mugina", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 1695.69, "company": "Cyberdyne", "name": "Karl Lee", "city": "Mugina", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 1351.17, "company": "Umbrella", "name": "Carol Fischer", "city": "Phoenix", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 7065.19, "company": "Hooli", "name": "Heidi Novak", "city": "Phoenix", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 4427.41, "company": "Wonka Co", "name": "Karl Diaz", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 8411.55, "company": "Hooli", "name": "Carol Smith", "city": "Phoenix", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": NaN, "company": "Soylent Corp", "name": "Bob Johnson", "city": "Mugina", "is_active": false}, {"unit": "cm", "job_title": "Product Manager", "amount": 1626.85, "company": "Stark Industries", "name": "Alice Wong", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 1656.8, "company": "Cyberdyne", "name": "Ivan Novak", "city": "Phoenix", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 3360.81, "company": "Hooli", "name": "Ivan Fischer", "city": "Phoenix", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 3472.24, "company": "Cyberdyne", "name": "Heidi Khan", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 7657.08, "company": "Hooli", "name": "Sara Johnson", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 7116.97, "company": "Umbrella", "name": "Priya Fischer", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Vice President", "amount": 7954.51, "company": "Soylent Corp", "name": "Heidi Khan", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": NaN, "company": "Hooli", "name": "Heidi Johnson", "city": "Mugina", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 4995.46, "company": "Cyberdyne", "name": "Heidi Wong", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 5916.47, "company": "Globex", "name": "Alice Fischer", "city": "Mugina", "is_active": false}, {"unit": "mL", "job_title": "Vice President", "amount": NaN, "company": "Umbrella", "name": "Sara Brandt", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 3640.92, "company": "Cyberdyne", "name": "Eve Johnson", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 427.02, "company": "Vehement", "name": "Alice Brandt", "city": "San Francisco", "is_active": false}, {"unit": "cm", "job_title": "Vice President", "amount": NaN, "company": "Initech", "name": "Bob Fischer", "city": "Mugina", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 8163.49, "company": "Hooli", "name": "David Johnson", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": NaN, "company": "Initech", "name": "Bob Reyes", "city": "Mugina", "is_active": false}, {"unit": "cm", "job_title": "Product Manager", "amount": 7693.03, "company": "Hooli", "name": "Bob Novak", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 1020.53, "company": "Wonka Co", "name": "Priya Petrov", "city": "Mugina", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 6846.46, "company": "Globex", "name": "Frank Diaz", "city": "Phoenix", "is_active": true}, {"unit": "mL", "job_title": "Product Manager", "amount": 1197.91, "company": "Hooli", "name": "Eve Wong", "city": "Phoenix", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 499.08, "company": "Cyberdyne", "name": "Bob Cruz", "city": "Mugina", "is_active": true}, {"unit": "mL", "job_title": "Vice President", "amount": 5390.49, "company": "Stark Industries", "name": "Eve Petrov", "city": "Phoenix", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 3918.58, "company": "Initech", "name": "Sara Petrov", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 7217.74, "company": "Vehement", "name": "Eve Fischer", "city": "Mugina", "is_active": true}, {"unit": "cm", "job_title": "Product Manager", "amount": 2764.68, "company": "Initech", "name": "Lena Khan", "city": "San Francisco", "is_active": false}, {"unit": "mL", "job_title": "Product Manager", "amount": 4170.6, "company": "Initech", "name": "David Adams", "city": "San Francisco", "is_active": true}, {"unit": "cm", "job_title": "Vice President", "amount": 1497.49, "company": "Initech", "name": "Judy Park", "city": "Phoenix", "is_active": false}, {"unit": "cm", "job_title": "Product Manager", "amount": 4303.99, "company": "Initech", "name": "Frank Park", "city": "San Francisco", "is_active": true}, {"unit": "mL", "job_title": "Product Manager", "amount": 6308.75, "company": "Cyberdyne", "name": "Ivan Smith", "city": "Phoenix", "is_active": true}, {"unit": "mL", "job_title": "Product Manager", "amount": 4656.36, "company": "Hooli", "name": "Frank Smith", "city": "San Francisco", "is_active": true}], "dirty_cols": ["unit", "job_title", "amount", "company", "name", "city", "is_active", "notes2"], "clean_cols": ["unit", "job_title", "amount", "company", "name", "city", "is_active"], "plan": {"dataset_summary": "51 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ML": "mL", "milliliter": "mL", "Cetnimeters": "cm", "centimeter": "cm", "MILLILITER": "mL", "centimeters": "cm", "mls": "mL", "CM": "cm", "ml": "mL", "Cm": "cm", "Ml": "mL", "CENTIMETER": "cm", "MLS": "mL", "Milliliter": "mL", "millilitter": "mL", "millilietr": "mL", "CEENTIMETER": "cm"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"V.P.": "Vice President", "vice president": "Vice President", "prod manager": "Product Manager", "Prod Manager": "Product Manager", "Vice Pres": "Vice President", "Prod Mgr": "Product Manager", "V..P.": "Vice President", "VP": "Vice President", "PM": "Product Manager", "VICE PRESIDENT": "Vice President", "v.p.": "Vice President", "PROD MGR": "Product Manager", "vp": "Vice President", "V.A.": "Vice President", "pm": "Product Manager", "Pm": "Product Manager", "Vice Prjs": "Vice President", "product manager": "Product Manager", "prod mgr": "Product Manager", "prod mggr": "Product Manager"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"San Fran": "San Francisco", "Sf": "San Francisco", "PHOMNIX": "Phoenix", "mugina": "Mugina", "sf": "San Francisco", "san fran": "San Francisco", "Mugipa": "Mugina", "phoenix": "Phoenix", "PHOENIX": "Phoenix", "MUGINA": "Mugina", "phhenix": "Phoenix", "SF": "San Francisco", "SAN FRAN": "San Francisco", "san francisco": "San Francisco", "poenix": "Phoenix", "sae fran": "San Francisco"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"department": "peoplle", "state": "LA", "job_title": "   Chieu Executive Officer", "signup_date": "4/21/2023", "is_active": "0", "country": "RU", "email": "odmjcf@corp.io", "city": "Crespo", "notes2": ""}, {"department": " OPS ", "state": " TN ", "job_title": "Chief Executive Officer", "signup_date": "5/11/2023", "is_active": "Yes", "country": "Republic of uriname", "email": "pgpnfi@corp.io", "city": "Crespo", "notes2": ""}, {"department": "R&D", "state": "La   ", "job_title": " AE ", "signup_date": "44998", "is_active": "false", "country": "Gm", "email": "dhamb@corp.io", "city": "daanbantayan", "notes2": ""}, {"department": "  ops", "state": "TN", "job_title": "   Account Executive", "signup_date": "6/19/2023", "is_active": "TRUE", "country": "Russga  ", "email": "   JBJPOA@EXAMPLE.COM", "city": "Crespo", "notes2": ""}, {"department": "Dev", "state": "American Samoa", "job_title": "Chief Executive Officer", "signup_date": "45207", "is_active": "N", "country": "RUS", "email": "   LEMHBJO@EXAMPLE.COM", "city": "crespo", "notes2": ""}, {"department": "People pOs", "state": "TN", "job_title": "ENG MGR", "signup_date": "2023-12-17", "is_active": "T", "country": "Suriname", "email": "HNEHGCD@TEST.ORG", "city": "Severomorsk   ", "notes2": ""}, {"department": " Dev ", "state": "american samoa  ", "job_title": "CHIEF EXECUTIVE OFFICER", "signup_date": "8/23/2023", "is_active": "T", "country": "ru", "email": "geio@corp.io", "city": "Cresspo", "notes2": ""}, {"department": "OPERATIONS", "state": "Tennessee", "job_title": "ae", "signup_date": "45192", "is_active": "F", "country": " Sr", "email": "ikbhil@mail.com", "city": " daanbantayan ", "notes2": ""}, {"department": "eng.", "state": "South Carolina", "job_title": "CEO  ", "signup_date": "45118", "is_active": "false", "country": "RU   ", "email": "GFOPAEH@MAIL.COM", "city": "Daanbantayan   ", "notes2": ""}, {"department": "Mazketing", "state": " Tennessee", "job_title": "Chief Executive Offier ", "signup_date": "2023-11-08", "is_active": "TRUE", "country": "SUR   ", "email": "   BPNDNE@MAIL.COM", "city": "Sevreomorsk", "notes2": ""}, {"department": "   dev", "state": "AS", "job_title": "Eng Mgr", "signup_date": "2023-05-03", "is_active": "0", "country": "AIA", "email": " JGHK@TEST.ORG ", "city": "crespo", "notes2": ""}, {"department": "HR", "state": "tn   ", "job_title": "CTO", "signup_date": "2023-06-04", "is_active": "Yes", "country": "BDI", "email": "adjh@mail.com", "city": "  Severomorsk", "notes2": ""}, {"department": " marketing ", "state": "American Samoa", "job_title": "acct exec", "signup_date": "2023-07-12", "is_active": "false", "country": "RU", "email": " bhdje@example.com", "city": "SEVEROMORSK", "notes2": ""}, {"department": "Ops   ", "state": "Americann Samoa", "job_title": "Engineering Manager", "signup_date": "45217", "is_active": "false", "country": "bi", "email": "JPLI@EXAMPLE.COM", "city": "Severomorsk", "notes2": ""}, {"department": "growth", "state": "SC", "job_title": "   c.e.o.", "signup_date": "7 Feb 2023", "is_active": "TRUE", "country": "Republic of Suriname", "email": "hhfdoj@example.com", "city": "Crespo", "notes2": ""}, {"department": "Ops", "state": "SOUT CAROLINA", "job_title": "C.E.O.", "signup_date": "44959", "is_active": "true", "country": " AI", "email": "EJHJF@CORP.IO  ", "city": "Crespo", "notes2": ""}, {"department": "HR", "state": "Tennessee", "job_title": "engineering mgr", "signup_date": "17 Apr 2023", "is_active": "TRUE", "country": "RUS", "email": "   HEFCAJ@CORP.IO", "city": " crespo", "notes2": ""}, {"department": "ops", "state": " AS ", "job_title": "Chief Executive Officer", "signup_date": "9 Feb 2023", "is_active": "No", "country": "Republic  of Suriname", "email": "ocegm@test.org   ", "city": "Daanbantayan", "notes2": ""}, {"department": "MKT", "state": "LA", "job_title": "chief executive officer", "signup_date": "2023-06-18", "is_active": "true", "country": " RU ", "email": "pjkjp@example.com", "city": "Crespo", "notes2": ""}, {"department": "R&D", "state": "AS", "job_title": "CHIEF TECHNOLOGY OFFICER", "signup_date": "2023-07-11", "is_active": "N", "country": "AI", "email": " GNNN@EXAMPLE.COM", "city": "famagusta", "notes2": ""}, {"department": "Mkt ", "state": "LA", "job_title": "cto", "signup_date": "5 Jan 2023", "is_active": "0", "country": "Sr", "email": " bhdim@corp.io", "city": "crespo", "notes2": ""}, {"department": " Ops ", "state": " la ", "job_title": "account executive", "signup_date": "9/16/2023", "is_active": "Y", "country": "Republic of Burundi", "email": " COPELJ@EXAMPLE.COM ", "city": "Cespo", "notes2": ""}, {"department": "Ops", "state": "as", "job_title": "Cto", "signup_date": "2023-12-11", "is_active": "T", "country": "SR", "email": "cchghoh@corp.io", "city": " Daanbantayan", "notes2": ""}, {"department": "DEV", "state": "SC  ", "job_title": "Engineering Mgr", "signup_date": "2023-03-22", "is_active": "F", "country": "AIA", "email": "bckcn@corp.io", "city": "Daanbantayan", "notes2": ""}, {"department": "Operattions", "state": "tn", "job_title": "Chief Technology Officer", "signup_date": "2023-12-28", "is_active": "Y", "country": "russian federation", "email": "dfldbmn@example.com", "city": "Daanbantayan", "notes2": ""}, {"department": "PEOPLE  OPS", "state": "South Carolina", "job_title": "Eng Manaber", "signup_date": "4/24/2023", "is_active": "true", "country": "GMB", "email": "GMJJ@CORP.IO", "city": "crespo", "notes2": ""}, {"department": "eng.", "state": "la", "job_title": "Acct Exec", "signup_date": "13 Mar 2023", "is_active": "Yes", "country": "Russia", "email": " CEABGO@EXAMPLE.COM ", "city": "CRESPO", "notes2": ""}, {"department": "Operatons", "state": "SC", "job_title": "engineering mgr", "signup_date": "2023-07-26", "is_active": "No", "country": "anguipla", "email": "JOJGE@TEST.ORG", "city": "CRESPO", "notes2": ""}, {"department": "Engineering", "state": "Amercan  Samoa", "job_title": "   ceo", "signup_date": "2023-06-19", "is_active": "Yes", "country": "burundi", "email": "  BFCEJ@EXAMPLE.COM", "city": "Daanbantqyan", "notes2": ""}, {"department": "Eng ", "state": "Louisiana ", "job_title": "Engineering Mgr", "signup_date": "5/20/2023", "is_active": "true", "country": "   republic of the gambia", "email": " LJFLEF@MAIL.COM ", "city": "FAMAGUSTA", "notes2": ""}, {"department": "operations", "state": "TN", "job_title": "CTO", "signup_date": "2023-02-17", "is_active": "FALSE", "country": "sr   ", "email": " NJIE@TEST.ORG", "city": "Severomorsk", "notes2": ""}, {"department": "Dev", "state": "  AS", "job_title": "CEO", "signup_date": "44927", "is_active": "Y", "country": "aia ", "email": "  EJACOH@TEST.ORG", "city": "Severomorsk", "notes2": ""}], "clean": [{"department": "Human Resources", "state": "Louisiana", "job_title": "Chief Executive Officer", "signup_date": "2023-04-21", "is_active": false, "country": "Russian Federation", "email": "odmjcf@corp.io", "city": "Crespo"}, {"department": "Operations", "state": "Tennessee", "job_title": "Chief Executive Officer", "signup_date": "2023-05-11", "is_active": true, "country": "Suriname", "email": "pgpnfi@corp.io", "city": "Crespo"}, {"department": "Engineering", "state": "Louisiana", "job_title": "Account Executive", "signup_date": "2023-03-13", "is_active": false, "country": "Gambia", "email": "dhamb@corp.io", "city": "Daanbantayan"}, {"department": "Operations", "state": "Tennessee", "job_title": "Account Executive", "signup_date": "2023-06-19", "is_active": true, "country": "Russian Federation", "email": "jbjpoa@example.com", "city": "Crespo"}, {"department": "Engineering", "state": "American Samoa", "job_title": "Chief Executive Officer", "signup_date": "2023-10-08", "is_active": false, "country": "Russian Federation", "email": "lemhbjo@example.com", "city": "Crespo"}, {"department": "Human Resources", "state": "Tennessee", "job_title": "Engineering Manager", "signup_date": "2023-12-17", "is_active": true, "country": "Suriname", "email": "hnehgcd@test.org", "city": "Severomorsk"}, {"department": "Engineering", "state": "American Samoa", "job_title": "Chief Executive Officer", "signup_date": "2023-08-23", "is_active": true, "country": "Russian Federation", "email": "geio@corp.io", "city": "Crespo"}, {"department": "Operations", "state": "Tennessee", "job_title": "Account Executive", "signup_date": "2023-09-23", "is_active": false, "country": "Suriname", "email": "ikbhil@mail.com", "city": "Daanbantayan"}, {"department": "Engineering", "state": "South Carolina", "job_title": "Chief Executive Officer", "signup_date": "2023-07-11", "is_active": false, "country": "Russian Federation", "email": "gfopaeh@mail.com", "city": "Daanbantayan"}, {"department": "Marketing", "state": "Tennessee", "job_title": "Chief Executive Officer", "signup_date": "2023-11-08", "is_active": true, "country": "Suriname", "email": "bpndne@mail.com", "city": "Severomorsk"}, {"department": "Engineering", "state": "American Samoa", "job_title": "Engineering Manager", "signup_date": "2023-05-03", "is_active": false, "country": "Anguilla", "email": "jghk@test.org", "city": "Crespo"}, {"department": "Human Resources", "state": "Tennessee", "job_title": "Chief Technology Officer", "signup_date": "2023-06-04", "is_active": true, "country": "Burundi", "email": "adjh@mail.com", "city": "Severomorsk"}, {"department": "Marketing", "state": "American Samoa", "job_title": "Account Executive", "signup_date": "2023-07-12", "is_active": false, "country": "Russian Federation", "email": "bhdje@example.com", "city": "Severomorsk"}, {"department": "Operations", "state": "American Samoa", "job_title": "Engineering Manager", "signup_date": "2023-10-18", "is_active": false, "country": "Burundi", "email": "jpli@example.com", "city": "Severomorsk"}, {"department": "Marketing", "state": "South Carolina", "job_title": "Chief Executive Officer", "signup_date": "2023-02-07", "is_active": true, "country": "Suriname", "email": "hhfdoj@example.com", "city": "Crespo"}, {"department": "Operations", "state": "South Carolina", "job_title": "Chief Executive Officer", "signup_date": "2023-02-02", "is_active": true, "country": "Anguilla", "email": "ejhjf@corp.io", "city": "Crespo"}, {"department": "Human Resources", "state": "Tennessee", "job_title": "Engineering Manager", "signup_date": "2023-04-17", "is_active": true, "country": "Russian Federation", "email": "hefcaj@corp.io", "city": "Crespo"}, {"department": "Operations", "state": "American Samoa", "job_title": "Chief Executive Officer", "signup_date": "2023-02-09", "is_active": false, "country": "Suriname", "email": "ocegm@test.org", "city": "Daanbantayan"}, {"department": "Marketing", "state": "Louisiana", "job_title": "Chief Executive Officer", "signup_date": "2023-06-18", "is_active": true, "country": "Russian Federation", "email": "pjkjp@example.com", "city": "Crespo"}, {"department": "Engineering", "state": "American Samoa", "job_title": "Chief Technology Officer", "signup_date": "2023-07-11", "is_active": false, "country": "Anguilla", "email": "gnnn@example.com", "city": "Famagusta"}, {"department": "Marketing", "state": "Louisiana", "job_title": "Chief Technology Officer", "signup_date": "2023-01-05", "is_active": false, "country": "Suriname", "email": "bhdim@corp.io", "city": "Crespo"}, {"department": "Operations", "state": "Louisiana", "job_title": "Account Executive", "signup_date": "2023-09-16", "is_active": true, "country": "Burundi", "email": "copelj@example.com", "city": "Crespo"}, {"department": "Operations", "state": "American Samoa", "job_title": "Chief Technology Officer", "signup_date": "2023-12-11", "is_active": true, "country": "Suriname", "email": "cchghoh@corp.io", "city": "Daanbantayan"}, {"department": "Engineering", "state": "South Carolina", "job_title": "Engineering Manager", "signup_date": "2023-03-22", "is_active": false, "country": "Anguilla", "email": "bckcn@corp.io", "city": "Daanbantayan"}, {"department": "Operations", "state": "Tennessee", "job_title": "Chief Technology Officer", "signup_date": "2023-12-28", "is_active": true, "country": "Russian Federation", "email": "dfldbmn@example.com", "city": "Daanbantayan"}, {"department": "Human Resources", "state": "South Carolina", "job_title": "Engineering Manager", "signup_date": "2023-04-24", "is_active": true, "country": "Gambia", "email": "gmjj@corp.io", "city": "Crespo"}, {"department": "Engineering", "state": "Louisiana", "job_title": "Account Executive", "signup_date": "2023-03-13", "is_active": true, "country": "Russian Federation", "email": "ceabgo@example.com", "city": "Crespo"}, {"department": "Operations", "state": "South Carolina", "job_title": "Engineering Manager", "signup_date": "2023-07-26", "is_active": false, "country": "Anguilla", "email": "jojge@test.org", "city": "Crespo"}, {"department": "Engineering", "state": "American Samoa", "job_title": "Chief Executive Officer", "signup_date": "2023-06-19", "is_active": true, "country": "Burundi", "email": "bfcej@example.com", "city": "Daanbantayan"}, {"department": "Engineering", "state": "Louisiana", "job_title": "Engineering Manager", "signup_date": "2023-05-20", "is_active": true, "country": "Gambia", "email": "ljflef@mail.com", "city": "Famagusta"}, {"department": "Operations", "state": "Tennessee", "job_title": "Chief Technology Officer", "signup_date": "2023-02-17", "is_active": false, "country": "Suriname", "email": "njie@test.org", "city": "Severomorsk"}, {"department": "Engineering", "state": "American Samoa", "job_title": "Chief Executive Officer", "signup_date": "2023-01-01", "is_active": true, "country": "Anguilla", "email": "ejacoh@test.org", "city": "Severomorsk"}], "dirty_cols": ["department", "state", "job_title", "signup_date", "is_active", "country", "email", "city", "notes2"], "clean_cols": ["department", "state", "job_title", "signup_date", "is_active", "country", "email", "city"], "plan": {"dataset_summary": "32 rows × 9 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"peoplle": "Human Resources", "OPS": "Operations", "R&D": "Engineering", "ops": "Operations", "Dev": "Engineering", "People pOs": "Human Resources", "OPERATIONS": "Operations", "eng.": "Engineering", "Mazketing": "Marketing", "dev": "Engineering", "HR": "Human Resources", "marketing": "Marketing", "Ops": "Operations", "growth": "Marketing", "MKT": "Marketing", "Mkt": "Marketing", "DEV": "Engineering", "Operattions": "Operations", "PEOPLE OPS": "Human Resources", "Operatons": "Operations", "Eng": "Engineering", "operations": "Operations"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LA": "Louisiana", "TN": "Tennessee", "La": "Louisiana", "american samoa": "American Samoa", "AS": "American Samoa", "tn": "Tennessee", "Americann Samoa": "American Samoa", "SC": "South Carolina", "SOUT CAROLINA": "South Carolina", "la": "Louisiana", "as": "American Samoa", "Amercan Samoa": "American Samoa"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Chieu Executive Officer": "Chief Executive Officer", "AE": "Account Executive", "ENG MGR": "Engineering Manager", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "ae": "Account Executive", "CEO": "Chief Executive Officer", "Chief Executive Offier": "Chief Executive Officer", "Eng Mgr": "Engineering Manager", "CTO": "Chief Technology Officer", "acct exec": "Account Executive", "c.e.o.": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer", "engineering mgr": "Engineering Manager", "chief executive officer": "Chief Executive Officer", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "cto": "Chief Technology Officer", "account executive": "Account Executive", "Cto": "Chief Technology Officer", "Engineering Mgr": "Engineering Manager", "Eng Manaber": "Engineering Manager", "Acct Exec": "Account Executive", "ceo": "Chief Executive Officer"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"RU": "Russian Federation", "Republic of uriname": "Suriname", "Gm": "Gambia", "Russga": "Russian Federation", "RUS": "Russian Federation", "ru": "Russian Federation", "Sr": "Suriname", "SUR": "Suriname", "AIA": "Anguilla", "BDI": "Burundi", "bi": "Burundi", "Republic of Suriname": "Suriname", "AI": "Anguilla", "Republic of Burundi": "Burundi", "SR": "Suriname", "russian federation": "Russian Federation", "GMB": "Gambia", "Russia": "Russian Federation", "anguipla": "Anguilla", "burundi": "Burundi", "republic of the gambia": "Gambia", "sr": "Suriname", "aia": "Anguilla"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"daanbantayan": "Daanbantayan", "crespo": "Crespo", "Cresspo": "Crespo", "Sevreomorsk": "Severomorsk", "SEVEROMORSK": "Severomorsk", "famagusta": "Famagusta", "Cespo": "Crespo", "CRESPO": "Crespo", "Daanbantqyan": "Daanbantayan", "FAMAGUSTA": "Famagusta"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "true", "amount": "$2,104.26", "name": "Priya Wong", "phone": "581.399.5512", "city": "Manzini", "status": " P3 ", "job_title": "  prod mgr", "extra": ""}, {"is_active": "FALSE", "amount": "$4,548.98", "name": "Lena Petrov", "phone": "558-060-5372", "city": "HIGHTON", "status": "Medium", "job_title": "Admin Assistant", "extra": ""}, {"is_active": "false", "amount": "$8,406.28", "name": " Bob Petrov", "phone": "5532058823", "city": "  Highton", "status": "p2  ", "job_title": "Admin Assistant", "extra": ""}, {"is_active": "0", "amount": "1.731,26", "name": "Ivan  Novak", "phone": "565-592-0652", "city": "manzini", "status": "medium  ", "job_title": "Prod Mgr", "extra": ""}, {"is_active": "TRUE", "amount": "$7,229.92", "name": " Karl Khan", "phone": "514-179-3831", "city": "CDMX", "status": "P2", "job_title": "snr engineer", "extra": ""}, {"is_active": "N", "amount": "387,73", "name": "Frank Cruz  ", "phone": "597.926.7912", "city": "Severomorsk  ", "status": "P1 ", "job_title": "PM", "extra": ""}, {"is_active": "F", "amount": "953,69", "name": "Grace Wong", "phone": "5484009242", "city": " CDMX ", "status": "LOW", "job_title": "PRODUCT MANAGER", "extra": ""}, {"is_active": "Yes", "amount": "$2,208.39", "name": "Lena Adams", "phone": "579.678.6224", "city": "MANZINI", "status": "low", "job_title": "snr engineer", "extra": ""}, {"is_active": "true", "amount": "870,15", "name": "Omar Lee", "phone": "510-657-3142", "city": "  Highton", "status": "P2", "job_title": " PM ", "extra": ""}, {"is_active": "FALSE", "amount": "$5,897.60", "name": "Heidi  Ali", "phone": "(564)637-4338", "city": "JejuqCity", "status": "P3", "job_title": "Prod Mgr", "extra": ""}, {"is_active": "TRUE", "amount": "8.549,28", "name": "Sara  Khan", "phone": "598.418.5287", "city": "Manzini", "status": "low", "job_title": "Administrative Assistant   ", "extra": ""}, {"is_active": "false", "amount": "6.639,01", "name": "Priya Khan", "phone": "(580)841-9827", "city": "mexico city", "status": "HIGH", "job_title": "pm", "extra": ""}, {"is_active": "No", "amount": "$2,680.07", "name": "Ivan Fischer  ", "phone": "5330725513", "city": "Manzini", "status": "m  ", "job_title": "PROD MANAGER", "extra": ""}, {"is_active": "T", "amount": "7.762,92", "name": "Karl  Cruz", "phone": "(551)897-8118", "city": "Manyini", "status": "   Low", "job_title": "snr engineer", "extra": ""}, {"is_active": "No", "amount": "6.515,21", "name": "Mona Ali   ", "phone": "559-488-5737", "city": "manzini", "status": "H ", "job_title": "Prod  Mgr", "extra": ""}, {"is_active": "N", "amount": "2.302,39", "name": "Omar Adams", "phone": "(546)655-6340", "city": "severomorsk", "status": "   P1", "job_title": "Senior Eng", "extra": ""}, {"is_active": "N", "amount": "$3,662.27", "name": "Carol Reyes", "phone": "(591)828-6699", "city": "Manzini", "status": "Low", "job_title": "  Snr Engineer", "extra": ""}, {"is_active": "T", "amount": "null", "name": "Lena  Cruz", "phone": "5920054345", "city": "highton", "status": "Low", "job_title": "Admin", "extra": ""}, {"is_active": "N", "amount": "$875.98", "name": "Alice  Novak", "phone": "5249524631", "city": "Manzini", "status": "High", "job_title": "Product  Manager", "extra": ""}, {"is_active": "Yes", "amount": "271,34", "name": "Bob Reyes", "phone": "5044368529", "city": "Highton", "status": "hjgh", "job_title": "Senior Engineer", "extra": ""}, {"is_active": "Y", "amount": "$6,361.66", "name": "   Grace Adams", "phone": "544-281-7345", "city": "severomorsk", "status": "med", "job_title": "Prod Manager", "extra": ""}, {"is_active": "TRUE", "amount": "#N/A", "name": "   Sara Adams", "phone": "592.687.9650", "city": "Manzini", "status": "M", "job_title": "Snr Engineer", "extra": ""}, {"is_active": "Yes", "amount": "$6,658.73", "name": "  Heidi Lee", "phone": "5864119944", "city": "manzini   ", "status": "MED", "job_title": "  Administrrative Assistant", "extra": ""}, {"is_active": "F", "amount": "$8,308.79", "name": "Judy  Johnson", "phone": "558.410.8608", "city": "Manzini", "status": "medium", "job_title": "Admin", "extra": ""}, {"is_active": "Y", "amount": "#N/A", "name": "Carol Diaz", "phone": "5585751447", "city": "Manzini", "status": " high", "job_title": "Product Manager", "extra": ""}, {"is_active": "false", "amount": "$8,429.11", "name": "Frank Brandt", "phone": "519-031-6829", "city": "jeju city", "status": "m", "job_title": "Senior Eng", "extra": ""}, {"is_active": "Yes", "amount": "$1,947.08", "name": "Frank Reyes  ", "phone": "519.530.2789", "city": "JEJU CITY", "status": "LOW", "job_title": "Sr Engineer", "extra": ""}, {"is_active": "No", "amount": "#N/A", "name": "Priya Novak", "phone": "592-235-4181", "city": " Manzini ", "status": "low", "job_title": "AdminA ssistant", "extra": ""}, {"is_active": "T", "amount": "6.612,25", "name": "Alice Diaz", "phone": "519.829.8991", "city": "CDMX", "status": "low", "job_title": "PROD  MANAGER", "extra": ""}, {"is_active": "FALSE", "amount": "$6,384.17", "name": "Karl Cruz", "phone": "5821130887", "city": "Ciudad de Mexico", "status": "P2  ", "job_title": "Admin Asst", "extra": ""}, {"is_active": "T", "amount": "$1,809.13", "name": "Alice Adams", "phone": "532.736.1658", "city": "MANZINI", "status": "H", "job_title": "senior engineer", "extra": ""}, {"is_active": "TRUE", "amount": "1.827,09", "name": "Eve Smith", "phone": "5002310285", "city": "severomorsk", "status": "l", "job_title": " administrative assistant", "extra": ""}, {"is_active": "FALSE", "amount": "85,84", "name": "Lena  Moore", "phone": "528.337.7758", "city": "Severomorsk   ", "status": " M", "job_title": "admin  assistant", "extra": ""}, {"is_active": "FALSE", "amount": "2.110,81", "name": "Priya  Diaz", "phone": "533-967-0746", "city": "highton", "status": "l", "job_title": "Sr Engineer   ", "extra": ""}, {"is_active": "TRUE", "amount": "1.467,59", "name": "Sara Adams", "phone": "(596)727-0772", "city": "Severomorsk", "status": "high", "job_title": "admin asst", "extra": ""}], "clean": [{"is_active": true, "amount": 2104.26, "name": "Priya Wong", "phone": "(581) 399-5512", "city": "Manzini", "status": "Low", "job_title": "Product Manager"}, {"is_active": false, "amount": 4548.98, "name": "Lena Petrov", "phone": "(558) 060-5372", "city": "Highton", "status": "Medium", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 8406.28, "name": "Bob Petrov", "phone": "(553) 205-8823", "city": "Highton", "status": "Medium", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 1731.26, "name": "Ivan Novak", "phone": "(565) 592-0652", "city": "Manzini", "status": "Medium", "job_title": "Product Manager"}, {"is_active": true, "amount": 7229.92, "name": "Karl Khan", "phone": "(514) 179-3831", "city": "Mexico City", "status": "Medium", "job_title": "Senior Engineer"}, {"is_active": false, "amount": 387.73, "name": "Frank Cruz", "phone": "(597) 926-7912", "city": "Severomorsk", "status": "High", "job_title": "Product Manager"}, {"is_active": false, "amount": 953.69, "name": "Grace Wong", "phone": "(548) 400-9242", "city": "Mexico City", "status": "Low", "job_title": "Product Manager"}, {"is_active": true, "amount": 2208.39, "name": "Lena Adams", "phone": "(579) 678-6224", "city": "Manzini", "status": "Low", "job_title": "Senior Engineer"}, {"is_active": true, "amount": 870.15, "name": "Omar Lee", "phone": "(510) 657-3142", "city": "Highton", "status": "Medium", "job_title": "Product Manager"}, {"is_active": false, "amount": 5897.6, "name": "Heidi Ali", "phone": "(564) 637-4338", "city": "Jeju City", "status": "Low", "job_title": "Product Manager"}, {"is_active": true, "amount": 8549.28, "name": "Sara Khan", "phone": "(598) 418-5287", "city": "Manzini", "status": "Low", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 6639.01, "name": "Priya Khan", "phone": "(580) 841-9827", "city": "Mexico City", "status": "High", "job_title": "Product Manager"}, {"is_active": false, "amount": 2680.07, "name": "Ivan Fischer", "phone": "(533) 072-5513", "city": "Manzini", "status": "Medium", "job_title": "Product Manager"}, {"is_active": true, "amount": 7762.92, "name": "Karl Cruz", "phone": "(551) 897-8118", "city": "Manzini", "status": "Low", "job_title": "Senior Engineer"}, {"is_active": false, "amount": 6515.21, "name": "Mona Ali", "phone": "(559) 488-5737", "city": "Manzini", "status": "High", "job_title": "Product Manager"}, {"is_active": false, "amount": 2302.39, "name": "Omar Adams", "phone": "(546) 655-6340", "city": "Severomorsk", "status": "High", "job_title": "Senior Engineer"}, {"is_active": false, "amount": 3662.27, "name": "Carol Reyes", "phone": "(591) 828-6699", "city": "Manzini", "status": "Low", "job_title": "Senior Engineer"}, {"is_active": true, "amount": NaN, "name": "Lena Cruz", "phone": "(592) 005-4345", "city": "Highton", "status": "Low", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 875.98, "name": "Alice Novak", "phone": "(524) 952-4631", "city": "Manzini", "status": "High", "job_title": "Product Manager"}, {"is_active": true, "amount": 271.34, "name": "Bob Reyes", "phone": "(504) 436-8529", "city": "Highton", "status": "High", "job_title": "Senior Engineer"}, {"is_active": true, "amount": 6361.66, "name": "Grace Adams", "phone": "(544) 281-7345", "city": "Severomorsk", "status": "Medium", "job_title": "Product Manager"}, {"is_active": true, "amount": NaN, "name": "Sara Adams", "phone": "(592) 687-9650", "city": "Manzini", "status": "Medium", "job_title": "Senior Engineer"}, {"is_active": true, "amount": 6658.73, "name": "Heidi Lee", "phone": "(586) 411-9944", "city": "Manzini", "status": "Medium", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 8308.79, "name": "Judy Johnson", "phone": "(558) 410-8608", "city": "Manzini", "status": "Medium", "job_title": "Administrative Assistant"}, {"is_active": true, "amount": NaN, "name": "Carol Diaz", "phone": "(558) 575-1447", "city": "Manzini", "status": "High", "job_title": "Product Manager"}, {"is_active": false, "amount": 8429.11, "name": "Frank Brandt", "phone": "(519) 031-6829", "city": "Jeju City", "status": "Medium", "job_title": "Senior Engineer"}, {"is_active": true, "amount": 1947.08, "name": "Frank Reyes", "phone": "(519) 530-2789", "city": "Jeju City", "status": "Low", "job_title": "Senior Engineer"}, {"is_active": false, "amount": NaN, "name": "Priya Novak", "phone": "(592) 235-4181", "city": "Manzini", "status": "Low", "job_title": "Administrative Assistant"}, {"is_active": true, "amount": 6612.25, "name": "Alice Diaz", "phone": "(519) 829-8991", "city": "Mexico City", "status": "Low", "job_title": "Product Manager"}, {"is_active": false, "amount": 6384.17, "name": "Karl Cruz", "phone": "(582) 113-0887", "city": "Mexico City", "status": "Medium", "job_title": "Administrative Assistant"}, {"is_active": true, "amount": 1809.13, "name": "Alice Adams", "phone": "(532) 736-1658", "city": "Manzini", "status": "High", "job_title": "Senior Engineer"}, {"is_active": true, "amount": 1827.09, "name": "Eve Smith", "phone": "(500) 231-0285", "city": "Severomorsk", "status": "Low", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 85.84, "name": "Lena Moore", "phone": "(528) 337-7758", "city": "Severomorsk", "status": "Medium", "job_title": "Administrative Assistant"}, {"is_active": false, "amount": 2110.81, "name": "Priya Diaz", "phone": "(533) 967-0746", "city": "Highton", "status": "Low", "job_title": "Senior Engineer"}, {"is_active": true, "amount": 1467.59, "name": "Sara Adams", "phone": "(596) 727-0772", "city": "Severomorsk", "status": "High", "job_title": "Administrative Assistant"}], "dirty_cols": ["is_active", "amount", "name", "phone", "city", "status", "job_title", "extra"], "clean_cols": ["is_active", "amount", "name", "phone", "city", "status", "job_title"], "plan": {"dataset_summary": "35 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HIGHTON": "Highton", "manzini": "Manzini", "CDMX": "Mexico City", "MANZINI": "Manzini", "JejuqCity": "Jeju City", "mexico city": "Mexico City", "Manyini": "Manzini", "severomorsk": "Severomorsk", "highton": "Highton", "jeju city": "Jeju City", "JEJU CITY": "Jeju City", "Ciudad de Mexico": "Mexico City"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P3": "Low", "p2": "Medium", "medium": "Medium", "P2": "Medium", "P1": "High", "LOW": "Low", "low": "Low", "HIGH": "High", "m": "Medium", "H": "High", "hjgh": "High", "med": "Medium", "M": "Medium", "MED": "Medium", "high": "High", "l": "Low"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"prod mgr": "Product Manager", "Admin Assistant": "Administrative Assistant", "Prod Mgr": "Product Manager", "snr engineer": "Senior Engineer", "PM": "Product Manager", "PRODUCT MANAGER": "Product Manager", "pm": "Product Manager", "PROD MANAGER": "Product Manager", "Senior Eng": "Senior Engineer", "Snr Engineer": "Senior Engineer", "Admin": "Administrative Assistant", "Prod Manager": "Product Manager", "Administrrative Assistant": "Administrative Assistant", "Sr Engineer": "Senior Engineer", "AdminA ssistant": "Administrative Assistant", "Admin Asst": "Administrative Assistant", "senior engineer": "Senior Engineer", "administrative assistant": "Administrative Assistant", "admin assistant": "Administrative Assistant", "admin asst": "Administrative Assistant"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"email": " CIKK@CORP.IO ", "status": "Paid", "unit": "liter", "amount": "4.414,64"}, {"email": "BNNHHG@CORP.IO   ", "status": " pad ", "unit": "l", "amount": "$8,801.80"}, {"email": "dokd@mail.com", "status": "refund", "unit": "cm", "amount": "$4,159.56"}, {"email": "ejalk@mail.com", "status": "refund", "unit": "L", "amount": "$8,232.68"}, {"email": "NEAHC@TEST.ORG", "status": "refund", "unit": "L", "amount": "$3,129.50"}, {"email": "gpeg@test.org ", "status": "Refunded", "unit": " milliliter", "amount": "9999999"}, {"email": "ecmijlh@mail.com", "status": "Refunded", "unit": "lbs", "amount": "4.687,10"}, {"email": " AJBLD@CORP.IO ", "status": "paid", "unit": "milliliter ", "amount": "$4,249.02"}, {"email": "   KOGC@CORP.IO", "status": "  PAID", "unit": "cetimeter", "amount": "$7,010.58"}, {"email": "OFEO@EXAMPLE.COM", "status": "Paid", "unit": "l", "amount": "1.585,48"}, {"email": "BNNHHG@CORP.IO   ", "status": " pad ", "unit": "l", "amount": "$8,801.80"}], "clean": [{"email": "cikk@corp.io", "status": "Paid", "unit": "L", "amount": 4414.64}, {"email": "bnnhhg@corp.io", "status": "Paid", "unit": "L", "amount": 8801.8}, {"email": "dokd@mail.com", "status": "Refunded", "unit": "cm", "amount": 4159.56}, {"email": "ejalk@mail.com", "status": "Refunded", "unit": "L", "amount": 8232.68}, {"email": "neahc@test.org", "status": "Refunded", "unit": "L", "amount": 3129.5}, {"email": "gpeg@test.org", "status": "Refunded", "unit": "mL", "amount": 9999999.0}, {"email": "ecmijlh@mail.com", "status": "Refunded", "unit": "lb", "amount": 4687.1}, {"email": "ajbld@corp.io", "status": "Paid", "unit": "mL", "amount": 4249.02}, {"email": "kogc@corp.io", "status": "Paid", "unit": "cm", "amount": 7010.58}, {"email": "ofeo@example.com", "status": "Paid", "unit": "L", "amount": 1585.48}], "dirty_cols": ["email", "status", "unit", "amount"], "clean_cols": ["email", "status", "unit", "amount"], "plan": {"dataset_summary": "11 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pad": "Paid", "refund": "Refunded", "paid": "Paid", "PAID": "Paid"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liter": "L", "l": "L", "milliliter": "mL", "lbs": "lb", "cetimeter": "cm"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"country": " bahrain ", "is_active": "true", "company": "   Soylent Corp", "industry": "education", "job_title": "Administrative Assistant", "col_x": ""}, {"country": "  qa", "is_active": "false", "company": "Globex   ", "industry": "Education", "job_title": "snr engineer", "col_x": ""}, {"country": "Kingdom of Bahrain", "is_active": "N", "company": "Initech", "industry": "  Ed", "job_title": "Abmin", "col_x": ""}, {"country": " Bh ", "is_active": "Yes", "company": "Vehement", "industry": "it", "job_title": " engineering mgr", "col_x": ""}, {"country": "QAT", "is_active": "F", "company": " Umbrella", "industry": "edtech", "job_title": " Admin Asst", "col_x": ""}, {"country": "BH", "is_active": "1", "company": " Globex ", "industry": "Tech", "job_title": "Eng Manager", "col_x": ""}, {"country": "  Kingdom Of Bahrain", "is_active": "false", "company": "Umbrella", "industry": "manufacturing", "job_title": "  Senior Engineer", "col_x": ""}, {"country": "BHR", "is_active": "TRUE", "company": "Cyberdyne", "industry": "EDUCATION", "job_title": "Admin Assistant", "col_x": ""}, {"country": "Bahrain", "is_active": "FALSE", "company": "Hooli", "industry": "   education", "job_title": "Eng Manager", "col_x": ""}, {"country": "qatar", "is_active": "1", "company": " Umbrella ", "industry": "Manufacturing", "job_title": "Senior Eng", "col_x": ""}, {"country": "Kingdom of Bahrain", "is_active": "false", "company": " Wonka Co", "industry": "Tech", "job_title": "Engineering Mgr", "col_x": ""}, {"country": "Kingdom of Bahrain", "is_active": "TRUE", "company": "  Stark Industries", "industry": "mfg", "job_title": "Admin Assistant", "col_x": ""}, {"country": "   qat", "is_active": "F", "company": "  Stark Industries", "industry": "industrial", "job_title": "eng mgr", "col_x": ""}, {"country": "bh", "is_active": "No", "company": "   Soylent Corp", "industry": "Industrial", "job_title": "Addmin Assistant", "col_x": ""}, {"country": "STATE OF QATAR", "is_active": "T", "company": " Hooli ", "industry": "MANUFACTURING", "job_title": "Senior Engineer", "col_x": ""}, {"country": "  QATAR", "is_active": "0", "company": " Soylent Corp", "industry": "Manufacxuring", "job_title": "admin", "col_x": ""}, {"country": "bh", "is_active": "false", "company": "Globex", "industry": "EDUCATION", "job_title": "Admin Assistant", "col_x": ""}, {"country": "Bahrain", "is_active": "FALSE", "company": "Hooli", "industry": "   education", "job_title": "Eng Manager", "col_x": ""}, {"country": "", "is_active": "", "company": "", "industry": "", "job_title": "", "col_x": ""}], "clean": [{"country": "Bahrain", "is_active": true, "company": "Soylent Corp", "industry": "Education", "job_title": "Administrative Assistant"}, {"country": "Qatar", "is_active": false, "company": "Globex", "industry": "Education", "job_title": "Senior Engineer"}, {"country": "Bahrain", "is_active": false, "company": "Initech", "industry": "Education", "job_title": "Administrative Assistant"}, {"country": "Bahrain", "is_active": true, "company": "Vehement", "industry": "Technology", "job_title": "Engineering Manager"}, {"country": "Qatar", "is_active": false, "company": "Umbrella", "industry": "Education", "job_title": "Administrative Assistant"}, {"country": "Bahrain", "is_active": true, "company": "Globex", "industry": "Technology", "job_title": "Engineering Manager"}, {"country": "Bahrain", "is_active": false, "company": "Umbrella", "industry": "Manufacturing", "job_title": "Senior Engineer"}, {"country": "Bahrain", "is_active": true, "company": "Cyberdyne", "industry": "Education", "job_title": "Administrative Assistant"}, {"country": "Bahrain", "is_active": false, "company": "Hooli", "industry": "Education", "job_title": "Engineering Manager"}, {"country": "Qatar", "is_active": true, "company": "Umbrella", "industry": "Manufacturing", "job_title": "Senior Engineer"}, {"country": "Bahrain", "is_active": false, "company": "Wonka Co", "industry": "Technology", "job_title": "Engineering Manager"}, {"country": "Bahrain", "is_active": true, "company": "Stark Industries", "industry": "Manufacturing", "job_title": "Administrative Assistant"}, {"country": "Qatar", "is_active": false, "company": "Stark Industries", "industry": "Manufacturing", "job_title": "Engineering Manager"}, {"country": "Bahrain", "is_active": false, "company": "Soylent Corp", "industry": "Manufacturing", "job_title": "Administrative Assistant"}, {"country": "Qatar", "is_active": true, "company": "Hooli", "industry": "Manufacturing", "job_title": "Senior Engineer"}, {"country": "Qatar", "is_active": false, "company": "Soylent Corp", "industry": "Manufacturing", "job_title": "Administrative Assistant"}, {"country": "Bahrain", "is_active": false, "company": "Globex", "industry": "Education", "job_title": "Administrative Assistant"}], "dirty_cols": ["country", "is_active", "company", "industry", "job_title", "col_x"], "clean_cols": ["country", "is_active", "company", "industry", "job_title"], "plan": {"dataset_summary": "19 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bahrain": "Bahrain", "qa": "Qatar", "Kingdom of Bahrain": "Bahrain", "Bh": "Bahrain", "QAT": "Qatar", "BH": "Bahrain", "Kingdom Of Bahrain": "Bahrain", "BHR": "Bahrain", "qatar": "Qatar", "qat": "Qatar", "bh": "Bahrain", "STATE OF QATAR": "Qatar", "QATAR": "Qatar"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"education": "Education", "Ed": "Education", "it": "Technology", "edtech": "Education", "Tech": "Technology", "manufacturing": "Manufacturing", "EDUCATION": "Education", "mfg": "Manufacturing", "industrial": "Manufacturing", "Industrial": "Manufacturing", "MANUFACTURING": "Manufacturing", "Manufacxuring": "Manufacturing"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"snr engineer": "Senior Engineer", "Abmin": "Administrative Assistant", "engineering mgr": "Engineering Manager", "Admin Asst": "Administrative Assistant", "Eng Manager": "Engineering Manager", "Admin Assistant": "Administrative Assistant", "Senior Eng": "Senior Engineer", "Engineering Mgr": "Engineering Manager", "eng mgr": "Engineering Manager", "Addmin Assistant": "Administrative Assistant", "admin": "Administrative Assistant"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "(554)182-7349", "name": " David Diaz", "status": "won", "unit": "Lb", "city": "Osaka"}, {"phone": "537-106-9070", "name": "Carol Smith", "status": "WON", "unit": "inch", "city": "siena"}, {"phone": "5716529819", "name": "Frank Johnson   ", "status": "WON", "unit": "  in", "city": "Khanty-Mansiysk"}, {"phone": "(551)002-7863", "name": "Grace Ali ", "status": "Closed Won", "unit": "pounds  ", "city": "KHANTY-MANSIYSK"}, {"phone": "(536)282-0297", "name": "Omar Moore   ", "status": "IN PROGRESS", "unit": "inch", "city": "Osaka"}, {"phone": "500-991-2128", "name": "  Priya Petrov", "status": "In  Progress", "unit": "Kilometers", "city": "Siena"}, {"phone": "584-800-6388", "name": "  Carol Novak", "status": "Won", "unit": "  IN", "city": "sovetskiy  "}, {"phone": "560.653.6377", "name": "Mona  Smith", "status": "Closed Won", "unit": "inch", "city": "Sovetskiy"}, {"phone": "578-521-6583", "name": "Mona Moore", "status": "  in progerss", "unit": "kilameters", "city": "Seina"}, {"phone": "554-027-3911", "name": "Judy Cruz", "status": "won", "unit": "\"", "city": "osaka"}, {"phone": "(595)641-0282", "name": " Frank Novak", "status": "won", "unit": "Poundds", "city": "Osaka"}, {"phone": "5978476726", "name": "Priya Petrov", "status": "closed won   ", "unit": " Inch", "city": " KHANTY-MANSIYSK "}, {"phone": "595-111-7841", "name": " Sara Smith", "status": "in prgoress", "unit": " in", "city": "Sovetskiy"}, {"phone": "566.035.9814", "name": "  Ivan Park", "status": "Won", "unit": "Mi", "city": "OSAKA"}, {"phone": "570-045-0703", "name": "Omar Cruz", "status": "WON", "unit": "inch ", "city": "Sovetskiy"}, {"phone": "595.280.1697", "name": "   Karl Cruz", "status": "won   ", "unit": "lb", "city": "sovetskiy"}, {"phone": "532-185-8012", "name": "Carol Wong", "status": "in prgoress", "unit": "Km  ", "city": "Osaka"}, {"phone": "5544478114", "name": "Carol Lee", "status": " closed-won ", "unit": "   \"", "city": "KHANTYMANSIYCK"}, {"phone": "567.847.1337", "name": "Sara  Moore", "status": "  WON", "unit": " MILE ", "city": "KhantyMansiysk"}, {"phone": "524.068.2688", "name": "Eve  Wong", "status": "won", "unit": "   inches", "city": "Khanty-Mansiysk"}, {"phone": "(540)662-1871", "name": "Alice Fischer", "status": "   CLOSED-WON", "unit": " inches ", "city": " SOVETSKIY "}, {"phone": "597.588.6130", "name": "Karl Khan  ", "status": "wip", "unit": "mile", "city": "khanty-mansiysk"}, {"phone": "554-848-2702", "name": "Eve Reyes", "status": "Won", "unit": " Km", "city": "KhantyMansiysk"}, {"phone": "517.359.9916", "name": "Eve Wong", "status": "Closed-Won", "unit": "\"", "city": "Osaka"}, {"phone": "511.043.9598", "name": "Alice Reyes", "status": "won", "unit": "lb", "city": "Siena"}, {"phone": "5545761929", "name": "Lena Smith", "status": "won", "unit": "km ", "city": "Siena"}, {"phone": "(512)757-7854", "name": "David Brandt", "status": "Won", "unit": "lbs", "city": "KhantyMansiysk"}, {"phone": "(559)530-5206", "name": "   Heidi Khan", "status": "In Progress", "unit": "lbs", "city": "Khanty-Mansiysk"}, {"phone": "(540)884-6942", "name": "Judy Wong  ", "status": "WIP", "unit": "lb", "city": "  oaka"}, {"phone": "559.667.4031", "name": "   Judy Wong", "status": "wip", "unit": " pounds ", "city": " Sovetskiy "}, {"phone": "552.214.4419", "name": "Alice Cruz", "status": "  closed-won", "unit": "kilometers", "city": "Khanty-Mansiysk"}, {"phone": "557-733-3767", "name": "Priya Khan", "status": "IN PROGRESS", "unit": "Miles   ", "city": "siena"}, {"phone": "(517)393-3984", "name": "Ivan Smith", "status": "closed-wn", "unit": "LBS", "city": "siena"}, {"phone": "501-814-8605", "name": "David Lee", "status": "ongoing", "unit": "km", "city": " Osaka"}, {"phone": "501-307-0247", "name": "Omar Adams", "status": "Won", "unit": "mi", "city": "Sovxtskiy"}, {"phone": "553.863.3976", "name": "Priya Novak ", "status": "WIP", "unit": "MKLE", "city": "Sovetskiy"}, {"phone": "526-289-7304", "name": "Eve Moore", "status": "ongonig", "unit": "INCH", "city": "Siena"}, {"phone": "562-087-6740", "name": "Judy Ali", "status": "IN PROGRESS", "unit": "Lb", "city": "   Khanty-Mansiysk"}, {"phone": "(529)104-7133", "name": "Sara  Johnson", "status": "closed-won  ", "unit": "KM", "city": "Siena"}, {"phone": "540.551.9119", "name": "Eve Johnson", "status": "closed-won", "unit": "km", "city": "Sovetskiy"}, {"phone": "572-744-0643", "name": "Lena Diaz", "status": "Won", "unit": "miles  ", "city": "Siena"}, {"phone": "5990053695", "name": "Frank Novak", "status": "In Proress", "unit": "lb", "city": "Osaa"}, {"phone": "5783052799", "name": "Sara Smith", "status": "ongoing", "unit": "KILOMETER", "city": "Khanty-Mansiysk"}, {"phone": "(583)323-4635", "name": "Heidi Reyes", "status": " won ", "unit": " kilometers ", "city": "OSAKA"}, {"phone": "(503)798-1320", "name": "Lena Johnson", "status": "closed-won", "unit": "lb ", "city": " Siena "}, {"phone": "519.974.9430", "name": "Heidi Wong", "status": "won", "unit": "inches", "city": "Sovetskiy "}, {"phone": "(513)850-2726", "name": "Priya  Petrov", "status": "in progress", "unit": "pound", "city": "OSAKA"}, {"phone": "(516)396-9641", "name": "Eve Brandt", "status": "WIP", "unit": "mi", "city": "osaka"}, {"phone": "586.383.9873", "name": "Priya  Smith", "status": "   WON", "unit": "  Mi", "city": "OSAKA"}, {"phone": "538.223.8741", "name": "Judy Wong", "status": "Won", "unit": "MI", "city": "siena"}, {"phone": "502-495-7145", "name": "Omar Johnson", "status": "Closed Won", "unit": "inch", "city": " Khanty-Mansiysk "}, {"phone": "515-364-9716", "name": "   Lena Fischer", "status": " CLOSED WON", "unit": "MI", "city": "Sovetskiiy"}, {"phone": "5559426947", "name": "Eve Brandt", "status": " CLOESD-WON ", "unit": " inch", "city": "sovetskiy"}, {"phone": "554-152-4121", "name": "Lena Smith", "status": "wip", "unit": "km", "city": "osaka"}, {"phone": "5383731691", "name": "Mona Wong", "status": "CLOSED WON", "unit": " lb ", "city": "  Siena"}, {"phone": "5763693522", "name": "Ivan  Reyes", "status": "   ongoing", "unit": "in", "city": "Sovetskiy"}, {"phone": "543.464.3595", "name": "Grace Khan ", "status": "inprogress", "unit": "\"", "city": "Khanty-Mansiysk"}, {"phone": "585.619.0006", "name": "  Bob Wong", "status": "WON", "unit": "  Km", "city": "Sina   "}, {"phone": "554-027-3911", "name": "Judy Cruz", "status": "won", "unit": "\"", "city": "osaka"}, {"phone": "5990053695", "name": "Frank Novak", "status": "In Proress", "unit": "lb", "city": "Osaa"}], "clean": [{"phone": "(554) 182-7349", "name": "David Diaz", "status": "Won", "unit": "lb", "city": "Osaka"}, {"phone": "(537) 106-9070", "name": "Carol Smith", "status": "Won", "unit": "in", "city": "Siena"}, {"phone": "(571) 652-9819", "name": "Frank Johnson", "status": "Won", "unit": "in", "city": "Khanty-Mansiysk"}, {"phone": "(551) 002-7863", "name": "Grace Ali", "status": "Won", "unit": "lb", "city": "Khanty-Mansiysk"}, {"phone": "(536) 282-0297", "name": "Omar Moore", "status": "In Progress", "unit": "in", "city": "Osaka"}, {"phone": "(500) 991-2128", "name": "Priya Petrov", "status": "In Progress", "unit": "km", "city": "Siena"}, {"phone": "(584) 800-6388", "name": "Carol Novak", "status": "Won", "unit": "in", "city": "Sovetskiy"}, {"phone": "(560) 653-6377", "name": "Mona Smith", "status": "Won", "unit": "in", "city": "Sovetskiy"}, {"phone": "(578) 521-6583", "name": "Mona Moore", "status": "In Progress", "unit": "km", "city": "Siena"}, {"phone": "(554) 027-3911", "name": "Judy Cruz", "status": "Won", "unit": "in", "city": "Osaka"}, {"phone": "(595) 641-0282", "name": "Frank Novak", "status": "Won", "unit": "lb", "city": "Osaka"}, {"phone": "(597) 847-6726", "name": "Priya Petrov", "status": "Won", "unit": "in", "city": "Khanty-Mansiysk"}, {"phone": "(595) 111-7841", "name": "Sara Smith", "status": "In Progress", "unit": "in", "city": "Sovetskiy"}, {"phone": "(566) 035-9814", "name": "Ivan Park", "status": "Won", "unit": "mi", "city": "Osaka"}, {"phone": "(570) 045-0703", "name": "Omar Cruz", "status": "Won", "unit": "in", "city": "Sovetskiy"}, {"phone": "(595) 280-1697", "name": "Karl Cruz", "status": "Won", "unit": "lb", "city": "Sovetskiy"}, {"phone": "(532) 185-8012", "name": "Carol Wong", "status": "In Progress", "unit": "km", "city": "Osaka"}, {"phone": "(554) 447-8114", "name": "Carol Lee", "status": "Won", "unit": "in", "city": "Khanty-Mansiysk"}, {"phone": "(567) 847-1337", "name": "Sara Moore", "status": "Won", "unit": "mi", "city": "Khanty-Mansiysk"}, {"phone": "(524) 068-2688", "name": "Eve Wong", "status": "Won", "unit": "in", "city": "Khanty-Mansiysk"}, {"phone": "(540) 662-1871", "name": "Alice Fischer", "status": "Won", "unit": "in", "city": "Sovetskiy"}, {"phone": "(597) 588-6130", "name": "Karl Khan", "status": "In Progress", "unit": "mi", "city": "Khanty-Mansiysk"}, {"phone": "(554) 848-2702", "name": "Eve Reyes", "status": "Won", "unit": "km", "city": "Khanty-Mansiysk"}, {"phone": "(517) 359-9916", "name": "Eve Wong", "status": "Won", "unit": "in", "city": "Osaka"}, {"phone": "(511) 043-9598", "name": "Alice Reyes", "status": "Won", "unit": "lb", "city": "Siena"}, {"phone": "(554) 576-1929", "name": "Lena Smith", "status": "Won", "unit": "km", "city": "Siena"}, {"phone": "(512) 757-7854", "name": "David Brandt", "status": "Won", "unit": "lb", "city": "Khanty-Mansiysk"}, {"phone": "(559) 530-5206", "name": "Heidi Khan", "status": "In Progress", "unit": "lb", "city": "Khanty-Mansiysk"}, {"phone": "(540) 884-6942", "name": "Judy Wong", "status": "In Progress", "unit": "lb", "city": "Osaka"}, {"phone": "(559) 667-4031", "name": "Judy Wong", "status": "In Progress", "unit": "lb", "city": "Sovetskiy"}, {"phone": "(552) 214-4419", "name": "Alice Cruz", "status": "Won", "unit": "km", "city": "Khanty-Mansiysk"}, {"phone": "(557) 733-3767", "name": "Priya Khan", "status": "In Progress", "unit": "mi", "city": "Siena"}, {"phone": "(517) 393-3984", "name": "Ivan Smith", "status": "Won", "unit": "lb", "city": "Siena"}, {"phone": "(501) 814-8605", "name": "David Lee", "status": "In Progress", "unit": "km", "city": "Osaka"}, {"phone": "(501) 307-0247", "name": "Omar Adams", "status": "Won", "unit": "mi", "city": "Sovetskiy"}, {"phone": "(553) 863-3976", "name": "Priya Novak", "status": "In Progress", "unit": "mi", "city": "Sovetskiy"}, {"phone": "(526) 289-7304", "name": "Eve Moore", "status": "In Progress", "unit": "in", "city": "Siena"}, {"phone": "(562) 087-6740", "name": "Judy Ali", "status": "In Progress", "unit": "lb", "city": "Khanty-Mansiysk"}, {"phone": "(529) 104-7133", "name": "Sara Johnson", "status": "Won", "unit": "km", "city": "Siena"}, {"phone": "(540) 551-9119", "name": "Eve Johnson", "status": "Won", "unit": "km", "city": "Sovetskiy"}, {"phone": "(572) 744-0643", "name": "Lena Diaz", "status": "Won", "unit": "mi", "city": "Siena"}, {"phone": "(599) 005-3695", "name": "Frank Novak", "status": "In Progress", "unit": "lb", "city": "Osaka"}, {"phone": "(578) 305-2799", "name": "Sara Smith", "status": "In Progress", "unit": "km", "city": "Khanty-Mansiysk"}, {"phone": "(583) 323-4635", "name": "Heidi Reyes", "status": "Won", "unit": "km", "city": "Osaka"}, {"phone": "(503) 798-1320", "name": "Lena Johnson", "status": "Won", "unit": "lb", "city": "Siena"}, {"phone": "(519) 974-9430", "name": "Heidi Wong", "status": "Won", "unit": "in", "city": "Sovetskiy"}, {"phone": "(513) 850-2726", "name": "Priya Petrov", "status": "In Progress", "unit": "lb", "city": "Osaka"}, {"phone": "(516) 396-9641", "name": "Eve Brandt", "status": "In Progress", "unit": "mi", "city": "Osaka"}, {"phone": "(586) 383-9873", "name": "Priya Smith", "status": "Won", "unit": "mi", "city": "Osaka"}, {"phone": "(538) 223-8741", "name": "Judy Wong", "status": "Won", "unit": "mi", "city": "Siena"}, {"phone": "(502) 495-7145", "name": "Omar Johnson", "status": "Won", "unit": "in", "city": "Khanty-Mansiysk"}, {"phone": "(515) 364-9716", "name": "Lena Fischer", "status": "Won", "unit": "mi", "city": "Sovetskiy"}, {"phone": "(555) 942-6947", "name": "Eve Brandt", "status": "Won", "unit": "in", "city": "Sovetskiy"}, {"phone": "(554) 152-4121", "name": "Lena Smith", "status": "In Progress", "unit": "km", "city": "Osaka"}, {"phone": "(538) 373-1691", "name": "Mona Wong", "status": "Won", "unit": "lb", "city": "Siena"}, {"phone": "(576) 369-3522", "name": "Ivan Reyes", "status": "In Progress", "unit": "in", "city": "Sovetskiy"}, {"phone": "(543) 464-3595", "name": "Grace Khan", "status": "In Progress", "unit": "in", "city": "Khanty-Mansiysk"}, {"phone": "(585) 619-0006", "name": "Bob Wong", "status": "Won", "unit": "km", "city": "Siena"}], "dirty_cols": ["phone", "name", "status", "unit", "city"], "clean_cols": ["phone", "name", "status", "unit", "city"], "plan": {"dataset_summary": "60 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"won": "Won", "WON": "Won", "Closed Won": "Won", "IN PROGRESS": "In Progress", "in progerss": "In Progress", "closed won": "Won", "in prgoress": "In Progress", "closed-won": "Won", "CLOSED-WON": "Won", "wip": "In Progress", "Closed-Won": "Won", "WIP": "In Progress", "closed-wn": "Won", "ongoing": "In Progress", "ongonig": "In Progress", "In Proress": "In Progress", "in progress": "In Progress", "CLOSED WON": "Won", "CLOESD-WON": "Won", "inprogress": "In Progress"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Lb": "lb", "inch": "in", "pounds": "lb", "Kilometers": "km", "IN": "in", "kilameters": "km", "\"": "in", "Poundds": "lb", "Inch": "in", "Mi": "mi", "Km": "km", "MILE": "mi", "inches": "in", "mile": "mi", "lbs": "lb", "kilometers": "km", "Miles": "mi", "LBS": "lb", "MKLE": "mi", "INCH": "in", "KM": "km", "miles": "mi", "KILOMETER": "km", "pound": "lb", "MI": "mi"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"siena": "Siena", "KHANTY-MANSIYSK": "Khanty-Mansiysk", "sovetskiy": "Sovetskiy", "Seina": "Siena", "osaka": "Osaka", "OSAKA": "Osaka", "KHANTYMANSIYCK": "Khanty-Mansiysk", "KhantyMansiysk": "Khanty-Mansiysk", "SOVETSKIY": "Sovetskiy", "khanty-mansiysk": "Khanty-Mansiysk", "oaka": "Osaka", "Sovxtskiy": "Sovetskiy", "Osaa": "Osaka", "Sovetskiiy": "Sovetskiy", "Sina": "Siena"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "0", "rate": "77.2%", "phone": "559.066.1747", "status": "fee trial", "company": "Acme Inc", "unnamed": ""}, {"is_active": "1", "rate": "95.6%", "phone": "582.127.1796", "status": "TRIAL", "company": " Hooli ", "unnamed": ""}, {"is_active": "N", "rate": "11.0%", "phone": "(597)463-9982", "status": "trial", "company": "Hooli", "unnamed": ""}, {"is_active": "F", "rate": "8.7%", "phone": "(502)520-4325", "status": "paused", "company": "   Vehement", "unnamed": ""}, {"is_active": "TRUE", "rate": "70.8%", "phone": "510.199.8270", "status": "paused", "company": "Globex", "unnamed": ""}, {"is_active": "T", "rate": "19.0%", "phone": "5777769953", "status": "TRIAL", "company": "Soylent Corp ", "unnamed": ""}, {"is_active": "F", "rate": "32.8%", "phone": "5207794933", "status": "on hold", "company": " Globex", "unnamed": ""}, {"is_active": "true", "rate": "84.8%", "phone": "(573)066-2527", "status": "Paused", "company": "Hooli", "unnamed": ""}, {"is_active": "FALSE", "rate": "71.1%", "phone": "542-564-3038", "status": "Trial", "company": "   Acme Inc", "unnamed": ""}, {"is_active": "true", "rate": "78.8%", "phone": "(596)315-0480", "status": "trial", "company": "Umbrella", "unnamed": ""}], "clean": [{"is_active": false, "rate": 0.772, "phone": "(559) 066-1747", "status": "Trial", "company": "Acme Inc"}, {"is_active": true, "rate": 0.956, "phone": "(582) 127-1796", "status": "Trial", "company": "Hooli"}, {"is_active": false, "rate": 0.11, "phone": "(597) 463-9982", "status": "Trial", "company": "Hooli"}, {"is_active": false, "rate": 0.087, "phone": "(502) 520-4325", "status": "Paused", "company": "Vehement"}, {"is_active": true, "rate": 0.708, "phone": "(510) 199-8270", "status": "Paused", "company": "Globex"}, {"is_active": true, "rate": 0.19, "phone": "(577) 776-9953", "status": "Trial", "company": "Soylent Corp"}, {"is_active": false, "rate": 0.32799999999999996, "phone": "(520) 779-4933", "status": "Paused", "company": "Globex"}, {"is_active": true, "rate": 0.848, "phone": "(573) 066-2527", "status": "Paused", "company": "Hooli"}, {"is_active": false, "rate": 0.711, "phone": "(542) 564-3038", "status": "Trial", "company": "Acme Inc"}, {"is_active": true, "rate": 0.7879999999999999, "phone": "(596) 315-0480", "status": "Trial", "company": "Umbrella"}], "dirty_cols": ["is_active", "rate", "phone", "status", "company", "unnamed"], "clean_cols": ["is_active", "rate", "phone", "status", "company"], "plan": {"dataset_summary": "10 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"fee trial": "Trial", "TRIAL": "Trial", "trial": "Trial", "paused": "Paused", "on hold": "Paused"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"status": " closey lost", "company": "Stark Industries", "country": "BV", "city": "Nowshera Kalan", "email": " pnphlaf@example.com ", "unit": "mile", "is_active": "Y", "notes2": ""}, {"status": " Lost ", "company": "Umbrella", "country": "Bangladesh", "city": "Nowshera Kalan", "email": " NKEI@EXAMPLE.COM ", "unit": "mi", "is_active": "N", "notes2": ""}, {"status": " In Progress", "company": "Wonka  Co", "country": "BVT", "city": "villasis", "email": "lnbk@mail.com", "unit": "kilometers", "is_active": "FALSE", "notes2": ""}, {"status": "Closed  Lost", "company": "Wonka  Co", "country": "Bouvet Island", "city": "Villassis", "email": "PNEF@CORP.IO  ", "unit": " Km ", "is_active": "N", "notes2": ""}, {"status": "Closed Lost", "company": "Hooli", "country": "PeopleS' Republic Of Bangladesh", "city": "Nowshera Kalan", "email": "KDHK@EXAMPLE.COM", "unit": "mils", "is_active": "TRUE", "notes2": ""}, {"status": "Wip", "company": "Acme Inc", "country": "BGD", "city": "Nowsshera Kalan", "email": "JEMD@EXAMPLE.COM", "unit": "Km", "is_active": "F", "notes2": ""}, {"status": "in-progress", "company": "Acme Inc   ", "country": "bv", "city": "Villasis", "email": "  PPBH@TEST.ORG", "unit": "mile", "is_active": "No", "notes2": ""}, {"status": " ongoing ", "company": "Stark Industries  ", "country": "   Bouvet Island", "city": "nowshera kalan", "email": "fkok@corp.io", "unit": "  kilometer", "is_active": "Yes", "notes2": ""}, {"status": "CLOSED LOST  ", "company": " Acme Inc", "country": "BV", "city": "Nowshera Kalan", "email": "eppgig@corp.io   ", "unit": "milse", "is_active": "T", "notes2": ""}, {"status": "lost", "company": "Umbrella ", "country": "Bd", "city": "nowshera  kalan", "email": " ileplan@mail.com", "unit": "in", "is_active": "Y", "notes2": ""}, {"status": "WIP  ", "company": "Globex", "country": "BVT", "city": " nowshera kalan", "email": "OAAEP@EXAMPLE.COM", "unit": " KILOMETERS ", "is_active": "Y", "notes2": ""}, {"status": "WIP", "company": "Hooli", "country": "BD", "city": "Villasis", "email": "CENCGJN@CORP.IO", "unit": "Inches", "is_active": "false", "notes2": ""}, {"status": "Lost", "company": "Cyberdyne", "country": "BV   ", "city": "villasis", "email": "defm@test.org", "unit": " Km", "is_active": "F", "notes2": ""}, {"status": "Ongoing", "company": " Umbrella", "country": "bvt", "city": "nowshera kalan", "email": "ogedo@test.org", "unit": "mi", "is_active": "T", "notes2": ""}, {"status": "closed lost", "company": "Stark  Industries", "country": "Bangladesh", "city": "  NOWSHERA KALAN", "email": "  LAOL@TEST.ORG", "unit": "Km", "is_active": "1", "notes2": ""}, {"status": "  lost", "company": "   Hooli", "country": "BGD", "city": "Nowshera aKlan", "email": "aoahad@test.org", "unit": "INCHES", "is_active": "true", "notes2": ""}, {"status": "lost", "company": "Globex", "country": "BGD", "city": "Nowshera Kalan", "email": "dddc@corp.io", "unit": "MILE", "is_active": "T", "notes2": ""}, {"status": "Closed Lost", "company": "Stark  Industries", "country": " PeopleS Republic Of Bangltdesh", "city": "nowshera kaln", "email": "AKNJPEI@MAIL.COM", "unit": " \"", "is_active": "No", "notes2": ""}, {"status": "ongoing", "company": "Acme Inc  ", "country": "PEOPLE'S REPUBLIC OF BANGLADESH", "city": "Villasis", "email": "EBGNEE@CORP.IO", "unit": "Inch", "is_active": "Y", "notes2": ""}, {"status": "Wip", "company": "Soylent Corp", "country": "BGD", "city": "Villasis", "email": "MEGMOEP@MAIL.COM", "unit": "Km", "is_active": "0", "notes2": ""}, {"status": "In  Progress", "company": "Soylent Corp   ", "country": "People'S  Republic Of Bangladesh", "city": "Nowshera Kalan", "email": "CEPDDB@MAIL.COM  ", "unit": "kilometers", "is_active": "Y", "notes2": ""}, {"status": " lot ", "company": "Hooli  ", "country": "BVT ", "city": "Villasis   ", "email": "jjbce@mail.com", "unit": " mi ", "is_active": "N", "notes2": ""}, {"status": "lost", "company": " Initech ", "country": "BGD", "city": "Nowshera  Kalan", "email": "fphpd@mail.com", "unit": " miles ", "is_active": "Y", "notes2": ""}, {"status": "in progress", "company": "Initech", "country": "BV", "city": "   villasis", "email": "IHAKHK@TEST.ORG   ", "unit": "km", "is_active": "No", "notes2": ""}, {"status": "  In-Progress", "company": "Umbrella", "country": " BV", "city": "nowshera kalan  ", "email": "lklpdbi@example.com", "unit": "mi", "is_active": "TRUE", "notes2": ""}, {"status": "  Closed Lost", "company": "Stark Industries", "country": "Bangladesh", "city": "Nowshera  Kalan", "email": "AMHEP@TEST.ORG   ", "unit": "miles", "is_active": "No", "notes2": ""}, {"status": "In Prgress", "company": "Cyberdyne", "country": "bv", "city": "NOWSHERA KALAN", "email": "HLMCJO@TEST.ORG", "unit": "KILOMETERS", "is_active": "Yes", "notes2": ""}, {"status": "Lost", "company": "Acme Inc  ", "country": "   bvt", "city": "Villasis", "email": "  BGLJO@EXAMPLE.COM", "unit": "kilometer", "is_active": "T", "notes2": ""}, {"status": "Lost   ", "company": "Soylent Corp", "country": " BGD", "city": " VILLASIS", "email": "ncbo@mail.com", "unit": "\"  ", "is_active": "T", "notes2": ""}, {"status": "Closed Lost  ", "company": "Initech", "country": "BGD", "city": "Nowshera Kalan", "email": "HPKK@EXAMPLE.COM  ", "unit": "KM", "is_active": "T", "notes2": ""}, {"status": "Closed Lost", "company": "Stark Industries   ", "country": "Peole's Republic of Bangladesh", "city": "  villasis", "email": " GNDDGOA@EXAMPLE.COM", "unit": "Mile", "is_active": "FALSE", "notes2": ""}, {"status": "Closed  Lost", "company": "Acme Inc", "country": "BVT   ", "city": "Villasis", "email": "jfhpfj@corp.io", "unit": "mile", "is_active": "0", "notes2": ""}, {"status": "in progress", "company": "Wonka Co", "country": "BV", "city": "Nowshera Kalan", "email": "DOJFN@EXAMPLE.COM", "unit": "inch", "is_active": "0", "notes2": ""}, {"status": "lost", "company": "Umbrella", "country": "People's Republic of Bnagladesh ", "city": "Nowshera Kalan", "email": " onih@mail.com ", "unit": "inches", "is_active": "N", "notes2": ""}, {"status": "ongoing", "company": " Vehement ", "country": "BGD", "city": "Villasis", "email": "nplml@test.org  ", "unit": "mle", "is_active": "F", "notes2": ""}, {"status": "Closed Lost", "company": "Wonka  Co", "country": " Bv ", "city": "Nkwshera Kalan", "email": "JMGFIL@CORP.IO", "unit": "   km", "is_active": "false", "notes2": ""}, {"status": "In  Progress", "company": "Soylent Corp  ", "country": "bouvet island", "city": "Villasis", "email": "lmagdbi@corp.io", "unit": "KILOMETER", "is_active": "F", "notes2": ""}, {"status": "In Progress", "company": "Wonka Co   ", "country": "   BOUVET ISLAND", "city": "Villasis", "email": "oneb@example.com", "unit": "inches", "is_active": "T", "notes2": ""}, {"status": "in progress", "company": "Soylent Corp", "country": "   Bouvet Island", "city": "Nowshera Kalan", "email": "JGAOF@CORP.IO", "unit": "miles   ", "is_active": "true", "notes2": ""}, {"status": "Closed Lost", "company": "Vehement", "country": "bgd", "city": "Nowshera Kalan", "email": "OELLLC@TEST.ORG", "unit": "Mi", "is_active": "T", "notes2": ""}, {"status": "ongoing", "company": "Initech", "country": "People's Republic of Bangladesh", "city": "nowshera  kalan", "email": "nheimaf@example.com", "unit": "mi", "is_active": "N", "notes2": ""}, {"status": " lost ", "company": "Umbrella", "country": " bgd ", "city": "Villasis", "email": "nbnpldg@test.org", "unit": " inches ", "is_active": "Y", "notes2": ""}, {"status": "ongoing", "company": "Soylent Corp", "country": "bangladesh", "city": "NOWSHERA KALAN", "email": "lmfg@mail.com", "unit": "mi", "is_active": "T", "notes2": ""}, {"status": "CLOSED LOST", "company": "  Cyberdyne", "country": "BV", "city": "   NOWSHERA KALAN", "email": "foaacg@corp.io  ", "unit": "kilmoeter", "is_active": "true", "notes2": ""}, {"status": "Lost ", "company": "Acme  Inc", "country": "bv", "city": "VILLASIS", "email": " GBMEP@CORP.IO", "unit": "miles", "is_active": "No", "notes2": ""}, {"status": "Closed Lost", "company": "Wonka Co", "country": "BD", "city": "Villasis", "email": "KEHI@CORP.IO", "unit": "mi", "is_active": "Y", "notes2": ""}, {"status": "Closed Lost", "company": "Soylent Corp", "country": "BGD   ", "city": "Nowshera Kalan", "email": "NLGICGI@CORP.IO ", "unit": "km ", "is_active": "Y", "notes2": ""}, {"status": "Lost", "company": "Acme Inc", "country": "BVT", "city": "NOWSHERA KALAN", "email": "CMHHEIF@MAIL.COM", "unit": "mi", "is_active": "0", "notes2": ""}, {"status": "in progress", "company": "Wonka Co", "country": "bouvet island", "city": "VILLASIS", "email": " IIGHM@EXAMPLE.COM ", "unit": "inch", "is_active": "Yes", "notes2": ""}, {"status": "lost", "company": "Soylent Corp", "country": "BVT", "city": "villasis", "email": "ALIFLC@MAIL.COM", "unit": "inch", "is_active": "FALSE", "notes2": ""}, {"status": " Lost", "company": "Umbrella", "country": "bangladesh", "city": "Villasis", "email": " BBAFK@TEST.ORG", "unit": "mi", "is_active": "T", "notes2": ""}, {"status": "lost", "company": "Stark  Industries", "country": " bd ", "city": "Villasis", "email": "oiecc@test.org", "unit": "KM", "is_active": "Y", "notes2": ""}, {"status": "Lost", "company": "   Cyberdyne", "country": "BOUVET ISLND  ", "city": "  Nowshera Kalan", "email": "  nhenkd@test.org", "unit": "mile", "is_active": "1", "notes2": ""}, {"status": " Lost", "company": "Hooli", "country": "Bouvet IIsland", "city": "Villasis", "email": " ECNDKO@TEST.ORG", "unit": "mile", "is_active": "0", "notes2": ""}], "clean": [{"status": "Lost", "company": "Stark Industries", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "pnphlaf@example.com", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Umbrella", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "nkei@example.com", "unit": "mi", "is_active": false}, {"status": "In Progress", "company": "Wonka Co", "country": "Bouvet Island", "city": "Villasis", "email": "lnbk@mail.com", "unit": "km", "is_active": false}, {"status": "Lost", "company": "Wonka Co", "country": "Bouvet Island", "city": "Villasis", "email": "pnef@corp.io", "unit": "km", "is_active": false}, {"status": "Lost", "company": "Hooli", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "kdhk@example.com", "unit": "mi", "is_active": true}, {"status": "In Progress", "company": "Acme Inc", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "jemd@example.com", "unit": "km", "is_active": false}, {"status": "In Progress", "company": "Acme Inc", "country": "Bouvet Island", "city": "Villasis", "email": "ppbh@test.org", "unit": "mi", "is_active": false}, {"status": "In Progress", "company": "Stark Industries", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "fkok@corp.io", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Acme Inc", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "eppgig@corp.io", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Umbrella", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "ileplan@mail.com", "unit": "in", "is_active": true}, {"status": "In Progress", "company": "Globex", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "oaaep@example.com", "unit": "km", "is_active": true}, {"status": "In Progress", "company": "Hooli", "country": "Bangladesh", "city": "Villasis", "email": "cencgjn@corp.io", "unit": "in", "is_active": false}, {"status": "Lost", "company": "Cyberdyne", "country": "Bouvet Island", "city": "Villasis", "email": "defm@test.org", "unit": "km", "is_active": false}, {"status": "In Progress", "company": "Umbrella", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "ogedo@test.org", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Stark Industries", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "laol@test.org", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Hooli", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "aoahad@test.org", "unit": "in", "is_active": true}, {"status": "Lost", "company": "Globex", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "dddc@corp.io", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Stark Industries", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "aknjpei@mail.com", "unit": "in", "is_active": false}, {"status": "In Progress", "company": "Acme Inc", "country": "Bangladesh", "city": "Villasis", "email": "ebgnee@corp.io", "unit": "in", "is_active": true}, {"status": "In Progress", "company": "Soylent Corp", "country": "Bangladesh", "city": "Villasis", "email": "megmoep@mail.com", "unit": "km", "is_active": false}, {"status": "In Progress", "company": "Soylent Corp", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "cepddb@mail.com", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Hooli", "country": "Bouvet Island", "city": "Villasis", "email": "jjbce@mail.com", "unit": "mi", "is_active": false}, {"status": "Lost", "company": "Initech", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "fphpd@mail.com", "unit": "mi", "is_active": true}, {"status": "In Progress", "company": "Initech", "country": "Bouvet Island", "city": "Villasis", "email": "ihakhk@test.org", "unit": "km", "is_active": false}, {"status": "In Progress", "company": "Umbrella", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "lklpdbi@example.com", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Stark Industries", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "amhep@test.org", "unit": "mi", "is_active": false}, {"status": "In Progress", "company": "Cyberdyne", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "hlmcjo@test.org", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Acme Inc", "country": "Bouvet Island", "city": "Villasis", "email": "bgljo@example.com", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Soylent Corp", "country": "Bangladesh", "city": "Villasis", "email": "ncbo@mail.com", "unit": "in", "is_active": true}, {"status": "Lost", "company": "Initech", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "hpkk@example.com", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Stark Industries", "country": "Bangladesh", "city": "Villasis", "email": "gnddgoa@example.com", "unit": "mi", "is_active": false}, {"status": "Lost", "company": "Acme Inc", "country": "Bouvet Island", "city": "Villasis", "email": "jfhpfj@corp.io", "unit": "mi", "is_active": false}, {"status": "In Progress", "company": "Wonka Co", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "dojfn@example.com", "unit": "in", "is_active": false}, {"status": "Lost", "company": "Umbrella", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "onih@mail.com", "unit": "in", "is_active": false}, {"status": "In Progress", "company": "Vehement", "country": "Bangladesh", "city": "Villasis", "email": "nplml@test.org", "unit": "mi", "is_active": false}, {"status": "Lost", "company": "Wonka Co", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "jmgfil@corp.io", "unit": "km", "is_active": false}, {"status": "In Progress", "company": "Soylent Corp", "country": "Bouvet Island", "city": "Villasis", "email": "lmagdbi@corp.io", "unit": "km", "is_active": false}, {"status": "In Progress", "company": "Wonka Co", "country": "Bouvet Island", "city": "Villasis", "email": "oneb@example.com", "unit": "in", "is_active": true}, {"status": "In Progress", "company": "Soylent Corp", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "jgaof@corp.io", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Vehement", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "oelllc@test.org", "unit": "mi", "is_active": true}, {"status": "In Progress", "company": "Initech", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "nheimaf@example.com", "unit": "mi", "is_active": false}, {"status": "Lost", "company": "Umbrella", "country": "Bangladesh", "city": "Villasis", "email": "nbnpldg@test.org", "unit": "in", "is_active": true}, {"status": "In Progress", "company": "Soylent Corp", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "lmfg@mail.com", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Cyberdyne", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "foaacg@corp.io", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Acme Inc", "country": "Bouvet Island", "city": "Villasis", "email": "gbmep@corp.io", "unit": "mi", "is_active": false}, {"status": "Lost", "company": "Wonka Co", "country": "Bangladesh", "city": "Villasis", "email": "kehi@corp.io", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Soylent Corp", "country": "Bangladesh", "city": "Nowshera Kalan", "email": "nlgicgi@corp.io", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Acme Inc", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "cmhheif@mail.com", "unit": "mi", "is_active": false}, {"status": "In Progress", "company": "Wonka Co", "country": "Bouvet Island", "city": "Villasis", "email": "iighm@example.com", "unit": "in", "is_active": true}, {"status": "Lost", "company": "Soylent Corp", "country": "Bouvet Island", "city": "Villasis", "email": "aliflc@mail.com", "unit": "in", "is_active": false}, {"status": "Lost", "company": "Umbrella", "country": "Bangladesh", "city": "Villasis", "email": "bbafk@test.org", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Stark Industries", "country": "Bangladesh", "city": "Villasis", "email": "oiecc@test.org", "unit": "km", "is_active": true}, {"status": "Lost", "company": "Cyberdyne", "country": "Bouvet Island", "city": "Nowshera Kalan", "email": "nhenkd@test.org", "unit": "mi", "is_active": true}, {"status": "Lost", "company": "Hooli", "country": "Bouvet Island", "city": "Villasis", "email": "ecndko@test.org", "unit": "mi", "is_active": false}], "dirty_cols": ["status", "company", "country", "city", "email", "unit", "is_active", "notes2"], "clean_cols": ["status", "company", "country", "city", "email", "unit", "is_active"], "plan": {"dataset_summary": "54 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"closey lost": "Lost", "Closed Lost": "Lost", "Wip": "In Progress", "in-progress": "In Progress", "ongoing": "In Progress", "CLOSED LOST": "Lost", "lost": "Lost", "WIP": "In Progress", "Ongoing": "In Progress", "closed lost": "Lost", "lot": "Lost", "in progress": "In Progress", "In-Progress": "In Progress", "In Prgress": "In Progress"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BV": "Bouvet Island", "BVT": "Bouvet Island", "PeopleS' Republic Of Bangladesh": "Bangladesh", "BGD": "Bangladesh", "bv": "Bouvet Island", "Bd": "Bangladesh", "BD": "Bangladesh", "bvt": "Bouvet Island", "PeopleS Republic Of Bangltdesh": "Bangladesh", "PEOPLE'S REPUBLIC OF BANGLADESH": "Bangladesh", "People'S Republic Of Bangladesh": "Bangladesh", "Peole's Republic of Bangladesh": "Bangladesh", "People's Republic of Bnagladesh": "Bangladesh", "Bv": "Bouvet Island", "bouvet island": "Bouvet Island", "BOUVET ISLAND": "Bouvet Island", "bgd": "Bangladesh", "People's Republic of Bangladesh": "Bangladesh", "bangladesh": "Bangladesh", "bd": "Bangladesh", "BOUVET ISLND": "Bouvet Island", "Bouvet IIsland": "Bouvet Island"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"villasis": "Villasis", "Villassis": "Villasis", "Nowsshera Kalan": "Nowshera Kalan", "nowshera kalan": "Nowshera Kalan", "NOWSHERA KALAN": "Nowshera Kalan", "Nowshera aKlan": "Nowshera Kalan", "nowshera kaln": "Nowshera Kalan", "VILLASIS": "Villasis", "Nkwshera Kalan": "Nowshera Kalan"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mile": "mi", "kilometers": "km", "Km": "km", "mils": "mi", "kilometer": "km", "milse": "mi", "KILOMETERS": "km", "Inches": "in", "INCHES": "in", "MILE": "mi", "\"": "in", "Inch": "in", "miles": "mi", "KM": "km", "Mile": "mi", "inch": "in", "inches": "in", "mle": "mi", "KILOMETER": "km", "Mi": "mi", "kilmoeter": "km"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"signup_date": "2023-11-14", "status": "   high", "currency": "huf", "company": "Cyberdyne", "is_active": "Yes", "industry": "MFG", "notes2": ""}, {"signup_date": "44943", "status": "h", "currency": "gibraltar pound ", "company": "Soylent  Corp", "is_active": "0", "industry": "Financial Services", "notes2": ""}, {"signup_date": "45052", "status": "High", "currency": "GOURDE", "company": "Initech", "is_active": "TRUE", "industry": "manufacturing", "notes2": ""}, {"signup_date": "45001", "status": " P1 ", "currency": "gip", "company": "Umbrella  ", "is_active": "1", "industry": "INDUSTRIAL ", "notes2": ""}, {"signup_date": "6 Feb 2023", "status": "MED", "currency": "Gibraltar Pound", "company": "Globex", "is_active": "1", "industry": "Industrial", "notes2": ""}, {"signup_date": "2023-05-13", "status": "P2", "currency": "  GNF", "company": " Vehement ", "is_active": "TRUE", "industry": "Banking", "notes2": ""}, {"signup_date": "44949", "status": "Medium", "currency": "Guinean Franc", "company": "Initech", "is_active": "Yes", "industry": "Finance", "notes2": ""}, {"signup_date": "2023-02-07", "status": "p2", "currency": "HUF", "company": "Cyberdyne", "is_active": "Yes", "industry": "FINANCIAL  SERRVICES", "notes2": ""}, {"signup_date": "44928", "status": " H ", "currency": "guinean franc", "company": " Globex ", "is_active": "T", "industry": "FINANCE", "notes2": ""}, {"signup_date": "2023-04-22", "status": "medium", "currency": "Guinean Franc", "company": "Soylent  Corp", "is_active": "FALSE", "industry": " Fintech ", "notes2": ""}, {"signup_date": "44981", "status": "P1", "currency": "HTG", "company": "   Umbrella", "is_active": "TRUE", "industry": "Manufacturing", "notes2": ""}, {"signup_date": "45250", "status": "H", "currency": "htg", "company": "Soylent Corp", "is_active": "0", "industry": "manufacturing", "notes2": ""}, {"signup_date": "44986", "status": "H  ", "currency": " gnf ", "company": "Hooli", "is_active": "false", "industry": "MANUFACTURING", "notes2": ""}, {"signup_date": "2023-07-19", "status": "HIGH   ", "currency": "gibraltar  pound", "company": " Globex", "is_active": "true", "industry": " FinServ ", "notes2": ""}, {"signup_date": "45213", "status": "P2", "currency": "Gip", "company": "Globex   ", "is_active": "FALSE", "industry": "finance", "notes2": ""}, {"signup_date": "18 Sep 2023", "status": "high", "currency": "gibraltar pound", "company": "Soylent  Corp", "is_active": "TRUE", "industry": "Finserv", "notes2": ""}, {"signup_date": "2023-10-07", "status": "MED", "currency": "Goure", "company": "Globex", "is_active": "0", "industry": "finserv", "notes2": ""}, {"signup_date": "45052", "status": "High", "currency": "GOURDE", "company": "Initech", "is_active": "TRUE", "industry": "manufacturing", "notes2": ""}, {"signup_date": "44943", "status": "h", "currency": "gibraltar pound ", "company": "Soylent  Corp", "is_active": "0", "industry": "Financial Services", "notes2": ""}], "clean": [{"signup_date": "2023-11-14", "status": "High", "currency": "HUF", "company": "Cyberdyne", "is_active": true, "industry": "Manufacturing"}, {"signup_date": "2023-01-17", "status": "High", "currency": "GIP", "company": "Soylent Corp", "is_active": false, "industry": "Financial Services"}, {"signup_date": "2023-05-06", "status": "High", "currency": "HTG", "company": "Initech", "is_active": true, "industry": "Manufacturing"}, {"signup_date": "2023-03-16", "status": "High", "currency": "GIP", "company": "Umbrella", "is_active": true, "industry": "Manufacturing"}, {"signup_date": "2023-02-06", "status": "Medium", "currency": "GIP", "company": "Globex", "is_active": true, "industry": "Manufacturing"}, {"signup_date": "2023-05-13", "status": "Medium", "currency": "GNF", "company": "Vehement", "is_active": true, "industry": "Financial Services"}, {"signup_date": "2023-01-23", "status": "Medium", "currency": "GNF", "company": "Initech", "is_active": true, "industry": "Financial Services"}, {"signup_date": "2023-02-07", "status": "Medium", "currency": "HUF", "company": "Cyberdyne", "is_active": true, "industry": "Financial Services"}, {"signup_date": "2023-01-02", "status": "High", "currency": "GNF", "company": "Globex", "is_active": true, "industry": "Financial Services"}, {"signup_date": "2023-04-22", "status": "Medium", "currency": "GNF", "company": "Soylent Corp", "is_active": false, "industry": "Financial Services"}, {"signup_date": "2023-02-24", "status": "High", "currency": "HTG", "company": "Umbrella", "is_active": true, "industry": "Manufacturing"}, {"signup_date": "2023-11-20", "status": "High", "currency": "HTG", "company": "Soylent Corp", "is_active": false, "industry": "Manufacturing"}, {"signup_date": "2023-03-01", "status": "High", "currency": "GNF", "company": "Hooli", "is_active": false, "industry": "Manufacturing"}, {"signup_date": "2023-07-19", "status": "High", "currency": "GIP", "company": "Globex", "is_active": true, "industry": "Financial Services"}, {"signup_date": "2023-10-14", "status": "Medium", "currency": "GIP", "company": "Globex", "is_active": false, "industry": "Financial Services"}, {"signup_date": "2023-09-18", "status": "High", "currency": "GIP", "company": "Soylent Corp", "is_active": true, "industry": "Financial Services"}, {"signup_date": "2023-10-07", "status": "Medium", "currency": "HTG", "company": "Globex", "is_active": false, "industry": "Financial Services"}], "dirty_cols": ["signup_date", "status", "currency", "company", "is_active", "industry", "notes2"], "clean_cols": ["signup_date", "status", "currency", "company", "is_active", "industry"], "plan": {"dataset_summary": "19 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"high": "High", "h": "High", "P1": "High", "MED": "Medium", "P2": "Medium", "p2": "Medium", "H": "High", "medium": "Medium", "HIGH": "High"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"huf": "HUF", "gibraltar pound": "GIP", "GOURDE": "HTG", "gip": "GIP", "Gibraltar Pound": "GIP", "Guinean Franc": "GNF", "guinean franc": "GNF", "htg": "HTG", "gnf": "GNF", "Gip": "GIP", "Goure": "HTG"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MFG": "Manufacturing", "manufacturing": "Manufacturing", "INDUSTRIAL": "Manufacturing", "Industrial": "Manufacturing", "Banking": "Financial Services", "Finance": "Financial Services", "FINANCIAL SERRVICES": "Financial Services", "FINANCE": "Financial Services", "Fintech": "Financial Services", "MANUFACTURING": "Manufacturing", "FinServ": "Financial Services", "finance": "Financial Services", "Finserv": "Financial Services", "finserv": "Financial Services"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"company": "  Vehement", "state": "VT", "job_title": "admin assistant"}, {"company": " Globex ", "state": "Maine", "job_title": "CTO"}, {"company": "Vehement", "state": "Gorgia", "job_title": "ADIN"}, {"company": "Cyberdyne", "state": " Maine", "job_title": "Admin Asst"}, {"company": "Cyberdyne", "state": "  WY", "job_title": "ADMINISTRATIVE ASSISTANT"}, {"company": " Initech", "state": "GA", "job_title": "rep"}, {"company": "Soylent Corp", "state": "VERMONT", "job_title": "Admin Assistant"}, {"company": "Vehement", "state": "vt", "job_title": "Chief Technology Officer"}, {"company": "Initech", "state": "new hampshire", "job_title": "Adin"}, {"company": "Cyberdyne ", "state": " WY ", "job_title": "  rep"}, {"company": "Hooli", "state": "MAINE", "job_title": "ADMIN  ASSISTANT"}, {"company": "Soylent Corp", "state": " WY", "job_title": "  Admin Asst"}, {"company": "Cyberdyne", "state": "Veamont", "job_title": "Sales Rep"}, {"company": " Umbrella ", "state": "Me", "job_title": " Rep "}, {"company": "Initech", "state": "new hampshire", "job_title": "Adin"}, {"company": "", "state": "", "job_title": ""}], "clean": [{"company": "Vehement", "state": "Vermont", "job_title": "Administrative Assistant"}, {"company": "Globex", "state": "Maine", "job_title": "Chief Technology Officer"}, {"company": "Vehement", "state": "Georgia", "job_title": "Administrative Assistant"}, {"company": "Cyberdyne", "state": "Maine", "job_title": "Administrative Assistant"}, {"company": "Cyberdyne", "state": "Wyoming", "job_title": "Administrative Assistant"}, {"company": "Initech", "state": "Georgia", "job_title": "Sales Representative"}, {"company": "Soylent Corp", "state": "Vermont", "job_title": "Administrative Assistant"}, {"company": "Vehement", "state": "Vermont", "job_title": "Chief Technology Officer"}, {"company": "Initech", "state": "New Hampshire", "job_title": "Administrative Assistant"}, {"company": "Cyberdyne", "state": "Wyoming", "job_title": "Sales Representative"}, {"company": "Hooli", "state": "Maine", "job_title": "Administrative Assistant"}, {"company": "Soylent Corp", "state": "Wyoming", "job_title": "Administrative Assistant"}, {"company": "Cyberdyne", "state": "Vermont", "job_title": "Sales Representative"}, {"company": "Umbrella", "state": "Maine", "job_title": "Sales Representative"}], "dirty_cols": ["company", "state", "job_title"], "clean_cols": ["company", "state", "job_title"], "plan": {"dataset_summary": "16 rows × 3 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"VT": "Vermont", "Gorgia": "Georgia", "WY": "Wyoming", "GA": "Georgia", "VERMONT": "Vermont", "vt": "Vermont", "new hampshire": "New Hampshire", "MAINE": "Maine", "Veamont": "Vermont", "Me": "Maine"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"admin assistant": "Administrative Assistant", "CTO": "Chief Technology Officer", "ADIN": "Administrative Assistant", "Admin Asst": "Administrative Assistant", "ADMINISTRATIVE ASSISTANT": "Administrative Assistant", "rep": "Sales Representative", "Admin Assistant": "Administrative Assistant", "Adin": "Administrative Assistant", "ADMIN ASSISTANT": "Administrative Assistant", "Sales Rep": "Sales Representative", "Rep": "Sales Representative"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "canadian dollar", "industry": "Real Estate", "state": "MO", "department": "Cust Support", "signup_date": "12/14/2023", "job_title": "Sales Representative", "status": "canceled", "is_active": "FALSE"}, {"currency": " BOB ", "industry": "Utilities", "state": "Mo  ", "department": "Suupport ", "signup_date": "2023-05-20", "job_title": "account executive", "status": "cancelled", "is_active": "Y"}, {"currency": "bob", "industry": "ENERKY", "state": "connecticut", "department": " IT ", "signup_date": "28 Jun 2023", "job_title": "Chief Executive Oficer   ", "status": "ACTIVE", "is_active": "No"}, {"currency": "boliviano", "industry": "REAL ESTATE", "state": "CT", "department": "Customer Support", "signup_date": "6/15/2023", "job_title": "ACCOUNT  EXECUTIVE", "status": "Triil", "is_active": "0"}, {"currency": "Boliviano", "industry": "Oil & Gas", "state": "missouri", "department": "biz dev", "signup_date": "45097", "job_title": "ceo", "status": "TRIAL", "is_active": "true"}, {"currency": "  BOB", "industry": "PropTech", "state": "MISSOURI", "department": "customer support", "signup_date": "2023-04-09", "job_title": "Sales Rep", "status": "freebtrial", "is_active": "T"}, {"currency": "cad", "industry": "RealEstate", "state": "   MO", "department": "I.T.  ", "signup_date": "1 May 2023", "job_title": "Acct Exec", "status": "   CNACELED", "is_active": "false"}, {"currency": "   Boliviano", "industry": " utilities ", "state": "Connecticut", "department": "Customer Support", "signup_date": "5/18/2023", "job_title": "Chief Executive Ofuicer", "status": "active", "is_active": "Yes"}, {"currency": " cad ", "industry": "Oil & Gas", "state": "Ct", "department": "Biz  Dev", "signup_date": "1 Apr 2023", "job_title": "rep", "status": "CHURNED", "is_active": "TRUE"}, {"currency": "Boliviano", "industry": "Real  Estate", "state": "missori", "department": "It", "signup_date": "10/13/2023", "job_title": "Rep", "status": "active", "is_active": "No"}, {"currency": "  bob", "industry": "Utilities", "state": "   Ct", "department": "biz dev", "signup_date": "1/25/2023", "job_title": "c.e.o.", "status": "active", "is_active": "FALSE"}, {"currency": "Boliviano", "industry": "oil & gas", "state": "MO", "department": "I.T.", "signup_date": "45187", "job_title": "Sales  Rep", "status": "Trial", "is_active": "1"}, {"currency": "bob", "industry": " proptech ", "state": "MO", "department": " Sales", "signup_date": "2023-07-28", "job_title": "ceo", "status": "active", "is_active": "FALSE"}, {"currency": "Canadian Dollar", "industry": "Real Estate", "state": "   Ct", "department": "CUSTOMER SUPPORT", "signup_date": "45037", "job_title": "ceo", "status": "trial", "is_active": "1"}, {"currency": "AOA", "industry": "Energy  ", "state": "CT", "department": "Sales", "signup_date": "2023-06-13", "job_title": "CHIEF EXECUTIVE OFFICER", "status": "   CHUNED", "is_active": "N"}, {"currency": "   Kwanza", "industry": "Energy", "state": "ct", "department": "Customer Support", "signup_date": "7/17/2023", "job_title": "chief executive officer", "status": "active", "is_active": "false"}, {"currency": "CAD", "industry": "PROPERTY", "state": "   CT", "department": "information technology", "signup_date": "2023-05-10", "job_title": "Sales Representative", "status": "Trial", "is_active": "FALSE"}, {"currency": "AOA", "industry": "Real Estate", "state": "ct  ", "department": "Sales", "signup_date": "2023-12-24", "job_title": "sales  rep", "status": "active", "is_active": "T"}, {"currency": "AOA", "industry": "Energy", "state": "MISSOURI", "department": " information technology", "signup_date": "7/19/2023", "job_title": " AE ", "status": "active", "is_active": "1"}, {"currency": "boliviano", "industry": "realestate", "state": "Missori", "department": "it", "signup_date": "8/20/2023", "job_title": " Chief Executive Officer", "status": "TRIAL", "is_active": "false"}, {"currency": "AOA", "industry": "enewgy", "state": "missouri", "department": "biz dev", "signup_date": "2023-11-18", "job_title": "c.e.o.", "status": "ACTIVE", "is_active": "No"}, {"currency": "AOA", "industry": "PropTech", "state": "Connecticut", "department": "Biz Dev", "signup_date": "12/13/2023", "job_title": "Acct Exec", "status": "triial", "is_active": "TRUE"}, {"currency": "AOA", "industry": "UTTILITIES", "state": " CT ", "department": "Biz Dev", "signup_date": "19 Nov 2023", "job_title": "C.E.O.", "status": "TRIAL", "is_active": "N"}, {"currency": "Canadian Dollar", "industry": "Oil  & Gas", "state": "MO", "department": "IT", "signup_date": "21 Jul 2023", "job_title": "sales rpe", "status": "ACTIVE", "is_active": "F"}, {"currency": "bob", "industry": "real estate", "state": "ct", "department": "support", "signup_date": "8/2/2023", "job_title": "Sales Rep", "status": "tzial", "is_active": "false"}, {"currency": "BOB", "industry": "Real  Estate", "state": "Missouri", "department": "IT", "signup_date": "2023-04-25", "job_title": "SALES REPRESENTATIVE", "status": "   TRIAL", "is_active": "N"}, {"currency": "boliviano", "industry": " Realestate ", "state": "Missouri", "department": "Information Technology", "signup_date": "11/4/2023", "job_title": "sales representative", "status": "CANCELED", "is_active": "FALSE"}, {"currency": " Kwanza ", "industry": "Utilities", "state": "MISSOURI", "department": "information technology", "signup_date": "2023-03-20", "job_title": "   CEO", "status": "active", "is_active": "false"}, {"currency": "BOB", "industry": "  oil & gas", "state": "MO", "department": "biz dev", "signup_date": "44983", "job_title": "Account Executive", "status": "active", "is_active": "1"}, {"currency": "canadan dollar", "industry": "OIL & GAS", "state": "Missouri", "department": "I.T.  ", "signup_date": "10 Oct 2023", "job_title": "sales representaitve", "status": "trial ", "is_active": "false"}, {"currency": "CAD", "industry": "RaelEstate  ", "state": "Mo", "department": "biz dev", "signup_date": "1 Jan 2023", "job_title": " AE ", "status": "ACTIVE", "is_active": "T"}, {"currency": "Canaadian Dollar", "industry": "Realestate  ", "state": "ct", "department": "i..", "signup_date": "7 Jul 2023", "job_title": "AE", "status": " churmed ", "is_active": "false"}, {"currency": "AOA", "industry": "Oil & GGas", "state": "Missouri ", "department": "Sales", "signup_date": "2023-04-10", "job_title": "CEO", "status": "Churned", "is_active": "TRUE"}, {"currency": "BOB", "industry": "Real Estaet  ", "state": "Ct", "department": "Customer Support", "signup_date": "2023-01-01", "job_title": "Ae", "status": "active   ", "is_active": "TRUE"}, {"currency": "BOB", "industry": "RealEstate", "state": "Connecticut  ", "department": "Biz Dev", "signup_date": "44959", "job_title": "ceo", "status": "Trial", "is_active": "true"}], "clean": [{"currency": "CAD", "industry": "Real Estate", "state": "Missouri", "department": "Customer Support", "signup_date": "2023-12-14", "job_title": "Sales Representative", "status": "Churned", "is_active": false}, {"currency": "BOB", "industry": "Energy", "state": "Missouri", "department": "Customer Support", "signup_date": "2023-05-20", "job_title": "Account Executive", "status": "Churned", "is_active": true}, {"currency": "BOB", "industry": "Energy", "state": "Connecticut", "department": "Information Technology", "signup_date": "2023-06-28", "job_title": "Chief Executive Officer", "status": "Active", "is_active": false}, {"currency": "BOB", "industry": "Real Estate", "state": "Connecticut", "department": "Customer Support", "signup_date": "2023-06-15", "job_title": "Account Executive", "status": "Trial", "is_active": false}, {"currency": "BOB", "industry": "Energy", "state": "Missouri", "department": "Sales", "signup_date": "2023-06-20", "job_title": "Chief Executive Officer", "status": "Trial", "is_active": true}, {"currency": "BOB", "industry": "Real Estate", "state": "Missouri", "department": "Customer Support", "signup_date": "2023-04-09", "job_title": "Sales Representative", "status": "Trial", "is_active": true}, {"currency": "CAD", "industry": "Real Estate", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-05-01", "job_title": "Account Executive", "status": "Churned", "is_active": false}, {"currency": "BOB", "industry": "Energy", "state": "Connecticut", "department": "Customer Support", "signup_date": "2023-05-18", "job_title": "Chief Executive Officer", "status": "Active", "is_active": true}, {"currency": "CAD", "industry": "Energy", "state": "Connecticut", "department": "Sales", "signup_date": "2023-04-01", "job_title": "Sales Representative", "status": "Churned", "is_active": true}, {"currency": "BOB", "industry": "Real Estate", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-10-13", "job_title": "Sales Representative", "status": "Active", "is_active": false}, {"currency": "BOB", "industry": "Energy", "state": "Connecticut", "department": "Sales", "signup_date": "2023-01-25", "job_title": "Chief Executive Officer", "status": "Active", "is_active": false}, {"currency": "BOB", "industry": "Energy", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-09-18", "job_title": "Sales Representative", "status": "Trial", "is_active": true}, {"currency": "BOB", "industry": "Real Estate", "state": "Missouri", "department": "Sales", "signup_date": "2023-07-28", "job_title": "Chief Executive Officer", "status": "Active", "is_active": false}, {"currency": "CAD", "industry": "Real Estate", "state": "Connecticut", "department": "Customer Support", "signup_date": "2023-04-21", "job_title": "Chief Executive Officer", "status": "Trial", "is_active": true}, {"currency": "AOA", "industry": "Energy", "state": "Connecticut", "department": "Sales", "signup_date": "2023-06-13", "job_title": "Chief Executive Officer", "status": "Churned", "is_active": false}, {"currency": "AOA", "industry": "Energy", "state": "Connecticut", "department": "Customer Support", "signup_date": "2023-07-17", "job_title": "Chief Executive Officer", "status": "Active", "is_active": false}, {"currency": "CAD", "industry": "Real Estate", "state": "Connecticut", "department": "Information Technology", "signup_date": "2023-05-10", "job_title": "Sales Representative", "status": "Trial", "is_active": false}, {"currency": "AOA", "industry": "Real Estate", "state": "Connecticut", "department": "Sales", "signup_date": "2023-12-24", "job_title": "Sales Representative", "status": "Active", "is_active": true}, {"currency": "AOA", "industry": "Energy", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-07-19", "job_title": "Account Executive", "status": "Active", "is_active": true}, {"currency": "BOB", "industry": "Real Estate", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-08-20", "job_title": "Chief Executive Officer", "status": "Trial", "is_active": false}, {"currency": "AOA", "industry": "Energy", "state": "Missouri", "department": "Sales", "signup_date": "2023-11-18", "job_title": "Chief Executive Officer", "status": "Active", "is_active": false}, {"currency": "AOA", "industry": "Real Estate", "state": "Connecticut", "department": "Sales", "signup_date": "2023-12-13", "job_title": "Account Executive", "status": "Trial", "is_active": true}, {"currency": "AOA", "industry": "Energy", "state": "Connecticut", "department": "Sales", "signup_date": "2023-11-19", "job_title": "Chief Executive Officer", "status": "Trial", "is_active": false}, {"currency": "CAD", "industry": "Energy", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-07-21", "job_title": "Sales Representative", "status": "Active", "is_active": false}, {"currency": "BOB", "industry": "Real Estate", "state": "Connecticut", "department": "Customer Support", "signup_date": "2023-08-02", "job_title": "Sales Representative", "status": "Trial", "is_active": false}, {"currency": "BOB", "industry": "Real Estate", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-04-25", "job_title": "Sales Representative", "status": "Trial", "is_active": false}, {"currency": "BOB", "industry": "Real Estate", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-11-04", "job_title": "Sales Representative", "status": "Churned", "is_active": false}, {"currency": "AOA", "industry": "Energy", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-03-20", "job_title": "Chief Executive Officer", "status": "Active", "is_active": false}, {"currency": "BOB", "industry": "Energy", "state": "Missouri", "department": "Sales", "signup_date": "2023-02-26", "job_title": "Account Executive", "status": "Active", "is_active": true}, {"currency": "CAD", "industry": "Energy", "state": "Missouri", "department": "Information Technology", "signup_date": "2023-10-10", "job_title": "Sales Representative", "status": "Trial", "is_active": false}, {"currency": "CAD", "industry": "Real Estate", "state": "Missouri", "department": "Sales", "signup_date": "2023-01-01", "job_title": "Account Executive", "status": "Active", "is_active": true}, {"currency": "CAD", "industry": "Real Estate", "state": "Connecticut", "department": "Information Technology", "signup_date": "2023-07-07", "job_title": "Account Executive", "status": "Churned", "is_active": false}, {"currency": "AOA", "industry": "Energy", "state": "Missouri", "department": "Sales", "signup_date": "2023-04-10", "job_title": "Chief Executive Officer", "status": "Churned", "is_active": true}, {"currency": "BOB", "industry": "Real Estate", "state": "Connecticut", "department": "Customer Support", "signup_date": "2023-01-01", "job_title": "Account Executive", "status": "Active", "is_active": true}, {"currency": "BOB", "industry": "Real Estate", "state": "Connecticut", "department": "Sales", "signup_date": "2023-02-02", "job_title": "Chief Executive Officer", "status": "Trial", "is_active": true}], "dirty_cols": ["currency", "industry", "state", "department", "signup_date", "job_title", "status", "is_active"], "clean_cols": ["currency", "industry", "state", "department", "signup_date", "job_title", "status", "is_active"], "plan": {"dataset_summary": "35 rows × 8 columns. 8 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"canadian dollar": "CAD", "bob": "BOB", "boliviano": "BOB", "Boliviano": "BOB", "cad": "CAD", "Canadian Dollar": "CAD", "Kwanza": "AOA", "canadan dollar": "CAD", "Canaadian Dollar": "CAD"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Utilities": "Energy", "ENERKY": "Energy", "REAL ESTATE": "Real Estate", "Oil & Gas": "Energy", "PropTech": "Real Estate", "RealEstate": "Real Estate", "utilities": "Energy", "oil & gas": "Energy", "proptech": "Real Estate", "PROPERTY": "Real Estate", "realestate": "Real Estate", "enewgy": "Energy", "UTTILITIES": "Energy", "real estate": "Real Estate", "Realestate": "Real Estate", "OIL & GAS": "Energy", "RaelEstate": "Real Estate", "Oil & GGas": "Energy", "Real Estaet": "Real Estate"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MO": "Missouri", "Mo": "Missouri", "connecticut": "Connecticut", "CT": "Connecticut", "missouri": "Missouri", "MISSOURI": "Missouri", "Ct": "Connecticut", "missori": "Missouri", "ct": "Connecticut", "Missori": "Missouri"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cust Support": "Customer Support", "Suupport": "Customer Support", "IT": "Information Technology", "biz dev": "Sales", "customer support": "Customer Support", "I.T.": "Information Technology", "Biz Dev": "Sales", "It": "Information Technology", "CUSTOMER SUPPORT": "Customer Support", "information technology": "Information Technology", "it": "Information Technology", "support": "Customer Support", "i..": "Information Technology"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"account executive": "Account Executive", "Chief Executive Oficer": "Chief Executive Officer", "ACCOUNT EXECUTIVE": "Account Executive", "ceo": "Chief Executive Officer", "Sales Rep": "Sales Representative", "Acct Exec": "Account Executive", "Chief Executive Ofuicer": "Chief Executive Officer", "rep": "Sales Representative", "Rep": "Sales Representative", "c.e.o.": "Chief Executive Officer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "chief executive officer": "Chief Executive Officer", "sales rep": "Sales Representative", "AE": "Account Executive", "C.E.O.": "Chief Executive Officer", "sales rpe": "Sales Representative", "SALES REPRESENTATIVE": "Sales Representative", "sales representative": "Sales Representative", "CEO": "Chief Executive Officer", "sales representaitve": "Sales Representative", "Ae": "Account Executive"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"canceled": "Churned", "cancelled": "Churned", "ACTIVE": "Active", "Triil": "Trial", "TRIAL": "Trial", "freebtrial": "Trial", "CNACELED": "Churned", "active": "Active", "CHURNED": "Churned", "trial": "Trial", "CHUNED": "Churned", "triial": "Trial", "tzial": "Trial", "CANCELED": "Churned", "churmed": "Churned"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"signup_date": "4 Oct 2023", "job_title": "  product manager", "city": " Shanwang ", "company": "Wonka Co", "is_active": "Y", "status": "Refunded   ", "unit": "litre"}, {"signup_date": "2023-11-05", "job_title": "ProdM gr", "city": "Germistton", "company": "Umbrella", "is_active": "0", "status": "  PAID", "unit": "liter"}, {"signup_date": "5 Mar 2023", "job_title": "Sales Represenative", "city": "tōkyō", "company": "Hooli", "is_active": "F", "status": "Overdue", "unit": "l"}, {"signup_date": "2 Nov 2023", "job_title": "Sales  Representative", "city": "san diego", "company": "Cyberdyne", "is_active": "N", "status": "Overdue ", "unit": "liter"}, {"signup_date": "1/1/2023", "job_title": "Sales Representative", "city": "   germiston", "company": "Umbrella  ", "is_active": "Yes", "status": " late ", "unit": "ml"}, {"signup_date": "2023-09-02", "job_title": "Rep", "city": "SAN DIEGO", "company": "Globex", "is_active": "FALSE", "status": " paid", "unit": "litre"}, {"signup_date": "45051", "job_title": "sales rep", "city": "SAN DIEGO", "company": "  Soylent Corp", "is_active": "FALSE", "status": "refunded", "unit": "liter"}, {"signup_date": "2023-02-12", "job_title": "Sales  Represcntative", "city": "  SHANWANG", "company": "Vehement", "is_active": "false", "status": "OVERDUE", "unit": "ML"}, {"signup_date": "45221", "job_title": "product manager", "city": "San Diego", "company": "Globex", "is_active": "Yes", "status": " refsnd ", "unit": "L"}, {"signup_date": "2023-01-28", "job_title": "Sales Rep   ", "city": "San Diego", "company": "  Soylent Corp", "is_active": "TRUE", "status": "refunded", "unit": " milliliter "}, {"signup_date": "5/24/2023", "job_title": "Prod  Mgr", "city": "Shanwang", "company": "   Vehement", "is_active": "false", "status": "Refunped", "unit": "   ml"}, {"signup_date": "20 May 2023", "job_title": "SALES REP", "city": "Sa Diego", "company": "Soylent Corp", "is_active": "FALSE", "status": "paid", "unit": "litre"}, {"signup_date": "12 Mar 2023", "job_title": "  Sales Rep", "city": "Germiston", "company": "Umbrella", "is_active": "F", "status": "refunded", "unit": "ml"}, {"signup_date": "22 Jun 2023", "job_title": "Rep", "city": "GERMISTON", "company": "Wonka Co", "is_active": "true", "status": "refunded", "unit": "ml"}, {"signup_date": "27 Feb 2023", "job_title": "Sales Rep", "city": "Germiston", "company": "Umbrella  ", "is_active": "1", "status": "paid", "unit": "LITERS"}, {"signup_date": "6 Oct 2023", "job_title": "Sales  Rep", "city": " Tokyo", "company": "  Hooli", "is_active": "TRUE", "status": "ltae", "unit": "mls"}, {"signup_date": "45268", "job_title": "Rep", "city": " San Diego", "company": "  Stark Industries", "is_active": "No", "status": "Overdeu", "unit": "mls  "}, {"signup_date": "6 Aug 2023", "job_title": "Sales Representative", "city": " Tokyo", "company": "Wonka Co", "is_active": "FALSE", "status": "paid ", "unit": "liter"}, {"signup_date": "45238", "job_title": "prod mgr", "city": "San Diego", "company": " Cyberdyne ", "is_active": "Y", "status": "paid", "unit": "liters"}, {"signup_date": "16 Aug 2023", "job_title": "SALES REPRESENTATIVE", "city": " Tokyo", "company": "Initech", "is_active": "T", "status": "Overdue", "unit": "L"}, {"signup_date": "17 Jul 2023", "job_title": "SALES REPRESENTATIVE", "city": "SAN DIEGO", "company": "   Cyberdyne", "is_active": "false", "status": "overdue", "unit": "L"}, {"signup_date": "4 Oct 2023", "job_title": "PROD MGR", "city": "  Gemriston", "company": "Wonka Co", "is_active": "N", "status": "Paid", "unit": "mls"}, {"signup_date": "2/13/2023", "job_title": "REP", "city": "shanwang", "company": "   Stark Industries", "is_active": "false", "status": "Overdue", "unit": "ml"}, {"signup_date": "2023-05-07", "job_title": "Prroduct Manager", "city": "Tky  ", "company": "Soylent Corp", "is_active": "N", "status": "Refunded", "unit": "Mls"}, {"signup_date": "2023-10-21", "job_title": "PM", "city": "SAN DIEGO", "company": "Stark Industries", "is_active": "T", "status": "Paid", "unit": "l"}, {"signup_date": "2023-06-02", "job_title": "sales  rep", "city": " TKY ", "company": "Initech   ", "is_active": "FALSE", "status": "Pvid ", "unit": "l"}, {"signup_date": "17 Sep 2023", "job_title": "Product Manager", "city": "tokyo", "company": "Soylent Corp", "is_active": "TRUE", "status": "REFUNDED", "unit": "L"}, {"signup_date": "2023-11-19", "job_title": "  Sales Rep", "city": "  geriston", "company": "Soylent Corp  ", "is_active": "1", "status": "   Refunded", "unit": "LITERS"}, {"signup_date": "2023-06-21", "job_title": "Prod Manager", "city": "  TOKYO", "company": "Acme Inc", "is_active": "F", "status": "paid", "unit": "liter"}, {"signup_date": "2023-08-20", "job_title": "Rep", "city": "TOAYO", "company": "Stark Industries", "is_active": "Yes", "status": "refunded", "unit": " milliliter"}, {"signup_date": "7/8/2023", "job_title": "Prod Mgr", "city": "tōkyō", "company": "Globex", "is_active": "true", "status": "refunded ", "unit": "liter"}, {"signup_date": "1 Dec 2023", "job_title": "prod mgr", "city": " Tōkyō ", "company": " Hooli ", "is_active": "Y", "status": "Overdue", "unit": "milliliter   "}, {"signup_date": "15 Feb 2023", "job_title": "Sales Representative", "city": "SHANWANG", "company": "Cyberdyne   ", "is_active": "T", "status": "paid", "unit": "L "}, {"signup_date": "45231", "job_title": "Prod Mgr", "city": "San Diego", "company": "Initech", "is_active": "T", "status": "REFUND", "unit": "   liter"}, {"signup_date": "45217", "job_title": "Prod Manager", "city": "SAN DIEGO", "company": "Globex", "is_active": "F", "status": "ovardue", "unit": " L "}, {"signup_date": "2023-09-23", "job_title": "SALES REPRESENTATIVE", "city": "Germiston  ", "company": "Cyberdyne   ", "is_active": "No", "status": "late  ", "unit": "milliliter"}, {"signup_date": "2023-07-12", "job_title": "PM", "city": "Tōkyō", "company": "   Stark Industries", "is_active": "FALSE", "status": "PAID", "unit": "   Ml"}, {"signup_date": "18 Apr 2023", "job_title": "Sales Rep", "city": "Tokyo", "company": "  Wonka Co", "is_active": "Yes", "status": " refund", "unit": "ltier"}, {"signup_date": "2023-12-03", "job_title": "REP", "city": "SAN DIEGO", "company": "Vehement  ", "is_active": "true", "status": "   paid", "unit": "l"}, {"signup_date": "1/18/2023", "job_title": "Prod Manager", "city": "tōkyō", "company": "   Hooli", "is_active": "TRUE", "status": "Paid", "unit": "mL"}, {"signup_date": "2023-07-14", "job_title": " prod mgr", "city": "Germiston", "company": "Umbrella", "is_active": "1", "status": "refunded", "unit": "  milliliter"}, {"signup_date": "1/9/2023", "job_title": "PROD MGR", "city": "Germiston ", "company": " Umbrella ", "is_active": "Yes", "status": "Paid", "unit": "mL"}, {"signup_date": "45194", "job_title": "   sales representative", "city": "tokyo", "company": " Initech ", "is_active": "Yes", "status": "PAID", "unit": "L"}, {"signup_date": "17 Sep 2023", "job_title": "Product Manager", "city": "tokyo", "company": "Soylent Corp", "is_active": "TRUE", "status": "REFUNDED", "unit": "L"}, {"signup_date": "7/8/2023", "job_title": "Prod Mgr", "city": "tōkyō", "company": "Globex", "is_active": "true", "status": "refunded ", "unit": "liter"}], "clean": [{"signup_date": "2023-10-04", "job_title": "Product Manager", "city": "Shanwang", "company": "Wonka Co", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-11-05", "job_title": "Product Manager", "city": "Germiston", "company": "Umbrella", "is_active": false, "status": "Paid", "unit": "L"}, {"signup_date": "2023-03-05", "job_title": "Sales Representative", "city": "Tokyo", "company": "Hooli", "is_active": false, "status": "Overdue", "unit": "L"}, {"signup_date": "2023-11-02", "job_title": "Sales Representative", "city": "San Diego", "company": "Cyberdyne", "is_active": false, "status": "Overdue", "unit": "L"}, {"signup_date": "2023-01-01", "job_title": "Sales Representative", "city": "Germiston", "company": "Umbrella", "is_active": true, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-09-02", "job_title": "Sales Representative", "city": "San Diego", "company": "Globex", "is_active": false, "status": "Paid", "unit": "L"}, {"signup_date": "2023-05-05", "job_title": "Sales Representative", "city": "San Diego", "company": "Soylent Corp", "is_active": false, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-02-12", "job_title": "Sales Representative", "city": "Shanwang", "company": "Vehement", "is_active": false, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-10-22", "job_title": "Product Manager", "city": "San Diego", "company": "Globex", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-01-28", "job_title": "Sales Representative", "city": "San Diego", "company": "Soylent Corp", "is_active": true, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-05-24", "job_title": "Product Manager", "city": "Shanwang", "company": "Vehement", "is_active": false, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-05-20", "job_title": "Sales Representative", "city": "San Diego", "company": "Soylent Corp", "is_active": false, "status": "Paid", "unit": "L"}, {"signup_date": "2023-03-12", "job_title": "Sales Representative", "city": "Germiston", "company": "Umbrella", "is_active": false, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-06-22", "job_title": "Sales Representative", "city": "Germiston", "company": "Wonka Co", "is_active": true, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-02-27", "job_title": "Sales Representative", "city": "Germiston", "company": "Umbrella", "is_active": true, "status": "Paid", "unit": "L"}, {"signup_date": "2023-10-06", "job_title": "Sales Representative", "city": "Tokyo", "company": "Hooli", "is_active": true, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-12-08", "job_title": "Sales Representative", "city": "San Diego", "company": "Stark Industries", "is_active": false, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-08-06", "job_title": "Sales Representative", "city": "Tokyo", "company": "Wonka Co", "is_active": false, "status": "Paid", "unit": "L"}, {"signup_date": "2023-11-08", "job_title": "Product Manager", "city": "San Diego", "company": "Cyberdyne", "is_active": true, "status": "Paid", "unit": "L"}, {"signup_date": "2023-08-16", "job_title": "Sales Representative", "city": "Tokyo", "company": "Initech", "is_active": true, "status": "Overdue", "unit": "L"}, {"signup_date": "2023-07-17", "job_title": "Sales Representative", "city": "San Diego", "company": "Cyberdyne", "is_active": false, "status": "Overdue", "unit": "L"}, {"signup_date": "2023-10-04", "job_title": "Product Manager", "city": "Germiston", "company": "Wonka Co", "is_active": false, "status": "Paid", "unit": "mL"}, {"signup_date": "2023-02-13", "job_title": "Sales Representative", "city": "Shanwang", "company": "Stark Industries", "is_active": false, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-05-07", "job_title": "Product Manager", "city": "Tokyo", "company": "Soylent Corp", "is_active": false, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-10-21", "job_title": "Product Manager", "city": "San Diego", "company": "Stark Industries", "is_active": true, "status": "Paid", "unit": "L"}, {"signup_date": "2023-06-02", "job_title": "Sales Representative", "city": "Tokyo", "company": "Initech", "is_active": false, "status": "Paid", "unit": "L"}, {"signup_date": "2023-09-17", "job_title": "Product Manager", "city": "Tokyo", "company": "Soylent Corp", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-11-19", "job_title": "Sales Representative", "city": "Germiston", "company": "Soylent Corp", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-06-21", "job_title": "Product Manager", "city": "Tokyo", "company": "Acme Inc", "is_active": false, "status": "Paid", "unit": "L"}, {"signup_date": "2023-08-20", "job_title": "Sales Representative", "city": "Tokyo", "company": "Stark Industries", "is_active": true, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-07-08", "job_title": "Product Manager", "city": "Tokyo", "company": "Globex", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-12-01", "job_title": "Product Manager", "city": "Tokyo", "company": "Hooli", "is_active": true, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-02-15", "job_title": "Sales Representative", "city": "Shanwang", "company": "Cyberdyne", "is_active": true, "status": "Paid", "unit": "L"}, {"signup_date": "2023-11-01", "job_title": "Product Manager", "city": "San Diego", "company": "Initech", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-10-18", "job_title": "Product Manager", "city": "San Diego", "company": "Globex", "is_active": false, "status": "Overdue", "unit": "L"}, {"signup_date": "2023-09-23", "job_title": "Sales Representative", "city": "Germiston", "company": "Cyberdyne", "is_active": false, "status": "Overdue", "unit": "mL"}, {"signup_date": "2023-07-12", "job_title": "Product Manager", "city": "Tokyo", "company": "Stark Industries", "is_active": false, "status": "Paid", "unit": "mL"}, {"signup_date": "2023-04-18", "job_title": "Sales Representative", "city": "Tokyo", "company": "Wonka Co", "is_active": true, "status": "Refunded", "unit": "L"}, {"signup_date": "2023-12-03", "job_title": "Sales Representative", "city": "San Diego", "company": "Vehement", "is_active": true, "status": "Paid", "unit": "L"}, {"signup_date": "2023-01-18", "job_title": "Product Manager", "city": "Tokyo", "company": "Hooli", "is_active": true, "status": "Paid", "unit": "mL"}, {"signup_date": "2023-07-14", "job_title": "Product Manager", "city": "Germiston", "company": "Umbrella", "is_active": true, "status": "Refunded", "unit": "mL"}, {"signup_date": "2023-01-09", "job_title": "Product Manager", "city": "Germiston", "company": "Umbrella", "is_active": true, "status": "Paid", "unit": "mL"}, {"signup_date": "2023-09-25", "job_title": "Sales Representative", "city": "Tokyo", "company": "Initech", "is_active": true, "status": "Paid", "unit": "L"}], "dirty_cols": ["signup_date", "job_title", "city", "company", "is_active", "status", "unit"], "clean_cols": ["signup_date", "job_title", "city", "company", "is_active", "status", "unit"], "plan": {"dataset_summary": "45 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"product manager": "Product Manager", "ProdM gr": "Product Manager", "Sales Represenative": "Sales Representative", "Rep": "Sales Representative", "sales rep": "Sales Representative", "Sales Represcntative": "Sales Representative", "Sales Rep": "Sales Representative", "Prod Mgr": "Product Manager", "SALES REP": "Sales Representative", "prod mgr": "Product Manager", "SALES REPRESENTATIVE": "Sales Representative", "PROD MGR": "Product Manager", "REP": "Sales Representative", "Prroduct Manager": "Product Manager", "PM": "Product Manager", "Prod Manager": "Product Manager", "sales representative": "Sales Representative"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Germistton": "Germiston", "tōkyō": "Tokyo", "san diego": "San Diego", "germiston": "Germiston", "SAN DIEGO": "San Diego", "SHANWANG": "Shanwang", "Sa Diego": "San Diego", "GERMISTON": "Germiston", "Gemriston": "Germiston", "shanwang": "Shanwang", "Tky": "Tokyo", "TKY": "Tokyo", "tokyo": "Tokyo", "geriston": "Germiston", "TOKYO": "Tokyo", "TOAYO": "Tokyo", "Tōkyō": "Tokyo"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PAID": "Paid", "late": "Overdue", "paid": "Paid", "refunded": "Refunded", "OVERDUE": "Overdue", "refsnd": "Refunded", "Refunped": "Refunded", "ltae": "Overdue", "Overdeu": "Overdue", "overdue": "Overdue", "Pvid": "Paid", "REFUNDED": "Refunded", "REFUND": "Refunded", "ovardue": "Overdue", "refund": "Refunded"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"litre": "L", "liter": "L", "l": "L", "ml": "mL", "ML": "mL", "milliliter": "mL", "LITERS": "L", "mls": "mL", "liters": "L", "Mls": "mL", "Ml": "mL", "ltier": "L"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"company": " Hooli ", "amount": "--", "status": "P1", "city": "DUKLI", "department": "biz dev", "notes2": ""}, {"company": "Umbrella", "amount": "1.190,11", "status": "high", "city": "Portland ", "department": "operations", "notes2": ""}, {"company": "Soylent Corp", "amount": "$7,858.77", "status": "med", "city": "dukli", "department": "sales", "notes2": ""}, {"company": "  Vehement", "amount": "8.164,02", "status": "High", "city": "portland", "department": "   OPS", "notes2": ""}, {"company": "Cyberdyne", "amount": "6.199,73", "status": "Medium", "city": "Ifrane", "department": "HR  ", "notes2": ""}, {"company": "Cyberdyne  ", "amount": "3.239,78", "status": "medium", "city": "Portland", "department": "HR", "notes2": ""}, {"company": "Vehement", "amount": "7.173,63", "status": "med", "city": "Wasvington D.C.", "department": "Human Resources", "notes2": ""}, {"company": "Initech", "amount": "1.079,59", "status": "High", "city": "Ifrane", "department": "HUMAN RESOURCES", "notes2": ""}, {"company": "Globex ", "amount": "$7,310.61", "status": " high ", "city": "DUKLI", "department": " legal & compliance", "notes2": ""}, {"company": " Initech ", "amount": "na", "status": "Medium", "city": "ifrane", "department": "Legal  & Compliance", "notes2": ""}, {"company": "Acme Inc  ", "amount": "$6,957.21", "status": "P1", "city": "Dukli", "department": "legal", "notes2": ""}, {"company": " Acme Inc", "amount": "3.633,24", "status": "M", "city": "Dukli", "department": "   operations", "notes2": ""}, {"company": "Acme Inc  ", "amount": "2.218,10", "status": "medium ", "city": "Ifrane", "department": "OPERATOINS", "notes2": ""}, {"company": " Hooli", "amount": "-", "status": " p1", "city": "DUKLI", "department": "HR ", "notes2": ""}, {"company": "Acme Inc", "amount": "None", "status": "P1 ", "city": "IFRANE", "department": "HR", "notes2": ""}, {"company": "Umbrella   ", "amount": "$6,597.63", "status": "p1", "city": "Washingtn DC", "department": " BI DEV", "notes2": ""}, {"company": "Acme Inc", "amount": "--", "status": "HIGH", "city": "Washington", "department": " People ", "notes2": ""}, {"company": "   Soylent Corp", "amount": "$5,650.50", "status": "medium", "city": "   Dukli", "department": "HR", "notes2": ""}, {"company": "  Vehement", "amount": "$4,799.47", "status": "M", "city": "  portland", "department": "lega l& compliance", "notes2": ""}, {"company": "Stark Industries", "amount": "5.079,45", "status": "HIGH", "city": "DC", "department": "Operations", "notes2": ""}, {"company": "Hooli ", "amount": "4.542,52", "status": "High", "city": "  IFRANE", "department": "biz dev", "notes2": ""}, {"company": "Wonka Co ", "amount": "$8,667.99", "status": "high", "city": "Washington  DC", "department": "OPERATGONS", "notes2": ""}, {"company": " Soylent Corp", "amount": "$2,647.41", "status": "med", "city": "WAHSINGTON", "department": "Sales", "notes2": ""}, {"company": "   Cyberdyne", "amount": "6.686,51", "status": "High ", "city": "DC", "department": "legal & compliance", "notes2": ""}, {"company": "Acme Inc", "amount": "$6,276.57", "status": "H", "city": "Dukli  ", "department": "OPERATIONS", "notes2": ""}, {"company": "Soylent Corp", "amount": "$7,113.42", "status": "M", "city": "Dukli", "department": "Biz Dev", "notes2": ""}, {"company": "Acme Inc", "amount": "N/A", "status": "p2", "city": " Portland ", "department": " Salse ", "notes2": ""}, {"company": "Globex", "amount": "$748.55", "status": "H", "city": "Ifrane", "department": "ops", "notes2": ""}, {"company": "Umbrella ", "amount": "1.096,35", "status": "P2", "city": " Prtland", "department": "legaal", "notes2": ""}, {"company": "Wonka Co", "amount": "6.687,32", "status": "Medium", "city": "Ifrrane", "department": " Biz Dev", "notes2": ""}, {"company": "Umbrella   ", "amount": "1.196,29", "status": "H", "city": "WASHINGTON D.C.", "department": "operations", "notes2": ""}, {"company": "Hooli", "amount": "3.241,40", "status": "high", "city": "Dukli", "department": "LEGAL", "notes2": ""}], "clean": [{"company": "Hooli", "amount": NaN, "status": "High", "city": "Dukli", "department": "Sales"}, {"company": "Umbrella", "amount": 1190.11, "status": "High", "city": "Portland", "department": "Operations"}, {"company": "Soylent Corp", "amount": 7858.77, "status": "Medium", "city": "Dukli", "department": "Sales"}, {"company": "Vehement", "amount": 8164.02, "status": "High", "city": "Portland", "department": "Operations"}, {"company": "Cyberdyne", "amount": 6199.73, "status": "Medium", "city": "Ifrane", "department": "Human Resources"}, {"company": "Cyberdyne", "amount": 3239.78, "status": "Medium", "city": "Portland", "department": "Human Resources"}, {"company": "Vehement", "amount": 7173.63, "status": "Medium", "city": "Washington", "department": "Human Resources"}, {"company": "Initech", "amount": 1079.59, "status": "High", "city": "Ifrane", "department": "Human Resources"}, {"company": "Globex", "amount": 7310.61, "status": "High", "city": "Dukli", "department": "Legal"}, {"company": "Initech", "amount": NaN, "status": "Medium", "city": "Ifrane", "department": "Legal"}, {"company": "Acme Inc", "amount": 6957.21, "status": "High", "city": "Dukli", "department": "Legal"}, {"company": "Acme Inc", "amount": 3633.24, "status": "Medium", "city": "Dukli", "department": "Operations"}, {"company": "Acme Inc", "amount": 2218.1, "status": "Medium", "city": "Ifrane", "department": "Operations"}, {"company": "Hooli", "amount": NaN, "status": "High", "city": "Dukli", "department": "Human Resources"}, {"company": "Acme Inc", "amount": NaN, "status": "High", "city": "Ifrane", "department": "Human Resources"}, {"company": "Umbrella", "amount": 6597.63, "status": "High", "city": "Washington", "department": "Sales"}, {"company": "Acme Inc", "amount": NaN, "status": "High", "city": "Washington", "department": "Human Resources"}, {"company": "Soylent Corp", "amount": 5650.5, "status": "Medium", "city": "Dukli", "department": "Human Resources"}, {"company": "Vehement", "amount": 4799.47, "status": "Medium", "city": "Portland", "department": "Legal"}, {"company": "Stark Industries", "amount": 5079.45, "status": "High", "city": "Washington", "department": "Operations"}, {"company": "Hooli", "amount": 4542.52, "status": "High", "city": "Ifrane", "department": "Sales"}, {"company": "Wonka Co", "amount": 8667.99, "status": "High", "city": "Washington", "department": "Operations"}, {"company": "Soylent Corp", "amount": 2647.41, "status": "Medium", "city": "Washington", "department": "Sales"}, {"company": "Cyberdyne", "amount": 6686.51, "status": "High", "city": "Washington", "department": "Legal"}, {"company": "Acme Inc", "amount": 6276.57, "status": "High", "city": "Dukli", "department": "Operations"}, {"company": "Soylent Corp", "amount": 7113.42, "status": "Medium", "city": "Dukli", "department": "Sales"}, {"company": "Acme Inc", "amount": NaN, "status": "Medium", "city": "Portland", "department": "Sales"}, {"company": "Globex", "amount": 748.55, "status": "High", "city": "Ifrane", "department": "Operations"}, {"company": "Umbrella", "amount": 1096.35, "status": "Medium", "city": "Portland", "department": "Legal"}, {"company": "Wonka Co", "amount": 6687.32, "status": "Medium", "city": "Ifrane", "department": "Sales"}, {"company": "Umbrella", "amount": 1196.29, "status": "High", "city": "Washington", "department": "Operations"}, {"company": "Hooli", "amount": 3241.4, "status": "High", "city": "Dukli", "department": "Legal"}], "dirty_cols": ["company", "amount", "status", "city", "department", "notes2"], "clean_cols": ["company", "amount", "status", "city", "department"], "plan": {"dataset_summary": "32 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P1": "High", "high": "High", "med": "Medium", "medium": "Medium", "M": "Medium", "p1": "High", "HIGH": "High", "H": "High", "p2": "Medium", "P2": "Medium"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"DUKLI": "Dukli", "dukli": "Dukli", "portland": "Portland", "Wasvington D.C.": "Washington", "ifrane": "Ifrane", "IFRANE": "Ifrane", "Washingtn DC": "Washington", "DC": "Washington", "Washington DC": "Washington", "WAHSINGTON": "Washington", "Prtland": "Portland", "Ifrrane": "Ifrane", "WASHINGTON D.C.": "Washington"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"biz dev": "Sales", "operations": "Operations", "sales": "Sales", "OPS": "Operations", "HR": "Human Resources", "HUMAN RESOURCES": "Human Resources", "legal & compliance": "Legal", "Legal & Compliance": "Legal", "legal": "Legal", "OPERATOINS": "Operations", "BI DEV": "Sales", "People": "Human Resources", "lega l& compliance": "Legal", "OPERATGONS": "Operations", "OPERATIONS": "Operations", "Biz Dev": "Sales", "Salse": "Sales", "ops": "Operations", "legaal": "Legal", "LEGAL": "Legal"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "Omar Brandt", "email": "  bphdfo@example.com", "amount": "$2,626.07", "state": "Mn", "country": " AS ", "unnamed": ""}, {"name": " Lena Smith", "email": "bhdmk@corp.io", "amount": "$3,797.19", "state": " South Carolina", "country": " ly ", "unnamed": ""}, {"name": "   Priya Moore", "email": "dnpb@corp.io", "amount": "8.579,56", "state": "South Carolina", "country": " mtq ", "unnamed": ""}, {"name": " Omar Cruz", "email": "kjdblb@mail.com", "amount": "$7,371.26", "state": "minnesota", "country": "DO", "unnamed": ""}, {"name": "   Bob Wong", "email": "ieloija@mail.com", "amount": "3.569,13", "state": "   MN", "country": "Libya", "unnamed": ""}, {"name": "  Priya Park", "email": "PAKIF@MAIL.COM", "amount": "8.824,98", "state": "Minnesota", "country": "LY", "unnamed": ""}, {"name": "Ivan Khan", "email": "KDKKLI@MAIL.COM", "amount": "6.199,89", "state": "Minnesota", "country": "Dominican Republic", "unnamed": ""}, {"name": "Heidi Reyes  ", "email": "pnof@mail.com", "amount": "8.080,55", "state": "MN", "country": " Mtq ", "unnamed": ""}, {"name": "Karl  Moore", "email": "FHPEE@CORP.IO", "amount": "TBD", "state": "Mn", "country": " as ", "unnamed": ""}, {"name": "Omar Reyes", "email": "PELEI@EXAMPLE.COM", "amount": "3.198,16", "state": " MN ", "country": "do", "unnamed": ""}, {"name": "Judy Lee", "email": " gmmp@example.com", "amount": "7.578,01", "state": "South Carolina", "country": "as", "unnamed": ""}, {"name": "  Alice Ali", "email": "bpnpj@example.com", "amount": "$1,681.94", "state": "   mn", "country": "Dominican Republic", "unnamed": ""}, {"name": "Mona  Wong", "email": "ekhd@mail.com", "amount": "5.622,67", "state": "Minnesota", "country": "Libya", "unnamed": ""}, {"name": "David Park", "email": "MBBJ@EXAMPLE.COM ", "amount": "7.553,39", "state": "MINNESOTA", "country": "ASM", "unnamed": ""}, {"name": "Omar Smith", "email": " iecamj@mail.com", "amount": "3.578,48", "state": "MN", "country": "LBY", "unnamed": ""}, {"name": "Karl Smith", "email": "khnhhg@test.org  ", "amount": "8.171,17", "state": "mn", "country": "  AS", "unnamed": ""}, {"name": "Grace Lee", "email": "ogefpd@test.org", "amount": "133,37", "state": "mn   ", "country": "Libya", "unnamed": ""}, {"name": "Omar Wong", "email": "inekfcj@mail.com", "amount": "5.509,89", "state": "South Carolina", "country": "MTQ ", "unnamed": ""}, {"name": " Lena Johnson", "email": "DKLJJE@MAIL.COM", "amount": "4.232,75", "state": "south carolina", "country": "LBY", "unnamed": ""}, {"name": "Omar  Johnson", "email": "ppglgh@test.org ", "amount": "$6,890.22", "state": "Minnesota", "country": "Do", "unnamed": ""}, {"name": "Mona  Diaz", "email": "KMFOAC@TEST.ORG", "amount": "487,82", "state": "south  carolina", "country": "MQ", "unnamed": ""}, {"name": "Mona Brandt", "email": "AOHBIL@TEST.ORG", "amount": "$7,221.59", "state": "   South Carolina", "country": "Mq", "unnamed": ""}, {"name": "Judy Wong", "email": "jdgona@example.com", "amount": "$1,160.06", "state": "Minnesota", "country": "MQ", "unnamed": ""}, {"name": "Carol Johnson", "email": "bfnhhjn@corp.io   ", "amount": "2.977,10", "state": "SC", "country": "ASM", "unnamed": ""}, {"name": "Mona Fischer", "email": "enhdc@corp.io", "amount": "6.461,92", "state": "Minnesota", "country": "Dom", "unnamed": ""}, {"name": "  Heidi Cruz", "email": "KLLHG@CORP.IO", "amount": "7.983,57", "state": "South Carolina  ", "country": "ly", "unnamed": ""}, {"name": "Frank Moore", "email": "lbkn@corp.io", "amount": "3.385,93", "state": " South CCarolina", "country": "MQ ", "unnamed": ""}, {"name": "Mona Smith", "email": " ghpakon@mail.com ", "amount": "None", "state": "Minnesota", "country": "DO", "unnamed": ""}, {"name": "Lena Wong   ", "email": "dljn@corp.io", "amount": "$5,756.36", "state": "MN   ", "country": "MTQ", "unnamed": ""}, {"name": "Priya Lee", "email": "eape@corp.io", "amount": "6.473,94", "state": "MN", "country": "do", "unnamed": ""}, {"name": "Ivan Diaz", "email": "pbfjmch@test.org", "amount": "--", "state": " SC ", "country": "Dom", "unnamed": ""}, {"name": " Eve Petrov", "email": "hjkgenb@corp.io", "amount": "545,24", "state": "South Carolina", "country": " MQ ", "unnamed": ""}, {"name": "Judy  Petrov", "email": "hcda@example.com", "amount": "$2,281.98", "state": " SC", "country": "dominican republic", "unnamed": ""}, {"name": "Grace  Johnson", "email": "imkmolg@example.com", "amount": "$5,824.83", "state": "MN", "country": "LBY", "unnamed": ""}], "clean": [{"name": "Omar Brandt", "email": "bphdfo@example.com", "amount": 2626.07, "state": "Minnesota", "country": "American Samoa"}, {"name": "Lena Smith", "email": "bhdmk@corp.io", "amount": 3797.19, "state": "South Carolina", "country": "Libya"}, {"name": "Priya Moore", "email": "dnpb@corp.io", "amount": 8579.56, "state": "South Carolina", "country": "Martinique"}, {"name": "Omar Cruz", "email": "kjdblb@mail.com", "amount": 7371.26, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Bob Wong", "email": "ieloija@mail.com", "amount": 3569.13, "state": "Minnesota", "country": "Libya"}, {"name": "Priya Park", "email": "pakif@mail.com", "amount": 8824.98, "state": "Minnesota", "country": "Libya"}, {"name": "Ivan Khan", "email": "kdkkli@mail.com", "amount": 6199.89, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Heidi Reyes", "email": "pnof@mail.com", "amount": 8080.55, "state": "Minnesota", "country": "Martinique"}, {"name": "Karl Moore", "email": "fhpee@corp.io", "amount": NaN, "state": "Minnesota", "country": "American Samoa"}, {"name": "Omar Reyes", "email": "pelei@example.com", "amount": 3198.16, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Judy Lee", "email": "gmmp@example.com", "amount": 7578.01, "state": "South Carolina", "country": "American Samoa"}, {"name": "Alice Ali", "email": "bpnpj@example.com", "amount": 1681.94, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Mona Wong", "email": "ekhd@mail.com", "amount": 5622.67, "state": "Minnesota", "country": "Libya"}, {"name": "David Park", "email": "mbbj@example.com", "amount": 7553.39, "state": "Minnesota", "country": "American Samoa"}, {"name": "Omar Smith", "email": "iecamj@mail.com", "amount": 3578.48, "state": "Minnesota", "country": "Libya"}, {"name": "Karl Smith", "email": "khnhhg@test.org", "amount": 8171.17, "state": "Minnesota", "country": "American Samoa"}, {"name": "Grace Lee", "email": "ogefpd@test.org", "amount": 133.37, "state": "Minnesota", "country": "Libya"}, {"name": "Omar Wong", "email": "inekfcj@mail.com", "amount": 5509.89, "state": "South Carolina", "country": "Martinique"}, {"name": "Lena Johnson", "email": "dkljje@mail.com", "amount": 4232.75, "state": "South Carolina", "country": "Libya"}, {"name": "Omar Johnson", "email": "ppglgh@test.org", "amount": 6890.22, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Mona Diaz", "email": "kmfoac@test.org", "amount": 487.82, "state": "South Carolina", "country": "Martinique"}, {"name": "Mona Brandt", "email": "aohbil@test.org", "amount": 7221.59, "state": "South Carolina", "country": "Martinique"}, {"name": "Judy Wong", "email": "jdgona@example.com", "amount": 1160.06, "state": "Minnesota", "country": "Martinique"}, {"name": "Carol Johnson", "email": "bfnhhjn@corp.io", "amount": 2977.1, "state": "South Carolina", "country": "American Samoa"}, {"name": "Mona Fischer", "email": "enhdc@corp.io", "amount": 6461.92, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Heidi Cruz", "email": "kllhg@corp.io", "amount": 7983.57, "state": "South Carolina", "country": "Libya"}, {"name": "Frank Moore", "email": "lbkn@corp.io", "amount": 3385.93, "state": "South Carolina", "country": "Martinique"}, {"name": "Mona Smith", "email": "ghpakon@mail.com", "amount": NaN, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Lena Wong", "email": "dljn@corp.io", "amount": 5756.36, "state": "Minnesota", "country": "Martinique"}, {"name": "Priya Lee", "email": "eape@corp.io", "amount": 6473.94, "state": "Minnesota", "country": "Dominican Republic"}, {"name": "Ivan Diaz", "email": "pbfjmch@test.org", "amount": NaN, "state": "South Carolina", "country": "Dominican Republic"}, {"name": "Eve Petrov", "email": "hjkgenb@corp.io", "amount": 545.24, "state": "South Carolina", "country": "Martinique"}, {"name": "Judy Petrov", "email": "hcda@example.com", "amount": 2281.98, "state": "South Carolina", "country": "Dominican Republic"}, {"name": "Grace Johnson", "email": "imkmolg@example.com", "amount": 5824.83, "state": "Minnesota", "country": "Libya"}], "dirty_cols": ["name", "email", "amount", "state", "country", "unnamed"], "clean_cols": ["name", "email", "amount", "state", "country"], "plan": {"dataset_summary": "34 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mn": "Minnesota", "minnesota": "Minnesota", "MN": "Minnesota", "mn": "Minnesota", "MINNESOTA": "Minnesota", "south carolina": "South Carolina", "SC": "South Carolina", "South CCarolina": "South Carolina"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AS": "American Samoa", "ly": "Libya", "mtq": "Martinique", "DO": "Dominican Republic", "LY": "Libya", "Mtq": "Martinique", "as": "American Samoa", "do": "Dominican Republic", "ASM": "American Samoa", "LBY": "Libya", "MTQ": "Martinique", "Do": "Dominican Republic", "MQ": "Martinique", "Mq": "Martinique", "Dom": "Dominican Republic", "dominican republic": "Dominican Republic"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": " Jakarta", "unit": "MI", "currency": "GTQ", "phone": "505.489.2664", "email": "cfmkea@example.com"}, {"city": "jakarta", "unit": "   km", "currency": "  Quetzal", "phone": "(588)750-2608", "email": " NMCF@EXAMPLE.COM"}, {"city": "Kampong Dungun", "unit": "Miles", "currency": "BURUNDI FRANC  ", "phone": "5748247927", "email": "GMPHKBI@TEST.ORG"}, {"city": "Kampong Dungun", "unit": "KILOMETERS", "currency": " gtq ", "phone": "5722355153", "email": "KCAGM@MAIL.COM"}, {"city": "kampong dungnu", "unit": "CM", "currency": "BIF", "phone": "503.474.3358", "email": "APCND@CORP.IO"}, {"city": "   Kampong Dungun", "unit": "Mi", "currency": "BIF", "phone": "5502782682", "email": "KPKBIHK@TEST.ORG"}, {"city": " Jakarta ", "unit": "CENTIMETERS", "currency": "Burundi Franc", "phone": "573-569-6064", "email": "JBKK@TEST.ORG"}, {"city": "jakarsa", "unit": "km", "currency": "gtq", "phone": "5835976403", "email": "PLKE@CORP.IO"}, {"city": "SANTA TERESA", "unit": "Mi", "currency": "gtq", "phone": "594.536.5421", "email": "jphncha@example.com"}, {"city": "   Kampong Dungun", "unit": "centimeter", "currency": "burundi fbanc", "phone": "5574759149", "email": "pjhk@test.org"}, {"city": "   Kampong Dungun", "unit": "centimeter", "currency": "burundi fbanc", "phone": "5574759149", "email": "pjhk@test.org"}, {"city": " Jakarta ", "unit": "CENTIMETERS", "currency": "Burundi Franc", "phone": "573-569-6064", "email": "JBKK@TEST.ORG"}], "clean": [{"city": "Jakarta", "unit": "mi", "currency": "GTQ", "phone": "(505) 489-2664", "email": "cfmkea@example.com"}, {"city": "Jakarta", "unit": "km", "currency": "GTQ", "phone": "(588) 750-2608", "email": "nmcf@example.com"}, {"city": "Kampong Dungun", "unit": "mi", "currency": "BIF", "phone": "(574) 824-7927", "email": "gmphkbi@test.org"}, {"city": "Kampong Dungun", "unit": "km", "currency": "GTQ", "phone": "(572) 235-5153", "email": "kcagm@mail.com"}, {"city": "Kampong Dungun", "unit": "cm", "currency": "BIF", "phone": "(503) 474-3358", "email": "apcnd@corp.io"}, {"city": "Kampong Dungun", "unit": "mi", "currency": "BIF", "phone": "(550) 278-2682", "email": "kpkbihk@test.org"}, {"city": "Jakarta", "unit": "cm", "currency": "BIF", "phone": "(573) 569-6064", "email": "jbkk@test.org"}, {"city": "Jakarta", "unit": "km", "currency": "GTQ", "phone": "(583) 597-6403", "email": "plke@corp.io"}, {"city": "Santa Teresa", "unit": "mi", "currency": "GTQ", "phone": "(594) 536-5421", "email": "jphncha@example.com"}, {"city": "Kampong Dungun", "unit": "cm", "currency": "BIF", "phone": "(557) 475-9149", "email": "pjhk@test.org"}], "dirty_cols": ["city", "unit", "currency", "phone", "email"], "clean_cols": ["city", "unit", "currency", "phone", "email"], "plan": {"dataset_summary": "12 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"jakarta": "Jakarta", "kampong dungnu": "Kampong Dungun", "jakarsa": "Jakarta", "SANTA TERESA": "Santa Teresa"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MI": "mi", "Miles": "mi", "KILOMETERS": "km", "CM": "cm", "Mi": "mi", "CENTIMETERS": "cm", "centimeter": "cm"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Quetzal": "GTQ", "BURUNDI FRANC": "BIF", "gtq": "GTQ", "Burundi Franc": "BIF", "burundi fbanc": "BIF"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"country": "JE", "status": "P1", "company": "Hooli ", "city": " peicheng", "job_title": "Prod Mgr", "amount": "812,25", "email": "BLDDODM@MAIL.COM  "}, {"country": "TO ", "status": "Medium   ", "company": " Umbrella", "city": "Shahceheng", "job_title": "c.e.o.", "amount": "4.841,14", "email": "jmieg@mail.com"}, {"country": "   JE", "status": "  med", "company": "Globex", "city": " peicheng", "job_title": "Chief Executive Officer", "amount": "$5,806.69", "email": "HLMPEL@EXAMPLE.COM"}, {"country": "jey   ", "status": "med", "company": "  Vehement", "city": "Puerot Cabello", "job_title": "ceo", "amount": "3.265,84", "email": "klmjlag@test.org"}, {"country": " GIB ", "status": "  M", "company": "Hooli", "city": "Puerto Cabelo ", "job_title": "ceo", "amount": "7.331,25", "email": "   FDKBKPC@CORP.IO"}, {"country": "  TO", "status": "med", "company": "Acme Inc", "city": "Puerto Cabello", "job_title": "PM", "amount": "$7,183.52", "email": "kkip@corp.io"}, {"country": "Kingdom of Tonga", "status": "HIGH", "company": "Globex", "city": "   Puerto Cabello", "job_title": "ceo", "amount": "$8,653.87", "email": " KMED@CORP.IO "}, {"country": "gjbraltar", "status": "high   ", "company": "Vehement ", "city": "Shahecheng", "job_title": "C.E.O.", "amount": "$3,730.62", "email": "  kdgdj@corp.io"}, {"country": " Gi", "status": "M", "company": " Stark Industries", "city": "Puerto Cabello", "job_title": "PM", "amount": "3.432,25", "email": "hbnf@example.com"}, {"country": "To", "status": "meedium", "company": "  Wonka Co", "city": "Puerto Cabello", "job_title": "CEO", "amount": "$2,868.71", "email": "   opbplof@example.com"}, {"country": "JERSEY  ", "status": "mediium", "company": "Hooli", "city": "Peicheng", "job_title": "prod manager", "amount": "6.789,21", "email": " hnap@mail.com "}, {"country": "Jey", "status": "med", "company": "   Umbrella", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": "$3,759.57", "email": "efkjgj@corp.io"}, {"country": "JE", "status": " P2 ", "company": "Cyberdyne", "city": "Peicheng", "job_title": "pm", "amount": "$1,860.66", "email": "ldolf@corp.io"}, {"country": " GI ", "status": " medium ", "company": "Stark Industries", "city": "Puerto Cabello ", "job_title": "ceo", "amount": "330,44", "email": " eadjcl@corp.io "}, {"country": "   TO", "status": "M", "company": "  Hooli", "city": "Shahecheng", "job_title": "Produtc  Manager", "amount": "-", "email": "ifdn@test.org   "}, {"country": "Jersey", "status": "HIGH", "company": "Wonka  Co", "city": "shahecheng", "job_title": "   CHIEF EXECUTIVE OFFICER", "amount": "5.642,66", "email": "fhconn@mail.com"}, {"country": "Tonga", "status": "P1", "company": "Stark Industries ", "city": "peeicheng", "job_title": "Ceo", "amount": "464,57", "email": "ofnheh@test.org"}, {"country": "je", "status": "Medium", "company": " Umbrella ", "city": "peicheng", "job_title": "Chief Executive Officer", "amount": "0", "email": "PNBAK@TEST.ORG"}, {"country": "JEY", "status": "M", "company": "Wonka Co ", "city": "Shahecheng", "job_title": "PM", "amount": "4.957,58", "email": "lbapfc@test.org"}, {"country": "Gibraltar", "status": "high", "company": "Acme Inc  ", "city": "Peicheng", "job_title": "C.E.O.", "amount": "$8,376.65", "email": "hpjooma@test.org"}, {"country": "Kingdom Of Tonga", "status": "P1", "company": "Wonka Co", "city": "Shaehcheng", "job_title": "pm", "amount": "$2,345.01", "email": "  DONHO@EXAMPLE.COM"}, {"country": "GIB", "status": "p1", "company": "Soylent  Corp", "city": "Shahechenng", "job_title": " CEO ", "amount": "6.626,24", "email": "jmdk@corp.io"}, {"country": "TON", "status": "M", "company": "Stark Industries ", "city": "SHAHECHENG", "job_title": "PM", "amount": "null", "email": "MINIGKB@EXAMPLE.COM   "}, {"country": "Je", "status": " Medium ", "company": "Vehement", "city": "peicheng", "job_title": "C.E.O.", "amount": "$7,363.53", "email": "CCPNN@MAIL.COM"}, {"country": "gibrlatar", "status": " HIGH ", "company": "Stark Industries", "city": "Puerto Cabello", "job_title": "Prod Manager", "amount": "-", "email": "abejgge@mail.com "}, {"country": "TO", "status": "M", "company": "Initech", "city": "Puerto Cabello", "job_title": "C.E.O.", "amount": "247,59", "email": " hdfnm@corp.io "}, {"country": "  TON", "status": "  M", "company": "   Vehement", "city": "SHAHECHENG", "job_title": "chief executive officer  ", "amount": "-", "email": "   edaafpk@test.org"}, {"country": "  to", "status": " H ", "company": "Initech", "city": "   Shahecheng", "job_title": "C.E.O.", "amount": "$2,427.65", "email": "ECELDMK@CORP.IO"}, {"country": "JEY", "status": "P2", "company": "Cyberdyne   ", "city": " peicheng ", "job_title": "product manager", "amount": "1.613,82", "email": "chifmcd@example.com"}, {"country": "TONGA", "status": "M", "company": " Cyberdyne ", "city": "Puerto Cabello", "job_title": "chief executive officer", "amount": "2.867,84", "email": "neja@test.org"}, {"country": "tonga", "status": "MEDIUM ", "company": " Globex", "city": "Puerto Cabello", "job_title": "PRODDUCT MANAGER", "amount": "7.547,44", "email": " DFECHK@CORP.IO "}, {"country": "  gibraltar", "status": "medium", "company": " Umbrella", "city": "PUERTO CABELLO", "job_title": "Product Manager", "amount": "$1,021.80", "email": "codme@corp.io"}, {"country": "JEY", "status": "High", "company": "Soylent Corp   ", "city": "puerto caello", "job_title": "ceo", "amount": "2.487,38", "email": "GLJHI@CORP.IO"}, {"country": "  Tonga", "status": "M  ", "company": "Acme Inc", "city": "   Peichegn", "job_title": "Chief Executive Officer", "amount": "--", "email": "efnl@test.org  "}, {"country": "Gibralrar", "status": "H", "company": "Cyberdyne", "city": "SHAHECHENG", "job_title": "CEO", "amount": "$4,315.74", "email": "NMCF@EXAMPLE.COM"}, {"country": " TO ", "status": "high", "company": " Globex ", "city": "Puerto Cabello", "job_title": "pm", "amount": "5.927,02", "email": "IPCKFH@EXAMPLE.COM"}, {"country": "jey   ", "status": "med", "company": "  Vehement", "city": "Puerot Cabello", "job_title": "ceo", "amount": "3.265,84", "email": "klmjlag@test.org"}, {"country": " GIB ", "status": "  M", "company": "Hooli", "city": "Puerto Cabelo ", "job_title": "ceo", "amount": "7.331,25", "email": "   FDKBKPC@CORP.IO"}], "clean": [{"country": "Jersey", "status": "High", "company": "Hooli", "city": "Peicheng", "job_title": "Product Manager", "amount": 812.25, "email": "blddodm@mail.com"}, {"country": "Tonga", "status": "Medium", "company": "Umbrella", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": 4841.14, "email": "jmieg@mail.com"}, {"country": "Jersey", "status": "Medium", "company": "Globex", "city": "Peicheng", "job_title": "Chief Executive Officer", "amount": 5806.69, "email": "hlmpel@example.com"}, {"country": "Jersey", "status": "Medium", "company": "Vehement", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 3265.84, "email": "klmjlag@test.org"}, {"country": "Gibraltar", "status": "Medium", "company": "Hooli", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 7331.25, "email": "fdkbkpc@corp.io"}, {"country": "Tonga", "status": "Medium", "company": "Acme Inc", "city": "Puerto Cabello", "job_title": "Product Manager", "amount": 7183.52, "email": "kkip@corp.io"}, {"country": "Tonga", "status": "High", "company": "Globex", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 8653.87, "email": "kmed@corp.io"}, {"country": "Gibraltar", "status": "High", "company": "Vehement", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": 3730.62, "email": "kdgdj@corp.io"}, {"country": "Gibraltar", "status": "Medium", "company": "Stark Industries", "city": "Puerto Cabello", "job_title": "Product Manager", "amount": 3432.25, "email": "hbnf@example.com"}, {"country": "Tonga", "status": "Medium", "company": "Wonka Co", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 2868.71, "email": "opbplof@example.com"}, {"country": "Jersey", "status": "Medium", "company": "Hooli", "city": "Peicheng", "job_title": "Product Manager", "amount": 6789.21, "email": "hnap@mail.com"}, {"country": "Jersey", "status": "Medium", "company": "Umbrella", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 3759.57, "email": "efkjgj@corp.io"}, {"country": "Jersey", "status": "Medium", "company": "Cyberdyne", "city": "Peicheng", "job_title": "Product Manager", "amount": 1860.66, "email": "ldolf@corp.io"}, {"country": "Gibraltar", "status": "Medium", "company": "Stark Industries", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 330.44, "email": "eadjcl@corp.io"}, {"country": "Tonga", "status": "Medium", "company": "Hooli", "city": "Shahecheng", "job_title": "Product Manager", "amount": NaN, "email": "ifdn@test.org"}, {"country": "Jersey", "status": "High", "company": "Wonka Co", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": 5642.66, "email": "fhconn@mail.com"}, {"country": "Tonga", "status": "High", "company": "Stark Industries", "city": "Peicheng", "job_title": "Chief Executive Officer", "amount": 464.57, "email": "ofnheh@test.org"}, {"country": "Jersey", "status": "Medium", "company": "Umbrella", "city": "Peicheng", "job_title": "Chief Executive Officer", "amount": 0.0, "email": "pnbak@test.org"}, {"country": "Jersey", "status": "Medium", "company": "Wonka Co", "city": "Shahecheng", "job_title": "Product Manager", "amount": 4957.58, "email": "lbapfc@test.org"}, {"country": "Gibraltar", "status": "High", "company": "Acme Inc", "city": "Peicheng", "job_title": "Chief Executive Officer", "amount": 8376.65, "email": "hpjooma@test.org"}, {"country": "Tonga", "status": "High", "company": "Wonka Co", "city": "Shahecheng", "job_title": "Product Manager", "amount": 2345.01, "email": "donho@example.com"}, {"country": "Gibraltar", "status": "High", "company": "Soylent Corp", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": 6626.24, "email": "jmdk@corp.io"}, {"country": "Tonga", "status": "Medium", "company": "Stark Industries", "city": "Shahecheng", "job_title": "Product Manager", "amount": NaN, "email": "minigkb@example.com"}, {"country": "Jersey", "status": "Medium", "company": "Vehement", "city": "Peicheng", "job_title": "Chief Executive Officer", "amount": 7363.53, "email": "ccpnn@mail.com"}, {"country": "Gibraltar", "status": "High", "company": "Stark Industries", "city": "Puerto Cabello", "job_title": "Product Manager", "amount": NaN, "email": "abejgge@mail.com"}, {"country": "Tonga", "status": "Medium", "company": "Initech", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 247.59, "email": "hdfnm@corp.io"}, {"country": "Tonga", "status": "Medium", "company": "Vehement", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": NaN, "email": "edaafpk@test.org"}, {"country": "Tonga", "status": "High", "company": "Initech", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": 2427.65, "email": "eceldmk@corp.io"}, {"country": "Jersey", "status": "Medium", "company": "Cyberdyne", "city": "Peicheng", "job_title": "Product Manager", "amount": 1613.82, "email": "chifmcd@example.com"}, {"country": "Tonga", "status": "Medium", "company": "Cyberdyne", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 2867.84, "email": "neja@test.org"}, {"country": "Tonga", "status": "Medium", "company": "Globex", "city": "Puerto Cabello", "job_title": "Product Manager", "amount": 7547.44, "email": "dfechk@corp.io"}, {"country": "Gibraltar", "status": "Medium", "company": "Umbrella", "city": "Puerto Cabello", "job_title": "Product Manager", "amount": 1021.8, "email": "codme@corp.io"}, {"country": "Jersey", "status": "High", "company": "Soylent Corp", "city": "Puerto Cabello", "job_title": "Chief Executive Officer", "amount": 2487.38, "email": "gljhi@corp.io"}, {"country": "Tonga", "status": "Medium", "company": "Acme Inc", "city": "Peicheng", "job_title": "Chief Executive Officer", "amount": NaN, "email": "efnl@test.org"}, {"country": "Gibraltar", "status": "High", "company": "Cyberdyne", "city": "Shahecheng", "job_title": "Chief Executive Officer", "amount": 4315.74, "email": "nmcf@example.com"}, {"country": "Tonga", "status": "High", "company": "Globex", "city": "Puerto Cabello", "job_title": "Product Manager", "amount": 5927.02, "email": "ipckfh@example.com"}], "dirty_cols": ["country", "status", "company", "city", "job_title", "amount", "email"], "clean_cols": ["country", "status", "company", "city", "job_title", "amount", "email"], "plan": {"dataset_summary": "38 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"JE": "Jersey", "TO": "Tonga", "jey": "Jersey", "GIB": "Gibraltar", "Kingdom of Tonga": "Tonga", "gjbraltar": "Gibraltar", "Gi": "Gibraltar", "To": "Tonga", "JERSEY": "Jersey", "Jey": "Jersey", "GI": "Gibraltar", "je": "Jersey", "JEY": "Jersey", "Kingdom Of Tonga": "Tonga", "TON": "Tonga", "Je": "Jersey", "gibrlatar": "Gibraltar", "to": "Tonga", "TONGA": "Tonga", "tonga": "Tonga", "gibraltar": "Gibraltar", "Gibralrar": "Gibraltar"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P1": "High", "med": "Medium", "M": "Medium", "HIGH": "High", "high": "High", "meedium": "Medium", "mediium": "Medium", "P2": "Medium", "medium": "Medium", "p1": "High", "H": "High", "MEDIUM": "Medium"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"peicheng": "Peicheng", "Shahceheng": "Shahecheng", "Puerot Cabello": "Puerto Cabello", "Puerto Cabelo": "Puerto Cabello", "shahecheng": "Shahecheng", "peeicheng": "Peicheng", "Shaehcheng": "Shahecheng", "Shahechenng": "Shahecheng", "SHAHECHENG": "Shahecheng", "PUERTO CABELLO": "Puerto Cabello", "puerto caello": "Puerto Cabello", "Peichegn": "Peicheng"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Prod Mgr": "Product Manager", "c.e.o.": "Chief Executive Officer", "ceo": "Chief Executive Officer", "PM": "Product Manager", "C.E.O.": "Chief Executive Officer", "CEO": "Chief Executive Officer", "prod manager": "Product Manager", "pm": "Product Manager", "Produtc Manager": "Product Manager", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "Ceo": "Chief Executive Officer", "Prod Manager": "Product Manager", "chief executive officer": "Chief Executive Officer", "product manager": "Product Manager", "PRODDUCT MANAGER": "Product Manager"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"country": "Saint Martin French part", "amount": "2.318,77", "signup_date": "23 Sep 2023", "name": "Omar  Khan", "job_title": "AE", "notes2": ""}, {"country": "SHN", "amount": "3.671,82", "signup_date": "8 Aug 2023", "name": "Frank  Petrov", "job_title": "snr engineer", "notes2": ""}, {"country": " CZ", "amount": "$2,612.13", "signup_date": "45163", "name": "Ivan Petrov", "job_title": "Dir", "notes2": ""}, {"country": "Czech Republic", "amount": "1.595,88", "signup_date": "2023-08-12", "name": "Ivan Park", "job_title": "director", "notes2": ""}, {"country": "SH", "amount": "$454.32", "signup_date": "45285", "name": "  Karl Fischer", "job_title": "chief technology officer", "notes2": ""}, {"country": "saint helena, ascension and tristan da cunha", "amount": "#N/A", "signup_date": "5/23/2023", "name": "Eve  Cruz", "job_title": "Senior Enginneer", "notes2": ""}, {"country": "Saint Martin (French part)", "amount": "--", "signup_date": "4 Jan 2023", "name": "Bob Park", "job_title": "sr engineer", "notes2": ""}, {"country": "republic of poland ", "amount": "1.571,93", "signup_date": "1/6/2023", "name": "Omar Novak", "job_title": "SENIOR ENG", "notes2": ""}, {"country": "POL", "amount": "$1,176.42", "signup_date": "8/17/2023", "name": "Omar Moore ", "job_title": "Senior Eng", "notes2": ""}, {"country": "shn", "amount": "3.310,12", "signup_date": "2023-11-01", "name": " Sara Novak", "job_title": "Director", "notes2": ""}, {"country": "pol", "amount": "None", "signup_date": "6/7/2023", "name": "  Carol Novak", "job_title": "Acct Exec", "notes2": ""}, {"country": "shn", "amount": "5.511,09", "signup_date": "2023-05-27", "name": "Heidi Reyes", "job_title": " C.T.O.", "notes2": ""}, {"country": "Czechia", "amount": "6.746,39", "signup_date": "10/14/2023", "name": "  Priya Brandt", "job_title": "C.T.O.", "notes2": ""}, {"country": "czechia", "amount": "$7,826.20", "signup_date": "45212", "name": "  Mona Brandt", "job_title": "CTO", "notes2": ""}, {"country": "SH", "amount": "5.915,36", "signup_date": "12 Mar 2023", "name": "David Lee", "job_title": "chief technology officer  ", "notes2": ""}, {"country": "Czech Republic", "amount": "3.664,27", "signup_date": "3/13/2023", "name": "Lena Cruz", "job_title": "Dir.", "notes2": ""}, {"country": "  maf", "amount": "$2,687.44", "signup_date": "45187", "name": "Ivan Park", "job_title": "Dir", "notes2": ""}, {"country": "CZE ", "amount": "$8,923.88", "signup_date": "5/8/2023", "name": "Bob Wong", "job_title": "Senior Eng", "notes2": ""}, {"country": "Saint Martin (French Part)", "amount": "4.572,08", "signup_date": "2023-11-02", "name": "Alice Lee   ", "job_title": "Account Executive", "notes2": ""}, {"country": "PL", "amount": "2.059,63", "signup_date": "23 Nov 2023", "name": "Omar Ali", "job_title": "Senior Engineer", "notes2": ""}, {"country": "maf", "amount": "2.967,12", "signup_date": "2023-05-28", "name": "Sara Fischer", "job_title": "Senior Engineer", "notes2": ""}, {"country": "SHN", "amount": "$2,593.93", "signup_date": "45164", "name": " Carol Smith", "job_title": "CTO", "notes2": ""}, {"country": "pol", "amount": "3.057,29", "signup_date": "19 Dec 2023", "name": "  Lena Lee", "job_title": "ACCOUNT EXECUTIVE", "notes2": ""}, {"country": "Saint Martin (French part)", "amount": "$4,587.52", "signup_date": "2/14/2023", "name": "Sara Brandt", "job_title": " AE ", "notes2": ""}, {"country": "Cz", "amount": "966,75", "signup_date": "9 Sep 2023", "name": "Heidi Smith", "job_title": "Dir", "notes2": ""}, {"country": " sh ", "amount": "4.843,18", "signup_date": "2023-10-06", "name": "  Eve Ali", "job_title": "chief technology officer ", "notes2": ""}, {"country": "republic of poland", "amount": "3.401,79", "signup_date": "3/25/2023", "name": " Frank Novak", "job_title": "C.T.O.", "notes2": ""}, {"country": "   SHN", "amount": "1.406,16", "signup_date": "45113", "name": "Omar Smith  ", "job_title": "DIRECTOR", "notes2": ""}, {"country": "SH", "amount": "$1,428.44", "signup_date": "2023-02-07", "name": "Heidi  Khan", "job_title": "SENIOR ENGINEER", "notes2": ""}, {"country": " Cze ", "amount": "$2,198.69", "signup_date": "18 Aug 2023", "name": "Judy Johnson   ", "job_title": "Sr. Engineer", "notes2": ""}, {"country": "  maf", "amount": "3.011,67", "signup_date": "9/26/2023", "name": "Ivan Brandt", "job_title": "Account Executive", "notes2": ""}, {"country": "SH", "amount": "5.637,07", "signup_date": "9 Aug 2023", "name": "Carol Park ", "job_title": "Dir", "notes2": ""}, {"country": "saint martin (french part)", "amount": "382,30", "signup_date": "2023-03-18", "name": "Alice Cruz", "job_title": "senior eng", "notes2": ""}, {"country": "saint martin (french part)", "amount": "$2,641.76", "signup_date": "2023-07-17", "name": "Karl Novak", "job_title": "CTO", "notes2": ""}, {"country": " CZ ", "amount": "?", "signup_date": "11/5/2023", "name": "Carol  Diaz", "job_title": "Sr. Engineer", "notes2": ""}, {"country": "SH", "amount": "$454.32", "signup_date": "45285", "name": "  Karl Fischer", "job_title": "chief technology officer", "notes2": ""}, {"country": "republic of poland", "amount": "3.401,79", "signup_date": "3/25/2023", "name": " Frank Novak", "job_title": "C.T.O.", "notes2": ""}], "clean": [{"country": "Saint Martin (French part)", "amount": 2318.77, "signup_date": "2023-09-23", "name": "Omar Khan", "job_title": "Account Executive"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 3671.82, "signup_date": "2023-08-08", "name": "Frank Petrov", "job_title": "Senior Engineer"}, {"country": "Czechia", "amount": 2612.13, "signup_date": "2023-08-25", "name": "Ivan Petrov", "job_title": "Director"}, {"country": "Czechia", "amount": 1595.88, "signup_date": "2023-08-12", "name": "Ivan Park", "job_title": "Director"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 454.32, "signup_date": "2023-12-25", "name": "Karl Fischer", "job_title": "Chief Technology Officer"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": NaN, "signup_date": "2023-05-23", "name": "Eve Cruz", "job_title": "Senior Engineer"}, {"country": "Saint Martin (French part)", "amount": NaN, "signup_date": "2023-01-04", "name": "Bob Park", "job_title": "Senior Engineer"}, {"country": "Poland", "amount": 1571.93, "signup_date": "2023-01-06", "name": "Omar Novak", "job_title": "Senior Engineer"}, {"country": "Poland", "amount": 1176.42, "signup_date": "2023-08-17", "name": "Omar Moore", "job_title": "Senior Engineer"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 3310.12, "signup_date": "2023-11-01", "name": "Sara Novak", "job_title": "Director"}, {"country": "Poland", "amount": NaN, "signup_date": "2023-06-07", "name": "Carol Novak", "job_title": "Account Executive"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 5511.09, "signup_date": "2023-05-27", "name": "Heidi Reyes", "job_title": "Chief Technology Officer"}, {"country": "Czechia", "amount": 6746.39, "signup_date": "2023-10-14", "name": "Priya Brandt", "job_title": "Chief Technology Officer"}, {"country": "Czechia", "amount": 7826.2, "signup_date": "2023-10-13", "name": "Mona Brandt", "job_title": "Chief Technology Officer"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 5915.36, "signup_date": "2023-03-12", "name": "David Lee", "job_title": "Chief Technology Officer"}, {"country": "Czechia", "amount": 3664.27, "signup_date": "2023-03-13", "name": "Lena Cruz", "job_title": "Director"}, {"country": "Saint Martin (French part)", "amount": 2687.44, "signup_date": "2023-09-18", "name": "Ivan Park", "job_title": "Director"}, {"country": "Czechia", "amount": 8923.88, "signup_date": "2023-05-08", "name": "Bob Wong", "job_title": "Senior Engineer"}, {"country": "Saint Martin (French part)", "amount": 4572.08, "signup_date": "2023-11-02", "name": "Alice Lee", "job_title": "Account Executive"}, {"country": "Poland", "amount": 2059.63, "signup_date": "2023-11-23", "name": "Omar Ali", "job_title": "Senior Engineer"}, {"country": "Saint Martin (French part)", "amount": 2967.12, "signup_date": "2023-05-28", "name": "Sara Fischer", "job_title": "Senior Engineer"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 2593.93, "signup_date": "2023-08-26", "name": "Carol Smith", "job_title": "Chief Technology Officer"}, {"country": "Poland", "amount": 3057.29, "signup_date": "2023-12-19", "name": "Lena Lee", "job_title": "Account Executive"}, {"country": "Saint Martin (French part)", "amount": 4587.52, "signup_date": "2023-02-14", "name": "Sara Brandt", "job_title": "Account Executive"}, {"country": "Czechia", "amount": 966.75, "signup_date": "2023-09-09", "name": "Heidi Smith", "job_title": "Director"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 4843.18, "signup_date": "2023-10-06", "name": "Eve Ali", "job_title": "Chief Technology Officer"}, {"country": "Poland", "amount": 3401.79, "signup_date": "2023-03-25", "name": "Frank Novak", "job_title": "Chief Technology Officer"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 1406.16, "signup_date": "2023-07-06", "name": "Omar Smith", "job_title": "Director"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 1428.44, "signup_date": "2023-02-07", "name": "Heidi Khan", "job_title": "Senior Engineer"}, {"country": "Czechia", "amount": 2198.69, "signup_date": "2023-08-18", "name": "Judy Johnson", "job_title": "Senior Engineer"}, {"country": "Saint Martin (French part)", "amount": 3011.67, "signup_date": "2023-09-26", "name": "Ivan Brandt", "job_title": "Account Executive"}, {"country": "Saint Helena, Ascension and Tristan da Cunha", "amount": 5637.07, "signup_date": "2023-08-09", "name": "Carol Park", "job_title": "Director"}, {"country": "Saint Martin (French part)", "amount": 382.3, "signup_date": "2023-03-18", "name": "Alice Cruz", "job_title": "Senior Engineer"}, {"country": "Saint Martin (French part)", "amount": 2641.76, "signup_date": "2023-07-17", "name": "Karl Novak", "job_title": "Chief Technology Officer"}, {"country": "Czechia", "amount": NaN, "signup_date": "2023-11-05", "name": "Carol Diaz", "job_title": "Senior Engineer"}], "dirty_cols": ["country", "amount", "signup_date", "name", "job_title", "notes2"], "clean_cols": ["country", "amount", "signup_date", "name", "job_title"], "plan": {"dataset_summary": "37 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Saint Martin French part": "Saint Martin (French part)", "SHN": "Saint Helena, Ascension and Tristan da Cunha", "CZ": "Czechia", "Czech Republic": "Czechia", "SH": "Saint Helena, Ascension and Tristan da Cunha", "saint helena, ascension and tristan da cunha": "Saint Helena, Ascension and Tristan da Cunha", "republic of poland": "Poland", "POL": "Poland", "shn": "Saint Helena, Ascension and Tristan da Cunha", "pol": "Poland", "czechia": "Czechia", "maf": "Saint Martin (French part)", "CZE": "Czechia", "Saint Martin (French Part)": "Saint Martin (French part)", "PL": "Poland", "Cz": "Czechia", "sh": "Saint Helena, Ascension and Tristan da Cunha", "Cze": "Czechia", "saint martin (french part)": "Saint Martin (French part)"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AE": "Account Executive", "snr engineer": "Senior Engineer", "Dir": "Director", "director": "Director", "chief technology officer": "Chief Technology Officer", "Senior Enginneer": "Senior Engineer", "sr engineer": "Senior Engineer", "SENIOR ENG": "Senior Engineer", "Senior Eng": "Senior Engineer", "Acct Exec": "Account Executive", "C.T.O.": "Chief Technology Officer", "CTO": "Chief Technology Officer", "Dir.": "Director", "ACCOUNT EXECUTIVE": "Account Executive", "DIRECTOR": "Director", "SENIOR ENGINEER": "Senior Engineer", "Sr. Engineer": "Senior Engineer", "senior eng": "Senior Engineer"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "5.713,15", "is_active": "Y", "status": "Active   ", "country": "bz", "industry": "real  estate", "name": " Bob Adams", "unnamed": ""}, {"amount": "3.771,23", "is_active": "true", "status": "paused", "country": "  can", "industry": "Telecommunications", "name": "Carol  Lee", "unnamed": ""}, {"amount": "7.807,14", "is_active": "TRUE", "status": "paused", "country": "Republic of Belarus", "industry": "RealEstate", "name": "Bob Diaz   ", "unnamed": ""}, {"amount": "-", "is_active": "true", "status": "paused   ", "country": "BLZ", "industry": "RealEstate", "name": "Bob Diaz", "unnamed": ""}, {"amount": "$4,523.03", "is_active": "No", "status": "On Hold", "country": " BLZ", "industry": " property", "name": "Omar Smith   ", "unnamed": ""}, {"amount": "-", "is_active": "TRUE", "status": "active", "country": "bulgaria", "industry": "health", "name": "Sara Moore", "unnamed": ""}, {"amount": "$3,254.25", "is_active": "T", "status": "ACTIVE", "country": " BLZ ", "industry": "Real Esyate", "name": " Omar Brandt", "unnamed": ""}, {"amount": "542,11", "is_active": "false", "status": " On Hold", "country": "BGR", "industry": "Health Care", "name": "  Judy Smith", "unnamed": ""}, {"amount": "$2,828.08", "is_active": "true", "status": " Active ", "country": "BGR", "industry": "medical", "name": "Grace Petrov ", "unnamed": ""}, {"amount": "6.783,27", "is_active": "T", "status": " ACTIVE", "country": "Republic Of Bulgaria", "industry": "Heatlh  Care", "name": "Heidi Diaz", "unnamed": ""}, {"amount": "$6,946.19", "is_active": "TRUE", "status": "active", "country": "Can", "industry": "healthcare", "name": "Judy Lee", "unnamed": ""}, {"amount": "$1,938.21", "is_active": "FALSE", "status": "ACTIVE", "country": "BELIZE", "industry": "Protech", "name": "Grace Khan", "unnamed": ""}, {"amount": "6.306,54", "is_active": "false", "status": "PAUSED", "country": "BG", "industry": "Comms", "name": "Alice Diaz", "unnamed": ""}, {"amount": "$6,836.96", "is_active": "T", "status": "Active", "country": "Belarus", "industry": "Medical", "name": "Lena Ali", "unnamed": ""}, {"amount": "5.637,60", "is_active": "true", "status": "Paused", "country": "Republic of Belarus ", "industry": "PROPTECH", "name": "Judy  Novak", "unnamed": ""}, {"amount": "$4,493.91", "is_active": "N", "status": "active", "country": "BLZ", "industry": "TELCO", "name": "Karl Ali", "unnamed": ""}, {"amount": "$70.64", "is_active": "F", "status": "Paused ", "country": "CAN", "industry": "Healthcare", "name": "Ivan Wong   ", "unnamed": ""}, {"amount": "$2,079.77", "is_active": "false", "status": "Paused  ", "country": "Bulgaria", "industry": "PropTech   ", "name": "Eve Diaz", "unnamed": ""}, {"amount": "815,59", "is_active": "T", "status": "active", "country": "PCN", "industry": " telco ", "name": "  Karl Petrov", "unnamed": ""}, {"amount": "1.729,67", "is_active": "Yes", "status": "On Hold ", "country": "BLZ", "industry": "property", "name": "Mona  Brandt", "unnamed": ""}, {"amount": "$1,938.21", "is_active": "FALSE", "status": "ACTIVE", "country": "BELIZE", "industry": "Protech", "name": "Grace Khan", "unnamed": ""}, {"amount": "", "is_active": "", "status": "", "country": "", "industry": "", "name": "", "unnamed": ""}], "clean": [{"amount": 5713.15, "is_active": true, "status": "Active", "country": "Belize", "industry": "Real Estate", "name": "Bob Adams"}, {"amount": 3771.23, "is_active": true, "status": "Paused", "country": "Canada", "industry": "Telecommunications", "name": "Carol Lee"}, {"amount": 7807.14, "is_active": true, "status": "Paused", "country": "Belarus", "industry": "Real Estate", "name": "Bob Diaz"}, {"amount": NaN, "is_active": true, "status": "Paused", "country": "Belize", "industry": "Real Estate", "name": "Bob Diaz"}, {"amount": 4523.03, "is_active": false, "status": "Paused", "country": "Belize", "industry": "Real Estate", "name": "Omar Smith"}, {"amount": NaN, "is_active": true, "status": "Active", "country": "Bulgaria", "industry": "Healthcare", "name": "Sara Moore"}, {"amount": 3254.25, "is_active": true, "status": "Active", "country": "Belize", "industry": "Real Estate", "name": "Omar Brandt"}, {"amount": 542.11, "is_active": false, "status": "Paused", "country": "Bulgaria", "industry": "Healthcare", "name": "Judy Smith"}, {"amount": 2828.08, "is_active": true, "status": "Active", "country": "Bulgaria", "industry": "Healthcare", "name": "Grace Petrov"}, {"amount": 6783.27, "is_active": true, "status": "Active", "country": "Bulgaria", "industry": "Healthcare", "name": "Heidi Diaz"}, {"amount": 6946.19, "is_active": true, "status": "Active", "country": "Canada", "industry": "Healthcare", "name": "Judy Lee"}, {"amount": 1938.21, "is_active": false, "status": "Active", "country": "Belize", "industry": "Real Estate", "name": "Grace Khan"}, {"amount": 6306.54, "is_active": false, "status": "Paused", "country": "Bulgaria", "industry": "Telecommunications", "name": "Alice Diaz"}, {"amount": 6836.96, "is_active": true, "status": "Active", "country": "Belarus", "industry": "Healthcare", "name": "Lena Ali"}, {"amount": 5637.6, "is_active": true, "status": "Paused", "country": "Belarus", "industry": "Real Estate", "name": "Judy Novak"}, {"amount": 4493.91, "is_active": false, "status": "Active", "country": "Belize", "industry": "Telecommunications", "name": "Karl Ali"}, {"amount": 70.64, "is_active": false, "status": "Paused", "country": "Canada", "industry": "Healthcare", "name": "Ivan Wong"}, {"amount": 2079.77, "is_active": false, "status": "Paused", "country": "Bulgaria", "industry": "Real Estate", "name": "Eve Diaz"}, {"amount": 815.59, "is_active": true, "status": "Active", "country": "Pitcairn", "industry": "Telecommunications", "name": "Karl Petrov"}, {"amount": 1729.67, "is_active": true, "status": "Paused", "country": "Belize", "industry": "Real Estate", "name": "Mona Brandt"}], "dirty_cols": ["amount", "is_active", "status", "country", "industry", "name", "unnamed"], "clean_cols": ["amount", "is_active", "status", "country", "industry", "name"], "plan": {"dataset_summary": "22 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paused": "Paused", "On Hold": "Paused", "active": "Active", "ACTIVE": "Active", "PAUSED": "Paused"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bz": "Belize", "can": "Canada", "Republic of Belarus": "Belarus", "BLZ": "Belize", "bulgaria": "Bulgaria", "BGR": "Bulgaria", "Republic Of Bulgaria": "Bulgaria", "Can": "Canada", "BELIZE": "Belize", "BG": "Bulgaria", "CAN": "Canada", "PCN": "Pitcairn"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"real estate": "Real Estate", "RealEstate": "Real Estate", "property": "Real Estate", "health": "Healthcare", "Real Esyate": "Real Estate", "Health Care": "Healthcare", "medical": "Healthcare", "Heatlh Care": "Healthcare", "healthcare": "Healthcare", "Protech": "Real Estate", "Comms": "Telecommunications", "Medical": "Healthcare", "PROPTECH": "Real Estate", "TELCO": "Telecommunications", "PropTech": "Real Estate", "telco": "Telecommunications"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"rate": "69.0%", "country": "liberia", "status": "low", "is_active": "0", "email": "  BDJBMMO@EXAMPLE.COM", "extra": ""}, {"rate": "32.6%", "country": "za", "status": "P2", "is_active": "1", "email": "AJDBFL@CORP.IO   ", "extra": ""}, {"rate": "54.4%", "country": "LBR", "status": "MED", "is_active": "FALSE", "email": " lclol@corp.io ", "extra": ""}, {"rate": "13.0%", "country": "LBR ", "status": " low", "is_active": "false", "email": "  faganfo@test.org", "extra": ""}, {"rate": "32.2%", "country": "Za", "status": "P3", "is_active": "TRUE", "email": "oknkck@test.org", "extra": ""}, {"rate": "11.0%", "country": "south africa", "status": "low", "is_active": "0", "email": "OHKDPG@CORP.IO", "extra": ""}, {"rate": "93.5%", "country": "REPUBLIC OF LIBERIA", "status": "medium", "is_active": "1", "email": "PJMLIA@TEST.ORG  ", "extra": ""}, {"rate": "78.5%", "country": "  ZA", "status": "M   ", "is_active": "No", "email": "bonimol@mail.com", "extra": ""}, {"rate": "46.8%", "country": "Republic of Liberia", "status": "low", "is_active": "false", "email": "   lbipkg@test.org", "extra": ""}, {"rate": "33.3%", "country": "liberia", "status": "Low", "is_active": "Y", "email": " DHDKNA@MAIL.COM ", "extra": ""}, {"rate": "42.1%", "country": "Liberia", "status": "P2", "is_active": "N", "email": "iejd@test.org", "extra": ""}, {"rate": "98.4%", "country": "South Africa", "status": "medium", "is_active": "No", "email": "dbkm@mail.com", "extra": ""}, {"rate": "49.9%", "country": "LBR", "status": "low", "is_active": "true", "email": "EABLAIF@TEST.ORG", "extra": ""}, {"rate": "75.9%", "country": " South Africa", "status": "med", "is_active": "No", "email": "DPNMOCB@TEST.ORG", "extra": ""}, {"rate": "90.9%", "country": " ZA ", "status": "l", "is_active": "false", "email": "PJFKBA@EXAMPLE.COM", "extra": ""}, {"rate": "55.5%", "country": "Republic of South Africa", "status": "low", "is_active": "F", "email": "JEODL@EXAMPLE.COM", "extra": ""}, {"rate": "85.5%", "country": "zaf", "status": "MEDIUM", "is_active": "Y", "email": "OENDMCG@MAIL.COM", "extra": ""}, {"rate": "40.9%", "country": " south africa", "status": "Low ", "is_active": "1", "email": "KJAJNP@EXAMPLE.COM", "extra": ""}, {"rate": "88.0%", "country": "republic of liberia ", "status": "P3", "is_active": "false", "email": "DIILMEN@CORP.IO", "extra": ""}, {"rate": "48.4%", "country": "lbr", "status": "Medium", "is_active": "Y", "email": "IBFEAME@EXAMPLE.COM", "extra": ""}, {"rate": "55.5%", "country": "Republic of Lberia", "status": "P3", "is_active": "T", "email": "hlfa@test.org", "extra": ""}, {"rate": "96.1%", "country": "zaf", "status": "M", "is_active": "F", "email": "KCGGFAO@CORP.IO", "extra": ""}, {"rate": "", "country": "", "status": "", "is_active": "", "email": "", "extra": ""}], "clean": [{"rate": 0.69, "country": "Liberia", "status": "Low", "is_active": false, "email": "bdjbmmo@example.com"}, {"rate": 0.326, "country": "South Africa", "status": "Medium", "is_active": true, "email": "ajdbfl@corp.io"}, {"rate": 0.544, "country": "Liberia", "status": "Medium", "is_active": false, "email": "lclol@corp.io"}, {"rate": 0.13, "country": "Liberia", "status": "Low", "is_active": false, "email": "faganfo@test.org"}, {"rate": 0.322, "country": "South Africa", "status": "Low", "is_active": true, "email": "oknkck@test.org"}, {"rate": 0.11, "country": "South Africa", "status": "Low", "is_active": false, "email": "ohkdpg@corp.io"}, {"rate": 0.935, "country": "Liberia", "status": "Medium", "is_active": true, "email": "pjmlia@test.org"}, {"rate": 0.785, "country": "South Africa", "status": "Medium", "is_active": false, "email": "bonimol@mail.com"}, {"rate": 0.46799999999999997, "country": "Liberia", "status": "Low", "is_active": false, "email": "lbipkg@test.org"}, {"rate": 0.33299999999999996, "country": "Liberia", "status": "Low", "is_active": true, "email": "dhdkna@mail.com"}, {"rate": 0.42100000000000004, "country": "Liberia", "status": "Medium", "is_active": false, "email": "iejd@test.org"}, {"rate": 0.9840000000000001, "country": "South Africa", "status": "Medium", "is_active": false, "email": "dbkm@mail.com"}, {"rate": 0.499, "country": "Liberia", "status": "Low", "is_active": true, "email": "eablaif@test.org"}, {"rate": 0.759, "country": "South Africa", "status": "Medium", "is_active": false, "email": "dpnmocb@test.org"}, {"rate": 0.909, "country": "South Africa", "status": "Low", "is_active": false, "email": "pjfkba@example.com"}, {"rate": 0.555, "country": "South Africa", "status": "Low", "is_active": false, "email": "jeodl@example.com"}, {"rate": 0.855, "country": "South Africa", "status": "Medium", "is_active": true, "email": "oendmcg@mail.com"}, {"rate": 0.409, "country": "South Africa", "status": "Low", "is_active": true, "email": "kjajnp@example.com"}, {"rate": 0.88, "country": "Liberia", "status": "Low", "is_active": false, "email": "diilmen@corp.io"}, {"rate": 0.484, "country": "Liberia", "status": "Medium", "is_active": true, "email": "ibfeame@example.com"}, {"rate": 0.555, "country": "Liberia", "status": "Low", "is_active": true, "email": "hlfa@test.org"}, {"rate": 0.961, "country": "South Africa", "status": "Medium", "is_active": false, "email": "kcggfao@corp.io"}], "dirty_cols": ["rate", "country", "status", "is_active", "email", "extra"], "clean_cols": ["rate", "country", "status", "is_active", "email"], "plan": {"dataset_summary": "23 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liberia": "Liberia", "za": "South Africa", "LBR": "Liberia", "Za": "South Africa", "south africa": "South Africa", "REPUBLIC OF LIBERIA": "Liberia", "ZA": "South Africa", "Republic of Liberia": "Liberia", "Republic of South Africa": "South Africa", "zaf": "South Africa", "republic of liberia": "Liberia", "lbr": "Liberia", "Republic of Lberia": "Liberia"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"low": "Low", "P2": "Medium", "MED": "Medium", "P3": "Low", "medium": "Medium", "M": "Medium", "med": "Medium", "l": "Low", "MEDIUM": "Medium"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"phone": "538.504.1592", "city": "INGELHEIM AM RHEIN", "email": "lpcflip@mail.com", "name": "Bob Diaz   ", "state": " IN"}, {"phone": "504-758-4673", "city": "  Tizi-n-Tleta", "email": " BLHJB@CORP.IO ", "name": "Ivan Smith", "state": "NEF JERSEY"}, {"phone": "5272287168", "city": "Ingelheim am Rhein", "email": "   EFEEDP@MAIL.COM", "name": "Karl  Lee", "state": "IN"}, {"phone": "(515)949-9201", "city": " tzii-n-tleta ", "email": "pgkip@test.org", "name": "Priya Moore", "state": "New Jersey"}, {"phone": "(587)616-0747", "city": "Adelaide", "email": "ocpaphc@test.org", "name": "Omar  Park", "state": " FL "}, {"phone": "(524)161-6146", "city": "INGELHEIM AM RHEIN", "email": "dmgalf@test.org", "name": "David Adams", "state": "florida"}, {"phone": "(598)109-9468", "city": "Tizi-n-Tleta", "email": "CPJGHP@EXAMPLE.COM", "name": "Bob Wong", "state": "fl"}, {"phone": "5605966083", "city": "Ingelheim Am Rhein ", "email": "NJLJIB@TEST.ORG", "name": "Judy Ali", "state": "in "}, {"phone": "5627197507", "city": "Adwlaide", "email": "EDHEK@CORP.IO ", "name": "Sara Khan", "state": "KY"}, {"phone": "539-831-6081", "city": "   adelaide", "email": " mfoh@test.org ", "name": "  Frank Brandt", "state": "FL"}, {"phone": "576-965-5691", "city": "ADELAIDE", "email": "OEFFPDJ@MAIL.COM", "name": "Carol Park", "state": "new jersey"}, {"phone": "590-832-0004", "city": "ADELAIDE   ", "email": "jkmho@mail.com", "name": "Judy Brandt", "state": "IN "}, {"phone": "502.070.1445", "city": "Adelaide", "email": "midadhe@test.org", "name": "Sara Johnson  ", "state": " Kentucky "}, {"phone": "5489540348", "city": "Tizi-n-Tleta", "email": "   DNFF@EXAMPLE.COM", "name": "Frank Diaz", "state": "Florida"}, {"phone": "576.496.9320", "city": "Adelaide", "email": "lcba@test.org  ", "name": " Omar Lee", "state": "Ky"}, {"phone": "514.855.9056", "city": "Adelaide ", "email": "fnhnkp@example.com", "name": "Priya Petrov", "state": "IN"}, {"phone": "581-535-3658", "city": "Ingelheim am Rhein", "email": " famo@mail.com ", "name": "   Frank Fischer", "state": "Kentucky"}, {"phone": "522-900-2672", "city": "tizi-n-tleta", "email": "HMCLA@CORP.IO   ", "name": "Ivan Brandt", "state": "NJ"}, {"phone": "5605966083", "city": "Ingelheim Am Rhein ", "email": "NJLJIB@TEST.ORG", "name": "Judy Ali", "state": "in "}], "clean": [{"phone": "(538) 504-1592", "city": "Ingelheim am Rhein", "email": "lpcflip@mail.com", "name": "Bob Diaz", "state": "Indiana"}, {"phone": "(504) 758-4673", "city": "Tizi-n-Tleta", "email": "blhjb@corp.io", "name": "Ivan Smith", "state": "New Jersey"}, {"phone": "(527) 228-7168", "city": "Ingelheim am Rhein", "email": "efeedp@mail.com", "name": "Karl Lee", "state": "Indiana"}, {"phone": "(515) 949-9201", "city": "Tizi-n-Tleta", "email": "pgkip@test.org", "name": "Priya Moore", "state": "New Jersey"}, {"phone": "(587) 616-0747", "city": "Adelaide", "email": "ocpaphc@test.org", "name": "Omar Park", "state": "Florida"}, {"phone": "(524) 161-6146", "city": "Ingelheim am Rhein", "email": "dmgalf@test.org", "name": "David Adams", "state": "Florida"}, {"phone": "(598) 109-9468", "city": "Tizi-n-Tleta", "email": "cpjghp@example.com", "name": "Bob Wong", "state": "Florida"}, {"phone": "(560) 596-6083", "city": "Ingelheim am Rhein", "email": "njljib@test.org", "name": "Judy Ali", "state": "Indiana"}, {"phone": "(562) 719-7507", "city": "Adelaide", "email": "edhek@corp.io", "name": "Sara Khan", "state": "Kentucky"}, {"phone": "(539) 831-6081", "city": "Adelaide", "email": "mfoh@test.org", "name": "Frank Brandt", "state": "Florida"}, {"phone": "(576) 965-5691", "city": "Adelaide", "email": "oeffpdj@mail.com", "name": "Carol Park", "state": "New Jersey"}, {"phone": "(590) 832-0004", "city": "Adelaide", "email": "jkmho@mail.com", "name": "Judy Brandt", "state": "Indiana"}, {"phone": "(502) 070-1445", "city": "Adelaide", "email": "midadhe@test.org", "name": "Sara Johnson", "state": "Kentucky"}, {"phone": "(548) 954-0348", "city": "Tizi-n-Tleta", "email": "dnff@example.com", "name": "Frank Diaz", "state": "Florida"}, {"phone": "(576) 496-9320", "city": "Adelaide", "email": "lcba@test.org", "name": "Omar Lee", "state": "Kentucky"}, {"phone": "(514) 855-9056", "city": "Adelaide", "email": "fnhnkp@example.com", "name": "Priya Petrov", "state": "Indiana"}, {"phone": "(581) 535-3658", "city": "Ingelheim am Rhein", "email": "famo@mail.com", "name": "Frank Fischer", "state": "Kentucky"}, {"phone": "(522) 900-2672", "city": "Tizi-n-Tleta", "email": "hmcla@corp.io", "name": "Ivan Brandt", "state": "New Jersey"}], "dirty_cols": ["phone", "city", "email", "name", "state"], "clean_cols": ["phone", "city", "email", "name", "state"], "plan": {"dataset_summary": "19 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"INGELHEIM AM RHEIN": "Ingelheim am Rhein", "tzii-n-tleta": "Tizi-n-Tleta", "Ingelheim Am Rhein": "Ingelheim am Rhein", "Adwlaide": "Adelaide", "adelaide": "Adelaide", "ADELAIDE": "Adelaide", "tizi-n-tleta": "Tizi-n-Tleta"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IN": "Indiana", "NEF JERSEY": "New Jersey", "FL": "Florida", "florida": "Florida", "fl": "Florida", "in": "Indiana", "KY": "Kentucky", "new jersey": "New Jersey", "Ky": "Kentucky", "NJ": "New Jersey"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "?", "city": "San Pedro de Jujuy", "currency": "Hong Koong Dollar", "company": "Vehement", "status": "open", "notes2": ""}, {"amount": "$667.26", "city": "Ner", "currency": "HKD", "company": "Wonka Co", "status": "CLOSED WON", "notes2": ""}, {"amount": "2.481,72", "city": "Toronto", "currency": " HKD ", "company": "Hooli", "status": "  Open", "notes2": ""}, {"amount": "null", "city": "Povorino", "currency": "etb", "company": "Soylent Corp", "status": " open ", "notes2": ""}, {"amount": "$7,961.84", "city": "toronto", "currency": "ethiopian birr", "company": " Initech ", "status": "Closed-Won", "notes2": ""}, {"amount": "5.869,93", "city": "Povorino", "currency": "ETB", "company": "Wonka Co", "status": "LOST", "notes2": ""}, {"amount": "$7,171.03", "city": "Ner   ", "currency": "hong kong dollar", "company": "Cyberdyne", "status": "open", "notes2": ""}, {"amount": "2.351,63", "city": "San Pedro de Jujuy", "currency": "ETB ", "company": "Stark Industries", "status": "lost  ", "notes2": ""}, {"amount": "8.445,08", "city": "Toronto", "currency": "ETHIOPIAN BIRR", "company": "Umbrella", "status": " new ", "notes2": ""}, {"amount": "None", "city": "ner ", "currency": "  ETB", "company": " Hooli ", "status": "open", "notes2": ""}, {"amount": "2.892,86", "city": "Ner", "currency": " hkd ", "company": "Umbrella", "status": "  won", "notes2": ""}, {"amount": "$2,320.68", "city": " San Pedro de Jujuy", "currency": "   hong  kong dollar", "company": "Umbrella", "status": "open", "notes2": ""}, {"amount": "2.368,52", "city": "SAN PEDRO DE JUJUY", "currency": "  Ethiopian Birr", "company": "Acme Inc", "status": " open ", "notes2": ""}, {"amount": "3.077,02", "city": "Toronto", "currency": "Ethiopian Birr", "company": "  Initech", "status": "closed-won", "notes2": ""}, {"amount": "2.957,09", "city": "Povorino", "currency": "Hkd", "company": "Acme Inc  ", "status": "lost", "notes2": ""}, {"amount": "7.540,80", "city": "   Toronot", "currency": "HKD", "company": "Globex", "status": "won", "notes2": ""}, {"amount": "$1,744.86", "city": "ner  ", "currency": "  Etb", "company": "  Cyberdyne", "status": "open", "notes2": ""}, {"amount": "$648.45", "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Cyberdyne   ", "status": " new ", "notes2": ""}, {"amount": "0", "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Umbrella", "status": "Lost", "notes2": ""}, {"amount": "5.092,88", "city": "NER", "currency": "ethiopian birr   ", "company": " Initech ", "status": "new", "notes2": ""}, {"amount": "TBD", "city": "  San Pedrro de Jujuy", "currency": "Eahiopian  Birr", "company": "Globex", "status": "LSOT", "notes2": ""}, {"amount": "$310.66", "city": "SAN PEDR ODE JUJUY", "currency": "Ethiopian Birr", "company": " Stark Industries", "status": "  open", "notes2": ""}, {"amount": "$301.97", "city": "Toronto", "currency": "hkd", "company": " Stark Industries", "status": "LOST", "notes2": ""}, {"amount": "$305.31", "city": "Povorino", "currency": "Hong Kong Dollar", "company": " Globex", "status": "Closed Lost", "notes2": ""}, {"amount": "3.566,41", "city": "Povorino", "currency": "ETB", "company": " Umbrella ", "status": "closed lost", "notes2": ""}, {"amount": "7.484,67", "city": "Ner", "currency": "Hong Kong Dollar", "company": "Stark  Industries", "status": "Lost", "notes2": ""}, {"amount": "6.335,26", "city": "Povorino", "currency": "ETB   ", "company": "   Acme Inc", "status": "Won", "notes2": ""}, {"amount": "$2,043.84", "city": "Toronto", "currency": "Hong Kong Dollr  ", "company": "  Umbrella", "status": "CLOSED WON", "notes2": ""}, {"amount": "6.812,23", "city": "Ner", "currency": "ETB", "company": "Vehement   ", "status": " new ", "notes2": ""}, {"amount": "5.522,06", "city": " Ner ", "currency": "Hong Kong Dollar", "company": "Initech  ", "status": "won", "notes2": ""}, {"amount": "$656.95", "city": " Ner ", "currency": "etb", "company": "Wonka Co", "status": "Won", "notes2": ""}], "clean": [{"amount": NaN, "city": "San Pedro de Jujuy", "currency": "HKD", "company": "Vehement", "status": "Open"}, {"amount": 667.26, "city": "Ner", "currency": "HKD", "company": "Wonka Co", "status": "Won"}, {"amount": 2481.72, "city": "Toronto", "currency": "HKD", "company": "Hooli", "status": "Open"}, {"amount": NaN, "city": "Povorino", "currency": "ETB", "company": "Soylent Corp", "status": "Open"}, {"amount": 7961.84, "city": "Toronto", "currency": "ETB", "company": "Initech", "status": "Won"}, {"amount": 5869.93, "city": "Povorino", "currency": "ETB", "company": "Wonka Co", "status": "Lost"}, {"amount": 7171.03, "city": "Ner", "currency": "HKD", "company": "Cyberdyne", "status": "Open"}, {"amount": 2351.63, "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Stark Industries", "status": "Lost"}, {"amount": 8445.08, "city": "Toronto", "currency": "ETB", "company": "Umbrella", "status": "Open"}, {"amount": NaN, "city": "Ner", "currency": "ETB", "company": "Hooli", "status": "Open"}, {"amount": 2892.86, "city": "Ner", "currency": "HKD", "company": "Umbrella", "status": "Won"}, {"amount": 2320.68, "city": "San Pedro de Jujuy", "currency": "HKD", "company": "Umbrella", "status": "Open"}, {"amount": 2368.52, "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Acme Inc", "status": "Open"}, {"amount": 3077.02, "city": "Toronto", "currency": "ETB", "company": "Initech", "status": "Won"}, {"amount": 2957.09, "city": "Povorino", "currency": "HKD", "company": "Acme Inc", "status": "Lost"}, {"amount": 7540.8, "city": "Toronto", "currency": "HKD", "company": "Globex", "status": "Won"}, {"amount": 1744.86, "city": "Ner", "currency": "ETB", "company": "Cyberdyne", "status": "Open"}, {"amount": 648.45, "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Cyberdyne", "status": "Open"}, {"amount": 0.0, "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Umbrella", "status": "Lost"}, {"amount": 5092.88, "city": "Ner", "currency": "ETB", "company": "Initech", "status": "Open"}, {"amount": NaN, "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Globex", "status": "Lost"}, {"amount": 310.66, "city": "San Pedro de Jujuy", "currency": "ETB", "company": "Stark Industries", "status": "Open"}, {"amount": 301.97, "city": "Toronto", "currency": "HKD", "company": "Stark Industries", "status": "Lost"}, {"amount": 305.31, "city": "Povorino", "currency": "HKD", "company": "Globex", "status": "Lost"}, {"amount": 3566.41, "city": "Povorino", "currency": "ETB", "company": "Umbrella", "status": "Lost"}, {"amount": 7484.67, "city": "Ner", "currency": "HKD", "company": "Stark Industries", "status": "Lost"}, {"amount": 6335.26, "city": "Povorino", "currency": "ETB", "company": "Acme Inc", "status": "Won"}, {"amount": 2043.84, "city": "Toronto", "currency": "HKD", "company": "Umbrella", "status": "Won"}, {"amount": 6812.23, "city": "Ner", "currency": "ETB", "company": "Vehement", "status": "Open"}, {"amount": 5522.06, "city": "Ner", "currency": "HKD", "company": "Initech", "status": "Won"}, {"amount": 656.95, "city": "Ner", "currency": "ETB", "company": "Wonka Co", "status": "Won"}], "dirty_cols": ["amount", "city", "currency", "company", "status", "notes2"], "clean_cols": ["amount", "city", "currency", "company", "status"], "plan": {"dataset_summary": "31 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"toronto": "Toronto", "ner": "Ner", "SAN PEDRO DE JUJUY": "San Pedro de Jujuy", "Toronot": "Toronto", "NER": "Ner", "San Pedrro de Jujuy": "San Pedro de Jujuy", "SAN PEDR ODE JUJUY": "San Pedro de Jujuy"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Hong Koong Dollar": "HKD", "etb": "ETB", "ethiopian birr": "ETB", "hong kong dollar": "HKD", "ETHIOPIAN BIRR": "ETB", "hkd": "HKD", "hong  kong dollar": "HKD", "Ethiopian Birr": "ETB", "Hkd": "HKD", "Etb": "ETB", "Eahiopian Birr": "ETB", "Hong Kong Dollar": "HKD", "Hong Kong Dollr": "HKD"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"open": "Open", "CLOSED WON": "Won", "Closed-Won": "Won", "LOST": "Lost", "lost": "Lost", "new": "Open", "won": "Won", "closed-won": "Won", "LSOT": "Lost", "Closed Lost": "Lost", "closed lost": "Lost"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"unit": "Km", "amount": "TBD", "phone": "(566)383-6564", "country": "Antigua and Barbuda", "currency": "   Egyptian Pound", "signup_date": "45282", "job_title": "product manager ", "name": " Judy Brandt"}, {"unit": "litre", "amount": "8.656,75", "phone": "528-642-1365", "country": "Faroe Islands", "currency": "£", "signup_date": "44971", "job_title": "prod mgr", "name": "Omar Fischer"}, {"unit": "km", "amount": "3.801,01", "phone": "519.861.8940", "country": "atg ", "currency": "£", "signup_date": "2023-11-12", "job_title": "Vice President", "name": "Alice Park  "}, {"unit": "km", "amount": "8.935,78", "phone": "554-167-3766", "country": "fro", "currency": "armenian dram", "signup_date": "18 Dec 2023", "job_title": "VP", "name": "Karl Park"}, {"unit": "LITERS", "amount": "$1,886.37", "phone": "520.827.0678", "country": "atg", "currency": "£", "signup_date": "45085", "job_title": "Prod Mnager", "name": "Sara Wong"}, {"unit": "Km", "amount": "$2,443.67", "phone": "513-983-6267", "country": "Antigua and arbuda", "currency": "pound sterling", "signup_date": "2023-02-22", "job_title": "Prod Manager", "name": "  Frank Brandt"}, {"unit": "kilometer", "amount": "6.890,48", "phone": "532-922-7172", "country": " FO", "currency": "gbp  ", "signup_date": "2 Apr 2023", "job_title": "prod mgr", "name": "Bob  Reyes"}, {"unit": "KILOMETER", "amount": "N/A", "phone": "5479340755", "country": "antigua and barbuda", "currency": "EGP", "signup_date": "2023-04-18", "job_title": "vice  president", "name": "  Heidi Adams"}, {"unit": "kilometers", "amount": "6.670,18", "phone": "5414514749", "country": "ATG   ", "currency": " amd ", "signup_date": "2023-08-16", "job_title": "v.p.", "name": "Judy  Novak"}, {"unit": "kilometer", "amount": "164,61", "phone": "599-785-3370", "country": " fro ", "currency": "£", "signup_date": "13 Nov 2023", "job_title": "Prod Manager", "name": "Eve Brandt "}, {"unit": "kiolmeter", "amount": "7.228,45", "phone": "5296335577", "country": "AG", "currency": "armenian dram", "signup_date": "2023-01-04", "job_title": "prod maanger", "name": "Lena Reyes"}, {"unit": "liters", "amount": "$8,439.43", "phone": "508-434-0408", "country": "ag", "currency": "egp", "signup_date": "7/16/2023", "job_title": "pm", "name": "Bob Petrov"}, {"unit": " Km", "amount": "--", "phone": "541.793.3780", "country": "AG", "currency": "AMD", "signup_date": "11 Dec 2023", "job_title": "Prod Mgr ", "name": "Bob Brandt "}, {"unit": " kilometer", "amount": "4.775,66", "phone": "5990837345", "country": "fro", "currency": "Amd", "signup_date": "1 Aug 2023", "job_title": "PM", "name": "David Lee"}, {"unit": " kilpmeter", "amount": "4.933,70", "phone": "596.506.5609", "country": "Antigua and Barbuda", "currency": "£", "signup_date": "12/5/2023", "job_title": " vp ", "name": "Priya  Novak"}, {"unit": "liter", "amount": "6.312,52", "phone": "566.657.9688", "country": "FRO ", "currency": "EGP", "signup_date": "8/12/2023", "job_title": "Product Manager ", "name": "Priya Lee"}, {"unit": "kilometers", "amount": "$5,650.63", "phone": "(535)801-3675", "country": "  ATG", "currency": "gbp", "signup_date": "7/22/2023", "job_title": "PROD MANAGER", "name": "   Heidi Johnson"}, {"unit": "km", "amount": "$5,283.19", "phone": "(558)497-4406", "country": "FO", "currency": "AMD", "signup_date": "3/20/2023", "job_title": "Vice President", "name": "Alice Adams"}, {"unit": "Km  ", "amount": "5.073,84", "phone": "540-836-4851", "country": "AG  ", "currency": "Armenian Dram", "signup_date": "12/14/2023", "job_title": "PM   ", "name": "Carol Lee"}, {"unit": "L", "amount": "None", "phone": "554-031-4199", "country": "AG", "currency": "EGYPTIAN POUND", "signup_date": "3/24/2023", "job_title": "vice presiddent", "name": "Eve  Lee"}, {"unit": " kilometers ", "amount": "$4,831.34", "phone": "508-614-2300", "country": "ATG", "currency": "AMD", "signup_date": "5/10/2023", "job_title": "prod mgr", "name": "Karl  Johnson"}, {"unit": "   liters", "amount": "$51.47", "phone": "5978279224", "country": "   ATG", "currency": " pound sterling", "signup_date": "45127", "job_title": "prod mgr", "name": "Ivan Adams"}, {"unit": "litre", "amount": "na", "phone": "(552)618-4435", "country": "ag", "currency": "  armenian dram", "signup_date": "2/11/2023", "job_title": "VP", "name": "  Frank Johnson"}, {"unit": "litre", "amount": "5.645,42", "phone": "556.959.0626", "country": "Faroe Islands", "currency": "egp", "signup_date": "7/7/2023", "job_title": "  product manager", "name": "Judy Wong   "}, {"unit": "kilometers", "amount": "$8,672.01", "phone": "5844703395", "country": "Atg", "currency": "egp", "signup_date": "2023-12-24", "job_title": "   product manager", "name": "Karl Lee  "}, {"unit": "litre", "amount": "$6,832.17", "phone": "(580)275-0296", "country": "Faroe Islands", "currency": "armenian dram", "signup_date": "2/25/2023", "job_title": "pm", "name": "Priya Johnson"}, {"unit": "km", "amount": "3.676,51", "phone": "517-697-5159", "country": "fo", "currency": "pound sterling", "signup_date": "45163", "job_title": "vp", "name": "Omar Novak"}, {"unit": "liter", "amount": "$5,819.21", "phone": "588-168-5362", "country": "antigua  and barbuda", "currency": "ARMENIAN DRAM", "signup_date": "14 Jul 2023", "job_title": "vice pres", "name": "Eve Petrov"}, {"unit": "liters", "amount": "4.872,19", "phone": "501.965.9755", "country": "Atg", "currency": "Armenian Dram", "signup_date": "2023-02-21", "job_title": "  Prod Manager", "name": "Heidi Wong"}, {"unit": "kilometer", "amount": "#N/A", "phone": "(535)582-7491", "country": "   Antigua ind Barbuda", "currency": "egp", "signup_date": "2023-11-09", "job_title": "VP.", "name": "Priya Lee   "}, {"unit": "KM", "amount": "$1,403.97", "phone": "5749799674", "country": "Fro", "currency": "£", "signup_date": "45079", "job_title": "prod manager", "name": "David Park"}, {"unit": "Km", "amount": "$8,648.63", "phone": "(534)513-6348", "country": "FO", "currency": "Egyptian Pound", "signup_date": "2023-06-07", "job_title": "Product Manager", "name": "Karl Khan"}, {"unit": "km", "amount": "1.741,33", "phone": "(540)553-2043", "country": "  Fraoe Islands", "currency": "gbp", "signup_date": "8/1/2023", "job_title": "VICEPRESIDENT", "name": "Lena Novak"}, {"unit": "L", "amount": "2.603,44", "phone": "5283691182", "country": "antigua and barbuda", "currency": "EGYPTIAN POUND", "signup_date": "5 Nov 2023", "job_title": "VP", "name": "Ivan Ali"}, {"unit": "LITERS", "amount": "3.024,93", "phone": "577-453-9934", "country": "Antigua And Barbuda", "currency": "Egyptian Pound", "signup_date": "45114", "job_title": " Vp", "name": " Karl Park"}, {"unit": "L", "amount": "8.098,26", "phone": "596.752.6716", "country": "AG", "currency": "egyptianpound", "signup_date": "2/26/2023", "job_title": "VP", "name": "Grace Moore"}, {"unit": " km ", "amount": "1.214,09", "phone": "597-335-2680", "country": " FO ", "currency": "£", "signup_date": "44969", "job_title": "   VP", "name": "Carol  Novak"}, {"unit": "km", "amount": "4.079,95", "phone": "5417065832", "country": "ag", "currency": "egyptian pound", "signup_date": "23 Aug 2023", "job_title": "Product Manager", "name": "Alice Petrov"}, {"unit": "l", "amount": "$2,470.10", "phone": "5925100871", "country": "faroe islands", "currency": "£", "signup_date": "5/3/2023", "job_title": "pm", "name": "Frank Diaz"}, {"unit": "km", "amount": "6.247,32", "phone": "553.297.5503", "country": "Faroe Islands", "currency": "EGP", "signup_date": "45060", "job_title": "vp", "name": "  Carol Adams"}, {"unit": "l", "amount": "$2,605.66", "phone": "570.036.4161", "country": " FRO", "currency": "armenian dram", "signup_date": "45254", "job_title": "prod manager", "name": "Judy Lee"}, {"unit": "kilometer", "amount": "164,61", "phone": "599-785-3370", "country": " fro ", "currency": "£", "signup_date": "13 Nov 2023", "job_title": "Prod Manager", "name": "Eve Brandt "}, {"unit": "kilometer", "amount": "6.890,48", "phone": "532-922-7172", "country": " FO", "currency": "gbp  ", "signup_date": "2 Apr 2023", "job_title": "prod mgr", "name": "Bob  Reyes"}, {"unit": "", "amount": "", "phone": "", "country": "", "currency": "", "signup_date": "", "job_title": "", "name": ""}], "clean": [{"unit": "km", "amount": NaN, "phone": "(566) 383-6564", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-12-22", "job_title": "Product Manager", "name": "Judy Brandt"}, {"unit": "L", "amount": 8656.75, "phone": "(528) 642-1365", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-02-14", "job_title": "Product Manager", "name": "Omar Fischer"}, {"unit": "km", "amount": 3801.01, "phone": "(519) 861-8940", "country": "Antigua and Barbuda", "currency": "GBP", "signup_date": "2023-11-12", "job_title": "Vice President", "name": "Alice Park"}, {"unit": "km", "amount": 8935.78, "phone": "(554) 167-3766", "country": "Faroe Islands", "currency": "AMD", "signup_date": "2023-12-18", "job_title": "Vice President", "name": "Karl Park"}, {"unit": "L", "amount": 1886.37, "phone": "(520) 827-0678", "country": "Antigua and Barbuda", "currency": "GBP", "signup_date": "2023-06-08", "job_title": "Product Manager", "name": "Sara Wong"}, {"unit": "km", "amount": 2443.67, "phone": "(513) 983-6267", "country": "Antigua and Barbuda", "currency": "GBP", "signup_date": "2023-02-22", "job_title": "Product Manager", "name": "Frank Brandt"}, {"unit": "km", "amount": 6890.48, "phone": "(532) 922-7172", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-04-02", "job_title": "Product Manager", "name": "Bob Reyes"}, {"unit": "km", "amount": NaN, "phone": "(547) 934-0755", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-04-18", "job_title": "Vice President", "name": "Heidi Adams"}, {"unit": "km", "amount": 6670.18, "phone": "(541) 451-4749", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-08-16", "job_title": "Vice President", "name": "Judy Novak"}, {"unit": "km", "amount": 164.61, "phone": "(599) 785-3370", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-11-13", "job_title": "Product Manager", "name": "Eve Brandt"}, {"unit": "km", "amount": 7228.45, "phone": "(529) 633-5577", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-01-04", "job_title": "Product Manager", "name": "Lena Reyes"}, {"unit": "L", "amount": 8439.43, "phone": "(508) 434-0408", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-07-16", "job_title": "Product Manager", "name": "Bob Petrov"}, {"unit": "km", "amount": NaN, "phone": "(541) 793-3780", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-12-11", "job_title": "Product Manager", "name": "Bob Brandt"}, {"unit": "km", "amount": 4775.66, "phone": "(599) 083-7345", "country": "Faroe Islands", "currency": "AMD", "signup_date": "2023-08-01", "job_title": "Product Manager", "name": "David Lee"}, {"unit": "km", "amount": 4933.7, "phone": "(596) 506-5609", "country": "Antigua and Barbuda", "currency": "GBP", "signup_date": "2023-12-05", "job_title": "Vice President", "name": "Priya Novak"}, {"unit": "L", "amount": 6312.52, "phone": "(566) 657-9688", "country": "Faroe Islands", "currency": "EGP", "signup_date": "2023-08-12", "job_title": "Product Manager", "name": "Priya Lee"}, {"unit": "km", "amount": 5650.63, "phone": "(535) 801-3675", "country": "Antigua and Barbuda", "currency": "GBP", "signup_date": "2023-07-22", "job_title": "Product Manager", "name": "Heidi Johnson"}, {"unit": "km", "amount": 5283.19, "phone": "(558) 497-4406", "country": "Faroe Islands", "currency": "AMD", "signup_date": "2023-03-20", "job_title": "Vice President", "name": "Alice Adams"}, {"unit": "km", "amount": 5073.84, "phone": "(540) 836-4851", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-12-14", "job_title": "Product Manager", "name": "Carol Lee"}, {"unit": "L", "amount": NaN, "phone": "(554) 031-4199", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-03-24", "job_title": "Vice President", "name": "Eve Lee"}, {"unit": "km", "amount": 4831.34, "phone": "(508) 614-2300", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-05-10", "job_title": "Product Manager", "name": "Karl Johnson"}, {"unit": "L", "amount": 51.47, "phone": "(597) 827-9224", "country": "Antigua and Barbuda", "currency": "GBP", "signup_date": "2023-07-20", "job_title": "Product Manager", "name": "Ivan Adams"}, {"unit": "L", "amount": NaN, "phone": "(552) 618-4435", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-02-11", "job_title": "Vice President", "name": "Frank Johnson"}, {"unit": "L", "amount": 5645.42, "phone": "(556) 959-0626", "country": "Faroe Islands", "currency": "EGP", "signup_date": "2023-07-07", "job_title": "Product Manager", "name": "Judy Wong"}, {"unit": "km", "amount": 8672.01, "phone": "(584) 470-3395", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-12-24", "job_title": "Product Manager", "name": "Karl Lee"}, {"unit": "L", "amount": 6832.17, "phone": "(580) 275-0296", "country": "Faroe Islands", "currency": "AMD", "signup_date": "2023-02-25", "job_title": "Product Manager", "name": "Priya Johnson"}, {"unit": "km", "amount": 3676.51, "phone": "(517) 697-5159", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-08-25", "job_title": "Vice President", "name": "Omar Novak"}, {"unit": "L", "amount": 5819.21, "phone": "(588) 168-5362", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-07-14", "job_title": "Vice President", "name": "Eve Petrov"}, {"unit": "L", "amount": 4872.19, "phone": "(501) 965-9755", "country": "Antigua and Barbuda", "currency": "AMD", "signup_date": "2023-02-21", "job_title": "Product Manager", "name": "Heidi Wong"}, {"unit": "km", "amount": NaN, "phone": "(535) 582-7491", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-11-09", "job_title": "Vice President", "name": "Priya Lee"}, {"unit": "km", "amount": 1403.97, "phone": "(574) 979-9674", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-06-02", "job_title": "Product Manager", "name": "David Park"}, {"unit": "km", "amount": 8648.63, "phone": "(534) 513-6348", "country": "Faroe Islands", "currency": "EGP", "signup_date": "2023-06-07", "job_title": "Product Manager", "name": "Karl Khan"}, {"unit": "km", "amount": 1741.33, "phone": "(540) 553-2043", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-08-01", "job_title": "Vice President", "name": "Lena Novak"}, {"unit": "L", "amount": 2603.44, "phone": "(528) 369-1182", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-11-05", "job_title": "Vice President", "name": "Ivan Ali"}, {"unit": "L", "amount": 3024.93, "phone": "(577) 453-9934", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-07-07", "job_title": "Vice President", "name": "Karl Park"}, {"unit": "L", "amount": 8098.26, "phone": "(596) 752-6716", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-02-26", "job_title": "Vice President", "name": "Grace Moore"}, {"unit": "km", "amount": 1214.09, "phone": "(597) 335-2680", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-02-12", "job_title": "Vice President", "name": "Carol Novak"}, {"unit": "km", "amount": 4079.95, "phone": "(541) 706-5832", "country": "Antigua and Barbuda", "currency": "EGP", "signup_date": "2023-08-23", "job_title": "Product Manager", "name": "Alice Petrov"}, {"unit": "L", "amount": 2470.1, "phone": "(592) 510-0871", "country": "Faroe Islands", "currency": "GBP", "signup_date": "2023-05-03", "job_title": "Product Manager", "name": "Frank Diaz"}, {"unit": "km", "amount": 6247.32, "phone": "(553) 297-5503", "country": "Faroe Islands", "currency": "EGP", "signup_date": "2023-05-14", "job_title": "Vice President", "name": "Carol Adams"}, {"unit": "L", "amount": 2605.66, "phone": "(570) 036-4161", "country": "Faroe Islands", "currency": "AMD", "signup_date": "2023-11-24", "job_title": "Product Manager", "name": "Judy Lee"}], "dirty_cols": ["unit", "amount", "phone", "country", "currency", "signup_date", "job_title", "name"], "clean_cols": ["unit", "amount", "phone", "country", "currency", "signup_date", "job_title", "name"], "plan": {"dataset_summary": "44 rows × 8 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Km": "km", "litre": "L", "LITERS": "L", "kilometer": "km", "KILOMETER": "km", "kilometers": "km", "kiolmeter": "km", "liters": "L", "kilpmeter": "km", "liter": "L", "KM": "km", "l": "L"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"atg": "Antigua and Barbuda", "fro": "Faroe Islands", "Antigua and arbuda": "Antigua and Barbuda", "FO": "Faroe Islands", "antigua and barbuda": "Antigua and Barbuda", "ATG": "Antigua and Barbuda", "AG": "Antigua and Barbuda", "ag": "Antigua and Barbuda", "FRO": "Faroe Islands", "Atg": "Antigua and Barbuda", "fo": "Faroe Islands", "Antigua ind Barbuda": "Antigua and Barbuda", "Fro": "Faroe Islands", "Fraoe Islands": "Faroe Islands", "Antigua And Barbuda": "Antigua and Barbuda", "faroe islands": "Faroe Islands"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Egyptian Pound": "EGP", "£": "GBP", "armenian dram": "AMD", "pound sterling": "GBP", "gbp": "GBP", "amd": "AMD", "egp": "EGP", "Amd": "AMD", "Armenian Dram": "AMD", "EGYPTIAN POUND": "EGP", "ARMENIAN DRAM": "AMD", "egyptianpound": "EGP", "egyptian pound": "EGP"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"product manager": "Product Manager", "prod mgr": "Product Manager", "VP": "Vice President", "Prod Mnager": "Product Manager", "Prod Manager": "Product Manager", "vice president": "Vice President", "v.p.": "Vice President", "prod maanger": "Product Manager", "pm": "Product Manager", "Prod Mgr": "Product Manager", "PM": "Product Manager", "vp": "Vice President", "PROD MANAGER": "Product Manager", "vice presiddent": "Vice President", "vice pres": "Vice President", "VP.": "Vice President", "prod manager": "Product Manager", "VICEPRESIDENT": "Vice President", "Vp": "Vice President"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"city": "sundern", "status": "overdue", "is_active": "T", "unit": "l", "signup_date": "2023-10-11", "email": "dmgfbm@corp.io"}, {"city": " balod", "status": "Refunded", "is_active": "F", "unit": "centimeters", "signup_date": "8/14/2023", "email": "oombme@mail.com"}, {"city": "balod", "status": "OVERDUE", "is_active": "Yes", "unit": "mls", "signup_date": "2023-11-01", "email": " HIJAI@MAIL.COM "}, {"city": "Langfang ", "status": "Pending  ", "is_active": "F", "unit": "Ml", "signup_date": "3 Aug 2023", "email": " pmfgn@mail.com "}, {"city": "Sunden", "status": "Pending", "is_active": "0", "unit": "Cm   ", "signup_date": "2023-03-05", "email": "achmjla@example.com"}, {"city": "Sundern   ", "status": "refund", "is_active": "N", "unit": "ml", "signup_date": "3 Nov 2023", "email": " JJEJJ@CORP.IO "}, {"city": "balod", "status": "refunded", "is_active": "TRUE", "unit": "l ", "signup_date": "2023-01-23", "email": "   KJJFBL@EXAMPLE.COM"}, {"city": " Sundern ", "status": "OVERDUE", "is_active": "F", "unit": "litir", "signup_date": "44985", "email": "ocnbb@example.com"}, {"city": "Langfang  ", "status": " Pending ", "is_active": "No", "unit": "Mls", "signup_date": "2023-07-27", "email": "NLFGMNB@TEST.ORG"}, {"city": "Balod", "status": "late  ", "is_active": "N", "unit": " CM", "signup_date": "2023-09-14", "email": "gmbdo@corp.io"}, {"city": " LANGFANG", "status": "Pending", "is_active": "F", "unit": "milliliter", "signup_date": "17 Sep 2023", "email": "PLFCCI@CORP.IO"}, {"city": "Sundern", "status": "REFUNDED", "is_active": "N", "unit": "Centimeters", "signup_date": "2023-03-21", "email": "bjbph@example.com"}, {"city": "langfang ", "status": "refunded", "is_active": "TRUE", "unit": "mls", "signup_date": "2023-08-02", "email": " iilf@example.com "}, {"city": "SUNDERN", "status": "Paid", "is_active": "1", "unit": "ml", "signup_date": "22 Sep 2023", "email": "akobl@corp.io"}, {"city": "Balod", "status": "overdue", "is_active": "0", "unit": "mls", "signup_date": "2023-10-05", "email": " hehb@example.com "}, {"city": "LANGFANG", "status": "due", "is_active": "0", "unit": "ml", "signup_date": "3 Jul 2023", "email": "LPICN@EXAMPLE.COM"}, {"city": "Sundern", "status": "late", "is_active": "FALSE", "unit": "l", "signup_date": "11/27/2023", "email": "CKBMO@EXAMPLE.COM"}, {"city": "sundern", "status": "PAID", "is_active": "FALSE", "unit": "ml", "signup_date": "45090", "email": "cbhp@test.org"}, {"city": "SUNDERN", "status": "Paid", "is_active": "TRUE", "unit": " milliliter ", "signup_date": "2023-06-27", "email": "alla@mail.com"}, {"city": "Balod", "status": "pending ", "is_active": "TRUE", "unit": "   l", "signup_date": "2023-12-25", "email": "pobako@example.com"}, {"city": "Langfang  ", "status": "Pending", "is_active": "false", "unit": " milxiliter ", "signup_date": "16 Oct 2023", "email": " bghjof@mail.com"}], "clean": [{"city": "Sundern", "status": "Overdue", "is_active": true, "unit": "L", "signup_date": "2023-10-11", "email": "dmgfbm@corp.io"}, {"city": "Balod", "status": "Refunded", "is_active": false, "unit": "cm", "signup_date": "2023-08-14", "email": "oombme@mail.com"}, {"city": "Balod", "status": "Overdue", "is_active": true, "unit": "mL", "signup_date": "2023-11-01", "email": "hijai@mail.com"}, {"city": "Langfang", "status": "Pending", "is_active": false, "unit": "mL", "signup_date": "2023-08-03", "email": "pmfgn@mail.com"}, {"city": "Sundern", "status": "Pending", "is_active": false, "unit": "cm", "signup_date": "2023-03-05", "email": "achmjla@example.com"}, {"city": "Sundern", "status": "Refunded", "is_active": false, "unit": "mL", "signup_date": "2023-11-03", "email": "jjejj@corp.io"}, {"city": "Balod", "status": "Refunded", "is_active": true, "unit": "L", "signup_date": "2023-01-23", "email": "kjjfbl@example.com"}, {"city": "Sundern", "status": "Overdue", "is_active": false, "unit": "L", "signup_date": "2023-02-28", "email": "ocnbb@example.com"}, {"city": "Langfang", "status": "Pending", "is_active": false, "unit": "mL", "signup_date": "2023-07-27", "email": "nlfgmnb@test.org"}, {"city": "Balod", "status": "Overdue", "is_active": false, "unit": "cm", "signup_date": "2023-09-14", "email": "gmbdo@corp.io"}, {"city": "Langfang", "status": "Pending", "is_active": false, "unit": "mL", "signup_date": "2023-09-17", "email": "plfcci@corp.io"}, {"city": "Sundern", "status": "Refunded", "is_active": false, "unit": "cm", "signup_date": "2023-03-21", "email": "bjbph@example.com"}, {"city": "Langfang", "status": "Refunded", "is_active": true, "unit": "mL", "signup_date": "2023-08-02", "email": "iilf@example.com"}, {"city": "Sundern", "status": "Paid", "is_active": true, "unit": "mL", "signup_date": "2023-09-22", "email": "akobl@corp.io"}, {"city": "Balod", "status": "Overdue", "is_active": false, "unit": "mL", "signup_date": "2023-10-05", "email": "hehb@example.com"}, {"city": "Langfang", "status": "Pending", "is_active": false, "unit": "mL", "signup_date": "2023-07-03", "email": "lpicn@example.com"}, {"city": "Sundern", "status": "Overdue", "is_active": false, "unit": "L", "signup_date": "2023-11-27", "email": "ckbmo@example.com"}, {"city": "Sundern", "status": "Paid", "is_active": false, "unit": "mL", "signup_date": "2023-06-13", "email": "cbhp@test.org"}, {"city": "Sundern", "status": "Paid", "is_active": true, "unit": "mL", "signup_date": "2023-06-27", "email": "alla@mail.com"}, {"city": "Balod", "status": "Pending", "is_active": true, "unit": "L", "signup_date": "2023-12-25", "email": "pobako@example.com"}, {"city": "Langfang", "status": "Pending", "is_active": false, "unit": "mL", "signup_date": "2023-10-16", "email": "bghjof@mail.com"}], "dirty_cols": ["city", "status", "is_active", "unit", "signup_date", "email"], "clean_cols": ["city", "status", "is_active", "unit", "signup_date", "email"], "plan": {"dataset_summary": "21 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sundern": "Sundern", "balod": "Balod", "Sunden": "Sundern", "LANGFANG": "Langfang", "langfang": "Langfang", "SUNDERN": "Sundern"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"overdue": "Overdue", "OVERDUE": "Overdue", "refund": "Refunded", "refunded": "Refunded", "late": "Overdue", "REFUNDED": "Refunded", "due": "Pending", "PAID": "Paid", "pending": "Pending"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"l": "L", "centimeters": "cm", "mls": "mL", "Ml": "mL", "Cm": "cm", "ml": "mL", "litir": "L", "Mls": "mL", "CM": "cm", "milliliter": "mL", "Centimeters": "cm", "milxiliter": "mL"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"amount": "8.275,87", "industry": "RealEstate", "city": "bunkyo   ", "name": "Judy Brandt", "currency": "bmd", "state": "MA", "extra": ""}, {"amount": "620,97", "industry": "Real Estate", "city": "Sherpur", "name": "Ivan Cruz", "currency": "bermudian  dollar", "state": "MA", "extra": ""}, {"amount": "6.241,62", "industry": "property", "city": "pokahci", "name": "Priya  Lee", "currency": "BMD", "state": "nebraska", "extra": ""}, {"amount": "$3,190.59", "industry": "Telco", "city": "SHEROUR", "name": "Frank Lee ", "currency": "Gourde   ", "state": " Michigan", "extra": ""}, {"amount": "None", "industry": " PROPERTY", "city": " sherpur ", "name": "Heidi Park", "currency": "GOURDE", "state": "MI", "extra": ""}, {"amount": "873,61", "industry": "TELECOM", "city": "Pokachi", "name": "  Mona Wong", "currency": "bmd", "state": "Nebraska", "extra": ""}, {"amount": "$7,612.36", "industry": "RealEstate", "city": "MOYALE", "name": "  Alice Reyes", "currency": "COLOMBIANfPESO", "state": "MA", "extra": ""}, {"amount": "?", "industry": "Telecommunications", "city": "MOYALE", "name": " David Brandt", "currency": "Gourde", "state": "MA ", "extra": ""}, {"amount": "$5,377.48", "industry": "proptech ", "city": "MOYALE", "name": "Sara Diaz", "currency": "  HTG", "state": "MI", "extra": ""}, {"amount": "$2,711.02", "industry": "Rel Estate  ", "city": "Moyale", "name": "Judy Reyes", "currency": "bmd", "state": "massachusetts", "extra": ""}, {"amount": "$3,742.77", "industry": "REAL ESTATE", "city": " Moyale", "name": " David Reyes", "currency": "gourde", "state": " MA ", "extra": ""}, {"amount": "2.363,07", "industry": " telecom", "city": "pokachi", "name": "  Bob Diaz", "currency": "   htg", "state": "MASSACHUSETTS", "extra": ""}, {"amount": "1.739,91", "industry": "PropTech", "city": "BUNKYO", "name": "Grace Reyes", "currency": "Gourde", "state": "NE  ", "extra": ""}, {"amount": "620,97", "industry": "Real Estate", "city": "Sherpur", "name": "Ivan Cruz", "currency": "bermudian  dollar", "state": "MA", "extra": ""}, {"amount": "", "industry": "", "city": "", "name": "", "currency": "", "state": "", "extra": ""}], "clean": [{"amount": 8275.87, "industry": "Real Estate", "city": "Bunkyo", "name": "Judy Brandt", "currency": "BMD", "state": "Massachusetts"}, {"amount": 620.97, "industry": "Real Estate", "city": "Sherpur", "name": "Ivan Cruz", "currency": "BMD", "state": "Massachusetts"}, {"amount": 6241.62, "industry": "Real Estate", "city": "Pokachi", "name": "Priya Lee", "currency": "BMD", "state": "Nebraska"}, {"amount": 3190.59, "industry": "Telecommunications", "city": "Sherpur", "name": "Frank Lee", "currency": "HTG", "state": "Michigan"}, {"amount": NaN, "industry": "Real Estate", "city": "Sherpur", "name": "Heidi Park", "currency": "HTG", "state": "Michigan"}, {"amount": 873.61, "industry": "Telecommunications", "city": "Pokachi", "name": "Mona Wong", "currency": "BMD", "state": "Nebraska"}, {"amount": 7612.36, "industry": "Real Estate", "city": "Moyale", "name": "Alice Reyes", "currency": "COP", "state": "Massachusetts"}, {"amount": NaN, "industry": "Telecommunications", "city": "Moyale", "name": "David Brandt", "currency": "HTG", "state": "Massachusetts"}, {"amount": 5377.48, "industry": "Real Estate", "city": "Moyale", "name": "Sara Diaz", "currency": "HTG", "state": "Michigan"}, {"amount": 2711.02, "industry": "Real Estate", "city": "Moyale", "name": "Judy Reyes", "currency": "BMD", "state": "Massachusetts"}, {"amount": 3742.77, "industry": "Real Estate", "city": "Moyale", "name": "David Reyes", "currency": "HTG", "state": "Massachusetts"}, {"amount": 2363.07, "industry": "Telecommunications", "city": "Pokachi", "name": "Bob Diaz", "currency": "HTG", "state": "Massachusetts"}, {"amount": 1739.91, "industry": "Real Estate", "city": "Bunkyo", "name": "Grace Reyes", "currency": "HTG", "state": "Nebraska"}], "dirty_cols": ["amount", "industry", "city", "name", "currency", "state", "extra"], "clean_cols": ["amount", "industry", "city", "name", "currency", "state"], "plan": {"dataset_summary": "15 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"RealEstate": "Real Estate", "property": "Real Estate", "Telco": "Telecommunications", "PROPERTY": "Real Estate", "TELECOM": "Telecommunications", "proptech": "Real Estate", "Rel Estate": "Real Estate", "REAL ESTATE": "Real Estate", "telecom": "Telecommunications", "PropTech": "Real Estate"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bunkyo": "Bunkyo", "pokahci": "Pokachi", "SHEROUR": "Sherpur", "sherpur": "Sherpur", "MOYALE": "Moyale", "pokachi": "Pokachi", "BUNKYO": "Bunkyo"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bmd": "BMD", "bermudian dollar": "BMD", "Gourde": "HTG", "GOURDE": "HTG", "COLOMBIANfPESO": "COP", "gourde": "HTG", "htg": "HTG"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MA": "Massachusetts", "nebraska": "Nebraska", "MI": "Michigan", "massachusetts": "Massachusetts", "MASSACHUSETTS": "Massachusetts", "NE": "Nebraska"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "Yes", "department": "Eng.", "amount": "7.475,59", "currency": "BWP ", "email": "gmlfpg@corp.io  ", "job_title": "Senior Engiineer"}, {"is_active": "0", "department": "Eng.", "amount": "-100", "currency": "bermudian  dollar", "email": "dedmp@example.com", "job_title": "Acct  Exec"}, {"is_active": "N", "department": "   Biz Dev", "amount": "$2,550.73", "currency": "bermudian dollar", "email": "kacef@example.com", "job_title": "C.T.O."}, {"is_active": "TRUE", "department": "  CS", "amount": "$8,306.40", "currency": "clp", "email": "HDAG@CORP.IO  ", "job_title": "Senior Eng"}, {"is_active": "true", "department": "Spuport", "amount": "$5,805.67", "currency": "DOP", "email": "NAHLA@TEST.ORG", "job_title": "CTO"}, {"is_active": "1", "department": "biz dev   ", "amount": "4.478,08", "currency": "bermudian dollar", "email": "chmekh@test.org", "job_title": "Rep"}, {"is_active": "Yes", "department": "Biz Dev", "amount": "5.337,32", "currency": "DOP", "email": " NJFNNC@EXAMPLE.COM", "job_title": "sales  repressentative"}, {"is_active": "T", "department": " R&D ", "amount": "6.241,76", "currency": "bwp", "email": "lige@mail.com", "job_title": "Rep"}, {"is_active": "FALSE", "department": "Dev", "amount": "1.892,63", "currency": " Chilean Peso", "email": "akgbn@test.org", "job_title": " SrjEngineer "}, {"is_active": "true", "department": "biz dev", "amount": "$3,905.98", "currency": "Dop", "email": "kblma@example.com ", "job_title": "C.T.O. "}, {"is_active": "No", "department": "Support", "amount": "$8,863.09", "currency": "   Bmd", "email": "BPIO@TEST.ORG", "job_title": "Sales  Representative"}, {"is_active": "No", "department": " eng. ", "amount": "$7,478.45", "currency": "CLP", "email": "AFMNHD@TEST.ORG", "job_title": "account executive"}, {"is_active": "Y", "department": "dev", "amount": "7.147,46", "currency": "BERMUDIAN DOLLAR", "email": "AIPDMML@EXAMPLE.COM", "job_title": "Sinior  Engineer"}, {"is_active": "No", "department": "Support", "amount": "$8,863.09", "currency": "   Bmd", "email": "BPIO@TEST.ORG", "job_title": "Sales  Representative"}, {"is_active": "true", "department": "biz dev", "amount": "$3,905.98", "currency": "Dop", "email": "kblma@example.com ", "job_title": "C.T.O. "}], "clean": [{"is_active": true, "department": "Engineering", "amount": 7475.59, "currency": "BWP", "email": "gmlfpg@corp.io", "job_title": "Senior Engineer"}, {"is_active": false, "department": "Engineering", "amount": -100.0, "currency": "BMD", "email": "dedmp@example.com", "job_title": "Account Executive"}, {"is_active": false, "department": "Sales", "amount": 2550.73, "currency": "BMD", "email": "kacef@example.com", "job_title": "Chief Technology Officer"}, {"is_active": true, "department": "Customer Support", "amount": 8306.4, "currency": "CLP", "email": "hdag@corp.io", "job_title": "Senior Engineer"}, {"is_active": true, "department": "Customer Support", "amount": 5805.67, "currency": "DOP", "email": "nahla@test.org", "job_title": "Chief Technology Officer"}, {"is_active": true, "department": "Sales", "amount": 4478.08, "currency": "BMD", "email": "chmekh@test.org", "job_title": "Sales Representative"}, {"is_active": true, "department": "Sales", "amount": 5337.32, "currency": "DOP", "email": "njfnnc@example.com", "job_title": "Sales Representative"}, {"is_active": true, "department": "Engineering", "amount": 6241.76, "currency": "BWP", "email": "lige@mail.com", "job_title": "Sales Representative"}, {"is_active": false, "department": "Engineering", "amount": 1892.63, "currency": "CLP", "email": "akgbn@test.org", "job_title": "Senior Engineer"}, {"is_active": true, "department": "Sales", "amount": 3905.98, "currency": "DOP", "email": "kblma@example.com", "job_title": "Chief Technology Officer"}, {"is_active": false, "department": "Customer Support", "amount": 8863.09, "currency": "BMD", "email": "bpio@test.org", "job_title": "Sales Representative"}, {"is_active": false, "department": "Engineering", "amount": 7478.45, "currency": "CLP", "email": "afmnhd@test.org", "job_title": "Account Executive"}, {"is_active": true, "department": "Engineering", "amount": 7147.46, "currency": "BMD", "email": "aipdmml@example.com", "job_title": "Senior Engineer"}], "dirty_cols": ["is_active", "department", "amount", "currency", "email", "job_title"], "clean_cols": ["is_active", "department", "amount", "currency", "email", "job_title"], "plan": {"dataset_summary": "15 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Eng.": "Engineering", "Biz Dev": "Sales", "CS": "Customer Support", "Spuport": "Customer Support", "biz dev": "Sales", "R&D": "Engineering", "Dev": "Engineering", "Support": "Customer Support", "eng.": "Engineering", "dev": "Engineering"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bermudian  dollar": "BMD", "bermudian dollar": "BMD", "clp": "CLP", "bwp": "BWP", "Chilean Peso": "CLP", "Dop": "DOP", "Bmd": "BMD", "BERMUDIAN DOLLAR": "BMD"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Senior Engiineer": "Senior Engineer", "Acct Exec": "Account Executive", "C.T.O.": "Chief Technology Officer", "Senior Eng": "Senior Engineer", "CTO": "Chief Technology Officer", "Rep": "Sales Representative", "sales repressentative": "Sales Representative", "SrjEngineer": "Senior Engineer", "account executive": "Account Executive", "Sinior Engineer": "Senior Engineer"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"signup_date": "2023-07-17", "amount": "8.565,44", "email": "gcjm@test.org", "company": "Stark Industries  ", "phone": "502-257-1554", "rate": "43.6%", "notes2": ""}, {"signup_date": "45117", "amount": "6.109,23", "email": "  gcahi@example.com", "company": "Hooli", "phone": "516.479.5912", "rate": "15.5%", "notes2": ""}, {"signup_date": "45248", "amount": "$5,947.29", "email": "cpnjmdm@corp.io", "company": " Hooli", "phone": "5660472204", "rate": "94.3%", "notes2": ""}, {"signup_date": "5/7/2023", "amount": "298,19", "email": "cpfh@mail.com", "company": "Initech", "phone": "597.628.1007", "rate": "53.2%", "notes2": ""}, {"signup_date": "45119", "amount": "$7,800.26", "email": "FOCJLDP@MAIL.COM", "company": "Initech ", "phone": "563.666.5275", "rate": "39.0%", "notes2": ""}, {"signup_date": "2023-05-02", "amount": "6.302,21", "email": "   apijo@corp.io", "company": "Hooli  ", "phone": "558-406-1103", "rate": "44.6%", "notes2": ""}, {"signup_date": "11/25/2023", "amount": "None", "email": "dbbop@test.org", "company": "Acme Inc", "phone": "512-444-4717", "rate": "17.3%", "notes2": ""}, {"signup_date": "8/1/2023", "amount": "$5,548.23", "email": "DJGL@CORP.IO", "company": "Cyberdyne   ", "phone": "595-742-5340", "rate": "72.6%", "notes2": ""}, {"signup_date": "2023-04-15", "amount": "$4,814.47", "email": "GICO@TEST.ORG", "company": "Initech", "phone": "(558)432-1028", "rate": "74.7%", "notes2": ""}, {"signup_date": "2023-01-20", "amount": "8.772,92", "email": "ajcfa@test.org", "company": "  Umbrella", "phone": "(515)006-1993", "rate": "20.7%", "notes2": ""}, {"signup_date": "45273", "amount": "4.136,21", "email": " JKCELDK@EXAMPLE.COM ", "company": " Stark Industries", "phone": "555-926-1500", "rate": "64.1%", "notes2": ""}, {"signup_date": "1 Mar 2023", "amount": "8.747,84", "email": " licihad@corp.io", "company": "  Soylent Corp", "phone": "(569)899-7529", "rate": "24.4%", "notes2": ""}, {"signup_date": "17 Mar 2023", "amount": "$6,657.76", "email": "FDCNOG@EXAMPLE.COM", "company": "Initech", "phone": "(581)234-6019", "rate": "83.9%", "notes2": ""}, {"signup_date": "2023-08-07", "amount": "$7,266.54", "email": "   cpnmml@test.org", "company": "Vehement", "phone": "597.345.6163", "rate": "66.1%", "notes2": ""}, {"signup_date": "2023-12-05", "amount": "$7,805.30", "email": "IBJLHHP@CORP.IO", "company": "  Wonka Co", "phone": "(558)195-3040", "rate": "86.1%", "notes2": ""}, {"signup_date": "5/9/2023", "amount": "$4,703.68", "email": " MDFJ@TEST.ORG ", "company": "Globex", "phone": "5884174157", "rate": "55.9%", "notes2": ""}, {"signup_date": "2 Nov 2023", "amount": "$8,175.68", "email": "EIHIGGN@CORP.IO", "company": "Acme Inc", "phone": "(545)847-4045", "rate": "29.2%", "notes2": ""}, {"signup_date": "2023-08-17", "amount": "$4,599.82", "email": "   PGMDOL@TEST.ORG", "company": " Cyberdyne ", "phone": "536.723.9337", "rate": "35.0%", "notes2": ""}, {"signup_date": "27 Nov 2023", "amount": "$7,677.85", "email": "kmiljio@example.com", "company": "Initech   ", "phone": "576.344.5497", "rate": "91.2%", "notes2": ""}, {"signup_date": "6 Nov 2023", "amount": "$4,812.97", "email": "mkeoija@corp.io", "company": "Umbrella", "phone": "5140500491", "rate": "17.1%", "notes2": ""}, {"signup_date": "1/14/2023", "amount": "$6,382.57", "email": "IAPNJ@MAIL.COM", "company": "Initech", "phone": "5902010955", "rate": "94.2%", "notes2": ""}, {"signup_date": "2023-11-14", "amount": "$896.79", "email": "fblkce@corp.io", "company": "   Soylent Corp", "phone": "514-169-9665", "rate": "67.6%", "notes2": ""}, {"signup_date": "44990", "amount": "$6,045.79", "email": "IKPHGA@EXAMPLE.COM", "company": "Wonka Co", "phone": "500-286-3327", "rate": "74.8%", "notes2": ""}, {"signup_date": "2023-11-19", "amount": "$5,698.35", "email": "ogfeml@mail.com   ", "company": "Soylent  Corp", "phone": "5041584711", "rate": "29.0%", "notes2": ""}, {"signup_date": "11 Dec 2023", "amount": "4.921,43", "email": "GCIFPAN@TEST.ORG", "company": "Wonka Co", "phone": "569-657-4907", "rate": "41.1%", "notes2": ""}, {"signup_date": "20 Mar 2023", "amount": "4.574,84", "email": " ncefbl@mail.com ", "company": "Globex", "phone": "5115740398", "rate": "49.8%", "notes2": ""}, {"signup_date": "12/11/2023", "amount": "2.176,94", "email": "ccdbdkb@example.com", "company": "Wonka Co", "phone": "502.086.7097", "rate": "31.7%", "notes2": ""}, {"signup_date": "11/3/2023", "amount": "$1,733.67", "email": " IBBA@CORP.IO ", "company": "   Globex", "phone": "595-075-8147", "rate": "96.9%", "notes2": ""}, {"signup_date": "4/20/2023", "amount": "5.410,04", "email": "IDNL@TEST.ORG", "company": "Cyberdyne", "phone": "536-940-2045", "rate": "24.6%", "notes2": ""}, {"signup_date": "11/7/2023", "amount": "$5,232.39", "email": "dfmo@mail.com  ", "company": " Soylent Corp", "phone": "(545)471-4627", "rate": "55.8%", "notes2": ""}, {"signup_date": "9/7/2023", "amount": "2.018,36", "email": "hgkhepj@corp.io", "company": "Acme Inc", "phone": "506-764-9331", "rate": "73.3%", "notes2": ""}, {"signup_date": "45272", "amount": "274,67", "email": "LAPDI@CORP.IO", "company": " Globex ", "phone": "502.730.7986", "rate": "19.3%", "notes2": ""}, {"signup_date": "18 Dec 2023", "amount": "2.699,61", "email": "DOBCLI@CORP.IO", "company": "Initech", "phone": "5612898761", "rate": "88.4%", "notes2": ""}, {"signup_date": "2023-01-10", "amount": "?", "email": "  kegi@test.org", "company": "  Acme Inc", "phone": "504-369-8586", "rate": "4.6%", "notes2": ""}, {"signup_date": "2023-03-03", "amount": "6.133,16", "email": "OHBG@EXAMPLE.COM", "company": "Soylent Corp", "phone": "587-288-5300", "rate": "35.4%", "notes2": ""}, {"signup_date": "45279", "amount": "$2,944.01", "email": "khplpp@example.com", "company": "Globex", "phone": "(595)493-0642", "rate": "69.8%", "notes2": ""}, {"signup_date": "44976", "amount": "4.853,79", "email": "NMNMCDL@MAIL.COM", "company": "Hooli", "phone": "5874995637", "rate": "41.9%", "notes2": ""}, {"signup_date": "45053", "amount": "7.751,91", "email": " MJEFPFJ@EXAMPLE.COM ", "company": "   Vehement", "phone": "516-691-5406", "rate": "93.7%", "notes2": ""}, {"signup_date": "5 Feb 2023", "amount": "TBD", "email": "ognog@corp.io ", "company": "Stark Industries", "phone": "(544)761-5093", "rate": "99.3%", "notes2": ""}, {"signup_date": "12 May 2023", "amount": "4.456,68", "email": "ccmo@test.org", "company": "Wonka Co", "phone": "5491139473", "rate": "66.9%", "notes2": ""}, {"signup_date": "44954", "amount": "$1,166.11", "email": "  knha@corp.io", "company": "   Initech", "phone": "569.217.9343", "rate": "54.6%", "notes2": ""}], "clean": [{"signup_date": "2023-07-17", "amount": 8565.44, "email": "gcjm@test.org", "company": "Stark Industries", "phone": "(502) 257-1554", "rate": 0.436}, {"signup_date": "2023-07-10", "amount": 6109.23, "email": "gcahi@example.com", "company": "Hooli", "phone": "(516) 479-5912", "rate": 0.155}, {"signup_date": "2023-11-18", "amount": 5947.29, "email": "cpnjmdm@corp.io", "company": "Hooli", "phone": "(566) 047-2204", "rate": 0.943}, {"signup_date": "2023-05-07", "amount": 298.19, "email": "cpfh@mail.com", "company": "Initech", "phone": "(597) 628-1007", "rate": 0.532}, {"signup_date": "2023-07-12", "amount": 7800.26, "email": "focjldp@mail.com", "company": "Initech", "phone": "(563) 666-5275", "rate": 0.39}, {"signup_date": "2023-05-02", "amount": 6302.21, "email": "apijo@corp.io", "company": "Hooli", "phone": "(558) 406-1103", "rate": 0.446}, {"signup_date": "2023-11-25", "amount": NaN, "email": "dbbop@test.org", "company": "Acme Inc", "phone": "(512) 444-4717", "rate": 0.17300000000000001}, {"signup_date": "2023-08-01", "amount": 5548.23, "email": "djgl@corp.io", "company": "Cyberdyne", "phone": "(595) 742-5340", "rate": 0.726}, {"signup_date": "2023-04-15", "amount": 4814.47, "email": "gico@test.org", "company": "Initech", "phone": "(558) 432-1028", "rate": 0.747}, {"signup_date": "2023-01-20", "amount": 8772.92, "email": "ajcfa@test.org", "company": "Umbrella", "phone": "(515) 006-1993", "rate": 0.207}, {"signup_date": "2023-12-13", "amount": 4136.21, "email": "jkceldk@example.com", "company": "Stark Industries", "phone": "(555) 926-1500", "rate": 0.6409999999999999}, {"signup_date": "2023-03-01", "amount": 8747.84, "email": "licihad@corp.io", "company": "Soylent Corp", "phone": "(569) 899-7529", "rate": 0.244}, {"signup_date": "2023-03-17", "amount": 6657.76, "email": "fdcnog@example.com", "company": "Initech", "phone": "(581) 234-6019", "rate": 0.8390000000000001}, {"signup_date": "2023-08-07", "amount": 7266.54, "email": "cpnmml@test.org", "company": "Vehement", "phone": "(597) 345-6163", "rate": 0.6609999999999999}, {"signup_date": "2023-12-05", "amount": 7805.3, "email": "ibjlhhp@corp.io", "company": "Wonka Co", "phone": "(558) 195-3040", "rate": 0.861}, {"signup_date": "2023-05-09", "amount": 4703.68, "email": "mdfj@test.org", "company": "Globex", "phone": "(588) 417-4157", "rate": 0.5589999999999999}, {"signup_date": "2023-11-02", "amount": 8175.68, "email": "eihiggn@corp.io", "company": "Acme Inc", "phone": "(545) 847-4045", "rate": 0.292}, {"signup_date": "2023-08-17", "amount": 4599.82, "email": "pgmdol@test.org", "company": "Cyberdyne", "phone": "(536) 723-9337", "rate": 0.35}, {"signup_date": "2023-11-27", "amount": 7677.85, "email": "kmiljio@example.com", "company": "Initech", "phone": "(576) 344-5497", "rate": 0.912}, {"signup_date": "2023-11-06", "amount": 4812.97, "email": "mkeoija@corp.io", "company": "Umbrella", "phone": "(514) 050-0491", "rate": 0.171}, {"signup_date": "2023-01-14", "amount": 6382.57, "email": "iapnj@mail.com", "company": "Initech", "phone": "(590) 201-0955", "rate": 0.9420000000000001}, {"signup_date": "2023-11-14", "amount": 896.79, "email": "fblkce@corp.io", "company": "Soylent Corp", "phone": "(514) 169-9665", "rate": 0.6759999999999999}, {"signup_date": "2023-03-05", "amount": 6045.79, "email": "ikphga@example.com", "company": "Wonka Co", "phone": "(500) 286-3327", "rate": 0.748}, {"signup_date": "2023-11-19", "amount": 5698.35, "email": "ogfeml@mail.com", "company": "Soylent Corp", "phone": "(504) 158-4711", "rate": 0.29}, {"signup_date": "2023-12-11", "amount": 4921.43, "email": "gcifpan@test.org", "company": "Wonka Co", "phone": "(569) 657-4907", "rate": 0.41100000000000003}, {"signup_date": "2023-03-20", "amount": 4574.84, "email": "ncefbl@mail.com", "company": "Globex", "phone": "(511) 574-0398", "rate": 0.498}, {"signup_date": "2023-12-11", "amount": 2176.94, "email": "ccdbdkb@example.com", "company": "Wonka Co", "phone": "(502) 086-7097", "rate": 0.317}, {"signup_date": "2023-11-03", "amount": 1733.67, "email": "ibba@corp.io", "company": "Globex", "phone": "(595) 075-8147", "rate": 0.9690000000000001}, {"signup_date": "2023-04-20", "amount": 5410.04, "email": "idnl@test.org", "company": "Cyberdyne", "phone": "(536) 940-2045", "rate": 0.24600000000000002}, {"signup_date": "2023-11-07", "amount": 5232.39, "email": "dfmo@mail.com", "company": "Soylent Corp", "phone": "(545) 471-4627", "rate": 0.5579999999999999}, {"signup_date": "2023-09-07", "amount": 2018.36, "email": "hgkhepj@corp.io", "company": "Acme Inc", "phone": "(506) 764-9331", "rate": 0.733}, {"signup_date": "2023-12-12", "amount": 274.67, "email": "lapdi@corp.io", "company": "Globex", "phone": "(502) 730-7986", "rate": 0.193}, {"signup_date": "2023-12-18", "amount": 2699.61, "email": "dobcli@corp.io", "company": "Initech", "phone": "(561) 289-8761", "rate": 0.884}, {"signup_date": "2023-01-10", "amount": NaN, "email": "kegi@test.org", "company": "Acme Inc", "phone": "(504) 369-8586", "rate": 0.046}, {"signup_date": "2023-03-03", "amount": 6133.16, "email": "ohbg@example.com", "company": "Soylent Corp", "phone": "(587) 288-5300", "rate": 0.354}, {"signup_date": "2023-12-19", "amount": 2944.01, "email": "khplpp@example.com", "company": "Globex", "phone": "(595) 493-0642", "rate": 0.698}, {"signup_date": "2023-02-19", "amount": 4853.79, "email": "nmnmcdl@mail.com", "company": "Hooli", "phone": "(587) 499-5637", "rate": 0.419}, {"signup_date": "2023-05-07", "amount": 7751.91, "email": "mjefpfj@example.com", "company": "Vehement", "phone": "(516) 691-5406", "rate": 0.937}, {"signup_date": "2023-02-05", "amount": NaN, "email": "ognog@corp.io", "company": "Stark Industries", "phone": "(544) 761-5093", "rate": 0.993}, {"signup_date": "2023-05-12", "amount": 4456.68, "email": "ccmo@test.org", "company": "Wonka Co", "phone": "(549) 113-9473", "rate": 0.669}, {"signup_date": "2023-01-28", "amount": 1166.11, "email": "knha@corp.io", "company": "Initech", "phone": "(569) 217-9343", "rate": 0.546}], "dirty_cols": ["signup_date", "amount", "email", "company", "phone", "rate", "notes2"], "clean_cols": ["signup_date", "amount", "email", "company", "phone", "rate"], "plan": {"dataset_summary": "41 rows × 7 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"city": "Loei", "signup_date": "45052", "currency": "HUF", "phone": "5650468291", "is_active": "Yes", "unnamed": ""}, {"city": "Aston", "signup_date": "4/24/2023", "currency": "CVE", "phone": "525-251-5026", "is_active": "0", "unnamed": ""}, {"city": "Calasiao", "signup_date": "44967", "currency": "CVE", "phone": "511-486-7995", "is_active": "Y", "unnamed": ""}, {"city": "Calasiao", "signup_date": "45238", "currency": "huf", "phone": "(574)431-8874", "is_active": "Y", "unnamed": ""}, {"city": "Calasiao", "signup_date": "10/24/2023", "currency": "BOB", "phone": "595-214-3836", "is_active": "N", "unnamed": ""}, {"city": "Asqon", "signup_date": "45191", "currency": "Cabo Verde Escudo", "phone": "5275663486", "is_active": "Y", "unnamed": ""}, {"city": "Calasiao", "signup_date": "45079", "currency": "BOB", "phone": "5401121227", "is_active": "TRUE", "unnamed": ""}, {"city": "Aston", "signup_date": "14 Jan 2023", "currency": "CABO VERDE ESCUDO", "phone": "500-310-0581", "is_active": "true", "unnamed": ""}, {"city": "LOEI", "signup_date": "5/5/2023", "currency": "FORINT", "phone": "531-139-0590", "is_active": "0", "unnamed": ""}, {"city": "ATON", "signup_date": "2/11/2023", "currency": "BOB", "phone": "(562)058-4828", "is_active": "No", "unnamed": ""}, {"city": " Calasiao", "signup_date": "45180", "currency": "BOB", "phone": "(556)937-0523", "is_active": "TRUE", "unnamed": ""}, {"city": " Calasiao", "signup_date": "2023-01-04", "currency": " BOB ", "phone": "589-539-6361", "is_active": "TRUE", "unnamed": ""}, {"city": "Tatarka", "signup_date": "6/8/2023", "currency": "Cabo Verde Escudo", "phone": "5153061808", "is_active": "TRUE", "unnamed": ""}, {"city": "Tatarak", "signup_date": "45281", "currency": "cabo verde escudo", "phone": "5197176295", "is_active": "N", "unnamed": ""}, {"city": "Loei", "signup_date": "4 Nov 2023", "currency": "cve  ", "phone": "583.483.9393", "is_active": "1", "unnamed": ""}, {"city": "Tatarka  ", "signup_date": "45270", "currency": "bob", "phone": "(573)284-1862", "is_active": "No", "unnamed": ""}, {"city": "Calasiao", "signup_date": "45191", "currency": "cabo verde escudo", "phone": "(561)326-7353", "is_active": "T", "unnamed": ""}, {"city": "loei", "signup_date": "45222", "currency": " forint ", "phone": "503.050.0278", "is_active": "0", "unnamed": ""}, {"city": "calasiiao", "signup_date": "1 Nov 2023", "currency": "CVE", "phone": "(597)036-7871", "is_active": "T", "unnamed": ""}, {"city": "Tatarka", "signup_date": "2023-02-03", "currency": "Boliviano", "phone": "(504)509-6222", "is_active": "TRUE", "unnamed": ""}, {"city": "aston", "signup_date": "23 Sep 2023", "currency": "Cabo Verde Escudo", "phone": "530-662-6778", "is_active": "F", "unnamed": ""}, {"city": "", "signup_date": "", "currency": "", "phone": "", "is_active": "", "unnamed": ""}], "clean": [{"city": "Loei", "signup_date": "2023-05-06", "currency": "HUF", "phone": "(565) 046-8291", "is_active": true}, {"city": "Aston", "signup_date": "2023-04-24", "currency": "CVE", "phone": "(525) 251-5026", "is_active": false}, {"city": "Calasiao", "signup_date": "2023-02-10", "currency": "CVE", "phone": "(511) 486-7995", "is_active": true}, {"city": "Calasiao", "signup_date": "2023-11-08", "currency": "HUF", "phone": "(574) 431-8874", "is_active": true}, {"city": "Calasiao", "signup_date": "2023-10-24", "currency": "BOB", "phone": "(595) 214-3836", "is_active": false}, {"city": "Aston", "signup_date": "2023-09-22", "currency": "CVE", "phone": "(527) 566-3486", "is_active": true}, {"city": "Calasiao", "signup_date": "2023-06-02", "currency": "BOB", "phone": "(540) 112-1227", "is_active": true}, {"city": "Aston", "signup_date": "2023-01-14", "currency": "CVE", "phone": "(500) 310-0581", "is_active": true}, {"city": "Loei", "signup_date": "2023-05-05", "currency": "HUF", "phone": "(531) 139-0590", "is_active": false}, {"city": "Aston", "signup_date": "2023-02-11", "currency": "BOB", "phone": "(562) 058-4828", "is_active": false}, {"city": "Calasiao", "signup_date": "2023-09-11", "currency": "BOB", "phone": "(556) 937-0523", "is_active": true}, {"city": "Calasiao", "signup_date": "2023-01-04", "currency": "BOB", "phone": "(589) 539-6361", "is_active": true}, {"city": "Tatarka", "signup_date": "2023-06-08", "currency": "CVE", "phone": "(515) 306-1808", "is_active": true}, {"city": "Tatarka", "signup_date": "2023-12-21", "currency": "CVE", "phone": "(519) 717-6295", "is_active": false}, {"city": "Loei", "signup_date": "2023-11-04", "currency": "CVE", "phone": "(583) 483-9393", "is_active": true}, {"city": "Tatarka", "signup_date": "2023-12-10", "currency": "BOB", "phone": "(573) 284-1862", "is_active": false}, {"city": "Calasiao", "signup_date": "2023-09-22", "currency": "CVE", "phone": "(561) 326-7353", "is_active": true}, {"city": "Loei", "signup_date": "2023-10-23", "currency": "HUF", "phone": "(503) 050-0278", "is_active": false}, {"city": "Calasiao", "signup_date": "2023-11-01", "currency": "CVE", "phone": "(597) 036-7871", "is_active": true}, {"city": "Tatarka", "signup_date": "2023-02-03", "currency": "BOB", "phone": "(504) 509-6222", "is_active": true}, {"city": "Aston", "signup_date": "2023-09-23", "currency": "CVE", "phone": "(530) 662-6778", "is_active": false}], "dirty_cols": ["city", "signup_date", "currency", "phone", "is_active", "unnamed"], "clean_cols": ["city", "signup_date", "currency", "phone", "is_active"], "plan": {"dataset_summary": "22 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Asqon": "Aston", "LOEI": "Loei", "ATON": "Aston", "Tatarak": "Tatarka", "loei": "Loei", "calasiiao": "Calasiao", "aston": "Aston"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"huf": "HUF", "Cabo Verde Escudo": "CVE", "CABO VERDE ESCUDO": "CVE", "FORINT": "HUF", "cabo verde escudo": "CVE", "cve": "CVE", "bob": "BOB", "forint": "HUF", "Boliviano": "BOB"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"city": "castro-urdiales", "currency": "Boliviano", "signup_date": "7 Feb 2023", "company": " Globex ", "name": "Alice Lee"}, {"city": "  Nakrekal", "currency": "COLOMBIAN PESO", "signup_date": "44966", "company": "Soylent Corp", "name": "Lena Moore"}, {"city": "castro-urdiales", "currency": "ETB", "signup_date": "5/1/2023", "company": "Wonka Co", "name": "Judy Johnson"}, {"city": "Castro-Urdiales", "currency": "Colombian Peso", "signup_date": "45275", "company": "Cyberdyne ", "name": "   Bob Smith"}, {"city": "castro-urdiales", "currency": "Colombian  Peso", "signup_date": "19 Apr 2023", "company": "Soylent Corp   ", "name": "Lena Reyes"}, {"city": "nakrekal", "currency": "ETHIOPIAN BIRR", "signup_date": "12/1/2023", "company": "Acme Inc   ", "name": "David Adams"}, {"city": "castro-urdiales", "currency": "bob", "signup_date": "25 Oct 2023", "company": " Vehement ", "name": " Grace Ali"}, {"city": "Ner", "currency": " cop", "signup_date": "2023-12-05", "company": " Initech ", "name": "Mona Lee"}, {"city": "Castro-Urdiales", "currency": "Guyana Dollar", "signup_date": "45186", "company": " Vehement ", "name": "  Carol Reyes"}, {"city": "Ner   ", "currency": " gyd", "signup_date": "45187", "company": " Vehement", "name": "Carol  Smith"}, {"city": " CASTRO-URDIALES", "currency": "bob", "signup_date": "2023-05-09", "company": "Vehement", "name": "David  Moore"}, {"city": " Castro-Urdiales", "currency": "ETB  ", "signup_date": "2023-01-17", "company": "Wonka Co", "name": "Karl  Adams"}], "clean": [{"city": "Castro-Urdiales", "currency": "BOB", "signup_date": "2023-02-07", "company": "Globex", "name": "Alice Lee"}, {"city": "Nakrekal", "currency": "COP", "signup_date": "2023-02-09", "company": "Soylent Corp", "name": "Lena Moore"}, {"city": "Castro-Urdiales", "currency": "ETB", "signup_date": "2023-05-01", "company": "Wonka Co", "name": "Judy Johnson"}, {"city": "Castro-Urdiales", "currency": "COP", "signup_date": "2023-12-15", "company": "Cyberdyne", "name": "Bob Smith"}, {"city": "Castro-Urdiales", "currency": "COP", "signup_date": "2023-04-19", "company": "Soylent Corp", "name": "Lena Reyes"}, {"city": "Nakrekal", "currency": "ETB", "signup_date": "2023-12-01", "company": "Acme Inc", "name": "David Adams"}, {"city": "Castro-Urdiales", "currency": "BOB", "signup_date": "2023-10-25", "company": "Vehement", "name": "Grace Ali"}, {"city": "Ner", "currency": "COP", "signup_date": "2023-12-05", "company": "Initech", "name": "Mona Lee"}, {"city": "Castro-Urdiales", "currency": "GYD", "signup_date": "2023-09-17", "company": "Vehement", "name": "Carol Reyes"}, {"city": "Ner", "currency": "GYD", "signup_date": "2023-09-18", "company": "Vehement", "name": "Carol Smith"}, {"city": "Castro-Urdiales", "currency": "BOB", "signup_date": "2023-05-09", "company": "Vehement", "name": "David Moore"}, {"city": "Castro-Urdiales", "currency": "ETB", "signup_date": "2023-01-17", "company": "Wonka Co", "name": "Karl Adams"}], "dirty_cols": ["city", "currency", "signup_date", "company", "name"], "clean_cols": ["city", "currency", "signup_date", "company", "name"], "plan": {"dataset_summary": "12 rows × 5 columns. 5 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"castro-urdiales": "Castro-Urdiales", "nakrekal": "Nakrekal", "CASTRO-URDIALES": "Castro-Urdiales"}, "rationale": "Unified 3 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Boliviano": "BOB", "COLOMBIAN PESO": "COP", "Colombian Peso": "COP", "ETHIOPIAN BIRR": "ETB", "bob": "BOB", "cop": "COP", "Guyana Dollar": "GYD", "gyd": "GYD"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"city": "Amparafaravola", "email": "oaaghk@mail.com", "department": " Fin", "unit": "mls", "signup_date": "5/1/2023", "company": "Acme Inc"}, {"city": "Amparafaravola", "email": "NNMIP@CORP.IO  ", "department": "DEV", "unit": "milliliter ", "signup_date": "3 Apr 2023", "company": " Cyberdyne "}, {"city": "Rome", "email": " BOLOJIG@CORP.IO ", "department": "Finane", "unit": "  liter", "signup_date": "28 Jan 2023", "company": "Stark Industries"}, {"city": "ROME", "email": "LAPDNN@CORP.IO", "department": " I.T. ", "unit": "ml", "signup_date": "21 Sep 2023", "company": "Initech"}, {"city": "RTME", "email": " mbhebi@example.com ", "department": "CUSTOMER SUPPORT", "unit": "l", "signup_date": "26 Feb 2023", "company": " Cyberdyne "}, {"city": "amparafaravola", "email": "FDHLEDD@MAIL.COM   ", "department": "CUSTOMER SUPPORT", "unit": "ML", "signup_date": "25 Mar 2023", "company": "Stark Industries "}, {"city": "rome", "email": "kdmja@example.com", "department": "i.tt.", "unit": "ML", "signup_date": "11/4/2023", "company": "Acme Inc"}, {"city": "Amparafaravolla", "email": "BOACAJK@EXAMPLE.COM", "department": " Fin", "unit": "   liters", "signup_date": "2023-04-23", "company": "Soylent Corp"}, {"city": "Santa Catarina ", "email": " KGOPMI@CORP.IO ", "department": "IT  ", "unit": "liter", "signup_date": "2023-05-02", "company": "Vehement"}, {"city": "  amparafaravola", "email": "HCMOL@EXAMPLE.COM", "department": "CUSTOMER SUPPRT", "unit": "liters", "signup_date": "2023-07-22", "company": "Initech"}, {"city": "Santa Catarina", "email": "CJHIP@MAIL.COM", "department": "fin", "unit": "ML", "signup_date": "7 Sep 2023", "company": " Vehement "}, {"city": "santa catarina", "email": "ljiij@test.org", "department": "CS", "unit": "mL", "signup_date": "2023-11-14", "company": "Acme Inc"}, {"city": "Amparafaravola", "email": " JJPBML@MAIL.COM ", "department": "Fiqance   ", "unit": "litre", "signup_date": "27 Nov 2023", "company": "Umbrella"}, {"city": " amparafaravola ", "email": " BMGKEGN@CORP.IO", "department": "Cust Support", "unit": "liter", "signup_date": "45171", "company": "Cyberdyne   "}, {"city": "Amparafaravola", "email": "fblh@test.org", "department": "INFORMATION TECHNOLOGY", "unit": "literrs", "signup_date": "10/3/2023", "company": "Stark Industries  "}, {"city": "rome", "email": "FBOLNM@TEST.ORG", "department": "Cust Support", "unit": "liters", "signup_date": "19 Jul 2023", "company": "Umbrella"}, {"city": "santa catarina", "email": "dohocog@example.com", "department": "INFORMATION TECHNOLOGY", "unit": "L", "signup_date": "27 Jul 2023", "company": "Globex"}, {"city": "santa catarina", "email": "   aadidk@mail.com", "department": " IT ", "unit": "mls", "signup_date": "2023-09-20", "company": "Acme Inc  "}, {"city": "amparafaravola", "email": "fnih@mail.com", "department": "dev", "unit": "liters", "signup_date": "9/2/2023", "company": "   Initech"}], "clean": [{"city": "Amparafaravola", "email": "oaaghk@mail.com", "department": "Finance", "unit": "mL", "signup_date": "2023-05-01", "company": "Acme Inc"}, {"city": "Amparafaravola", "email": "nnmip@corp.io", "department": "Engineering", "unit": "mL", "signup_date": "2023-04-03", "company": "Cyberdyne"}, {"city": "Rome", "email": "bolojig@corp.io", "department": "Finance", "unit": "L", "signup_date": "2023-01-28", "company": "Stark Industries"}, {"city": "Rome", "email": "lapdnn@corp.io", "department": "Information Technology", "unit": "mL", "signup_date": "2023-09-21", "company": "Initech"}, {"city": "Rome", "email": "mbhebi@example.com", "department": "Customer Support", "unit": "L", "signup_date": "2023-02-26", "company": "Cyberdyne"}, {"city": "Amparafaravola", "email": "fdhledd@mail.com", "department": "Customer Support", "unit": "mL", "signup_date": "2023-03-25", "company": "Stark Industries"}, {"city": "Rome", "email": "kdmja@example.com", "department": "Information Technology", "unit": "mL", "signup_date": "2023-11-04", "company": "Acme Inc"}, {"city": "Amparafaravola", "email": "boacajk@example.com", "department": "Finance", "unit": "L", "signup_date": "2023-04-23", "company": "Soylent Corp"}, {"city": "Santa Catarina", "email": "kgopmi@corp.io", "department": "Information Technology", "unit": "L", "signup_date": "2023-05-02", "company": "Vehement"}, {"city": "Amparafaravola", "email": "hcmol@example.com", "department": "Customer Support", "unit": "L", "signup_date": "2023-07-22", "company": "Initech"}, {"city": "Santa Catarina", "email": "cjhip@mail.com", "department": "Finance", "unit": "mL", "signup_date": "2023-09-07", "company": "Vehement"}, {"city": "Santa Catarina", "email": "ljiij@test.org", "department": "Customer Support", "unit": "mL", "signup_date": "2023-11-14", "company": "Acme Inc"}, {"city": "Amparafaravola", "email": "jjpbml@mail.com", "department": "Finance", "unit": "L", "signup_date": "2023-11-27", "company": "Umbrella"}, {"city": "Amparafaravola", "email": "bmgkegn@corp.io", "department": "Customer Support", "unit": "L", "signup_date": "2023-09-02", "company": "Cyberdyne"}, {"city": "Amparafaravola", "email": "fblh@test.org", "department": "Information Technology", "unit": "L", "signup_date": "2023-10-03", "company": "Stark Industries"}, {"city": "Rome", "email": "fbolnm@test.org", "department": "Customer Support", "unit": "L", "signup_date": "2023-07-19", "company": "Umbrella"}, {"city": "Santa Catarina", "email": "dohocog@example.com", "department": "Information Technology", "unit": "L", "signup_date": "2023-07-27", "company": "Globex"}, {"city": "Santa Catarina", "email": "aadidk@mail.com", "department": "Information Technology", "unit": "mL", "signup_date": "2023-09-20", "company": "Acme Inc"}, {"city": "Amparafaravola", "email": "fnih@mail.com", "department": "Engineering", "unit": "L", "signup_date": "2023-09-02", "company": "Initech"}], "dirty_cols": ["city", "email", "department", "unit", "signup_date", "company"], "clean_cols": ["city", "email", "department", "unit", "signup_date", "company"], "plan": {"dataset_summary": "19 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ROME": "Rome", "RTME": "Rome", "amparafaravola": "Amparafaravola", "rome": "Rome", "Amparafaravolla": "Amparafaravola", "santa catarina": "Santa Catarina"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fin": "Finance", "DEV": "Engineering", "Finane": "Finance", "I.T.": "Information Technology", "CUSTOMER SUPPORT": "Customer Support", "i.tt.": "Information Technology", "IT": "Information Technology", "CUSTOMER SUPPRT": "Customer Support", "fin": "Finance", "CS": "Customer Support", "Fiqance": "Finance", "Cust Support": "Customer Support", "INFORMATION TECHNOLOGY": "Information Technology", "dev": "Engineering"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mls": "mL", "milliliter": "mL", "liter": "L", "ml": "mL", "l": "L", "ML": "mL", "liters": "L", "litre": "L", "literrs": "L"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"city": " rio de janeiro", "country": "sweden", "rate": "82.0%", "currency": "CVE", "job_title": "admin assistant", "col_x": ""}, {"city": " Svendborg ", "country": "blr", "rate": "4.1%", "currency": "CVE", "job_title": "admin asst", "col_x": ""}, {"city": "svednborg", "country": "JAM", "rate": "41.4%", "currency": "Cabo Verde Escudo   ", "job_title": "Sr Engineer", "col_x": ""}, {"city": "   Rio", "country": "belarus  ", "rate": "85.5%", "currency": "Boliviano", "job_title": "CEO", "col_x": ""}, {"city": "Rio de Janeiro", "country": "finland", "rate": "60.7%", "currency": "BOB", "job_title": " ceo ", "col_x": ""}, {"city": "rio", "country": "JM", "rate": "82.8%", "currency": "  Boliiviano", "job_title": "CEO", "col_x": ""}, {"city": "svendborg", "country": "FI", "rate": "44.0%", "currency": "DJIBOUTI FRANC", "job_title": "Snr Engineer", "col_x": ""}, {"city": "   svendborg", "country": "Cyprus ", "rate": "8.5%", "currency": "bob", "job_title": "sr engineer", "col_x": ""}, {"city": "Svendborg", "country": "JAM", "rate": "16.6%", "currency": "DJIBOUTI FRANC   ", "job_title": "Admin Assitant", "col_x": ""}, {"city": "svendborg", "country": "Jamaica  ", "rate": "19.7%", "currency": "DJF", "job_title": "  Chief Executive Officer", "col_x": ""}, {"city": "rio", "country": "Jm  ", "rate": "28.8%", "currency": "djibouti franc", "job_title": "chief executive officer", "col_x": ""}, {"city": "   rib de janeiro", "country": "   Republic of Finland", "rate": "49.0%", "currency": " BTN", "job_title": "SENIOR ENGINEER", "col_x": ""}, {"city": "Rio de Janeiro", "country": "jm   ", "rate": "14.6%", "currency": " BOB ", "job_title": "admin assistant  ", "col_x": ""}, {"city": "Rio De Janeiro ", "country": "republic of cyprus", "rate": "80.3%", "currency": "bob ", "job_title": "  Admin Assistant", "col_x": ""}, {"city": "Svendborg", "country": "JAM", "rate": "16.6%", "currency": "DJIBOUTI FRANC   ", "job_title": "Admin Assitant", "col_x": ""}, {"city": "", "country": "", "rate": "", "currency": "", "job_title": "", "col_x": ""}], "clean": [{"city": "Rio de Janeiro", "country": "Sweden", "rate": 0.82, "currency": "CVE", "job_title": "Administrative Assistant"}, {"city": "Svendborg", "country": "Belarus", "rate": 0.040999999999999995, "currency": "CVE", "job_title": "Administrative Assistant"}, {"city": "Svendborg", "country": "Jamaica", "rate": 0.414, "currency": "CVE", "job_title": "Senior Engineer"}, {"city": "Rio de Janeiro", "country": "Belarus", "rate": 0.855, "currency": "BOB", "job_title": "Chief Executive Officer"}, {"city": "Rio de Janeiro", "country": "Finland", "rate": 0.607, "currency": "BOB", "job_title": "Chief Executive Officer"}, {"city": "Rio de Janeiro", "country": "Jamaica", "rate": 0.828, "currency": "BOB", "job_title": "Chief Executive Officer"}, {"city": "Svendborg", "country": "Finland", "rate": 0.44, "currency": "DJF", "job_title": "Senior Engineer"}, {"city": "Svendborg", "country": "Cyprus", "rate": 0.085, "currency": "BOB", "job_title": "Senior Engineer"}, {"city": "Svendborg", "country": "Jamaica", "rate": 0.166, "currency": "DJF", "job_title": "Administrative Assistant"}, {"city": "Svendborg", "country": "Jamaica", "rate": 0.19699999999999998, "currency": "DJF", "job_title": "Chief Executive Officer"}, {"city": "Rio de Janeiro", "country": "Jamaica", "rate": 0.28800000000000003, "currency": "DJF", "job_title": "Chief Executive Officer"}, {"city": "Rio de Janeiro", "country": "Finland", "rate": 0.49, "currency": "BTN", "job_title": "Senior Engineer"}, {"city": "Rio de Janeiro", "country": "Jamaica", "rate": 0.146, "currency": "BOB", "job_title": "Administrative Assistant"}, {"city": "Rio de Janeiro", "country": "Cyprus", "rate": 0.8029999999999999, "currency": "BOB", "job_title": "Administrative Assistant"}], "dirty_cols": ["city", "country", "rate", "currency", "job_title", "col_x"], "clean_cols": ["city", "country", "rate", "currency", "job_title"], "plan": {"dataset_summary": "16 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"rio de janeiro": "Rio de Janeiro", "svednborg": "Svendborg", "Rio": "Rio de Janeiro", "rio": "Rio de Janeiro", "svendborg": "Svendborg", "rib de janeiro": "Rio de Janeiro", "Rio De Janeiro": "Rio de Janeiro"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sweden": "Sweden", "blr": "Belarus", "JAM": "Jamaica", "belarus": "Belarus", "finland": "Finland", "JM": "Jamaica", "FI": "Finland", "Jm": "Jamaica", "Republic of Finland": "Finland", "jm": "Jamaica", "republic of cyprus": "Cyprus"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cabo Verde Escudo": "CVE", "Boliviano": "BOB", "Boliiviano": "BOB", "DJIBOUTI FRANC": "DJF", "bob": "BOB", "djibouti franc": "DJF"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"admin assistant": "Administrative Assistant", "admin asst": "Administrative Assistant", "Sr Engineer": "Senior Engineer", "CEO": "Chief Executive Officer", "ceo": "Chief Executive Officer", "Snr Engineer": "Senior Engineer", "sr engineer": "Senior Engineer", "Admin Assitant": "Administrative Assistant", "chief executive officer": "Chief Executive Officer", "SENIOR ENGINEER": "Senior Engineer", "Admin Assistant": "Administrative Assistant"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "4 Feb 2023", "company": "Wonka Co", "unit": "   litre", "amount": "$8,349.94", "city": "jeddah", "is_active": "1", "department": "Engineering", "col_x": ""}, {"signup_date": "13 Jun 2023", "company": "  Umbrella", "unit": " liter ", "amount": "6.077,39", "city": "jeddah", "is_active": "1", "department": "human  resources", "col_x": ""}, {"signup_date": "45000", "company": "   Stark Industries", "unit": "L", "amount": "$3,643.46", "city": "DANGILA", "is_active": "N", "department": "pm", "col_x": ""}, {"signup_date": "2023-10-14", "company": " Wonka Co", "unit": " L ", "amount": "1.274,31", "city": "JEDDAH", "is_active": "Y", "department": "Prod", "col_x": ""}, {"signup_date": "2023-07-09", "company": "Vehement", "unit": "pounds", "amount": "None", "city": " Dangila ", "is_active": "0", "department": "people", "col_x": ""}, {"signup_date": "5/6/2023", "company": " Globex ", "unit": "L", "amount": "?", "city": " Dangila ", "is_active": "0", "department": " product ", "col_x": ""}, {"signup_date": "2/23/2023", "company": "Umbrella", "unit": "Liters", "amount": "271,55", "city": "JEDDAH", "is_active": "F", "department": "   HR", "col_x": ""}, {"signup_date": "26 Jan 2023", "company": "Stark  Industries", "unit": "Liters", "amount": "2.471,45", "city": "Dagnila", "is_active": "F", "department": "  I.T.", "col_x": ""}, {"signup_date": "9/9/2023", "company": "Cyberdyne", "unit": "   pound", "amount": "7.936,83", "city": "Dangila", "is_active": "TRUE", "department": "Prod", "col_x": ""}, {"signup_date": "45053", "company": "  Soylent Corp", "unit": "LB", "amount": "0", "city": "Dangila", "is_active": "1", "department": "Prod", "col_x": ""}, {"signup_date": "2023-11-16", "company": "Cyberdyne", "unit": "lb", "amount": "2.574,44", "city": " dangila", "is_active": "TRUE", "department": "   Eg.", "col_x": ""}, {"signup_date": "2/18/2023", "company": "Vehement", "unit": "L", "amount": "474,16", "city": "Jeddah", "is_active": "Yes", "department": "Engineering", "col_x": ""}, {"signup_date": "2023-08-21", "company": "Umbrella", "unit": "literrs", "amount": "2.336,69", "city": "Dangia", "is_active": "Y", "department": "peopl ops", "col_x": ""}, {"signup_date": "4/2/2023", "company": "Globex", "unit": "L", "amount": "$4,786.71", "city": "  Dangila", "is_active": "F", "department": "Human Resources", "col_x": ""}, {"signup_date": "2023-02-14", "company": "Soylent  Corp", "unit": "l", "amount": "1.273,15", "city": "Dangila", "is_active": "true", "department": " PM ", "col_x": ""}, {"signup_date": "14 Aug 2023", "company": "Hooli ", "unit": "litre", "amount": "--", "city": "Dangila", "is_active": "1", "department": " Prod", "col_x": ""}, {"signup_date": "3 Dec 2023", "company": "Vehement  ", "unit": "L", "amount": "2.203,29", "city": "   dangola", "is_active": "N", "department": "Information  Technology", "col_x": ""}, {"signup_date": "45155", "company": " Initech ", "unit": "liter", "amount": "$5,104.60", "city": " Jeddah ", "is_active": "true", "department": "PM ", "col_x": ""}, {"signup_date": "2023-03-24", "company": "Acme  Inc", "unit": "lbs", "amount": "$1,662.10", "city": "JEDDAH", "is_active": "F", "department": "Pod", "col_x": ""}, {"signup_date": "45126", "company": "   Globex", "unit": "liter", "amount": "4.948,05", "city": " Jeddah ", "is_active": "false", "department": "People", "col_x": ""}, {"signup_date": "2023-08-05", "company": "Vehement", "unit": "LBS", "amount": "$7,479.41", "city": "Danglia", "is_active": "No", "department": "product", "col_x": ""}, {"signup_date": "6 Sep 2023", "company": "  Globex", "unit": "LBS", "amount": "$7,452.26", "city": "   JEDAH", "is_active": "Y", "department": "Product", "col_x": ""}, {"signup_date": "18 Sep 2023", "company": "Vehement ", "unit": "POUND", "amount": "$1,171.60", "city": "Dangila", "is_active": "No", "department": " Hr ", "col_x": ""}, {"signup_date": "2023-04-05", "company": "Acme Inc", "unit": " l ", "amount": "None", "city": "Dangila", "is_active": "TRUE", "department": "PRODUCT", "col_x": ""}, {"signup_date": "45268", "company": "Cyberdyne", "unit": "   litre", "amount": "$4,574.78", "city": "  Jeddah", "is_active": "TRUE", "department": "engineering", "col_x": ""}, {"signup_date": "2023-11-13", "company": "Vehement", "unit": "lb", "amount": "2.951,11", "city": "DANGILA", "is_active": "TRUE", "department": "IT..", "col_x": ""}, {"signup_date": "2023-02-01", "company": "Globex", "unit": "lb", "amount": "6.120,44", "city": "Dangila", "is_active": "false", "department": "PROD   ", "col_x": ""}, {"signup_date": "45119", "company": " Umbrella ", "unit": "L", "amount": "7.608,34", "city": "Jeddah ", "is_active": "FALSE", "department": "  People Ops", "col_x": ""}, {"signup_date": "11/6/2023", "company": " Stark Industries", "unit": "liter", "amount": "$1,841.05", "city": "JEDDAH", "is_active": "false", "department": " Engineering ", "col_x": ""}, {"signup_date": "2023-07-13", "company": "Globex", "unit": "l   ", "amount": "6.810,11", "city": " Jeddah ", "is_active": "Yes", "department": "Information Technology", "col_x": ""}, {"signup_date": "2023-06-12", "company": "  Stark Industries", "unit": "LB", "amount": "$5,257.19", "city": "Dangila   ", "is_active": "No", "department": "Prod", "col_x": ""}, {"signup_date": "45256", "company": " Initech ", "unit": "L", "amount": "$6,676.63", "city": "Jeddah", "is_active": "TRUE", "department": "I..T", "col_x": ""}, {"signup_date": "3/19/2023", "company": " Umbrella ", "unit": "Litre", "amount": "8.997,49", "city": "dangila", "is_active": "0", "department": "Eng.", "col_x": ""}, {"signup_date": "12/28/2023", "company": "  Initech", "unit": " liters", "amount": "$8,394.45", "city": "Dangila", "is_active": "TRUE", "department": "R&D", "col_x": ""}, {"signup_date": "23 Jul 2023", "company": " Stark Industries", "unit": "lb", "amount": "$5,546.38", "city": " Jeddah ", "is_active": "1", "department": "I.T.", "col_x": ""}, {"signup_date": "15 Jan 2023", "company": "Soylent Corp", "unit": "L", "amount": "$3,855.69", "city": "Dangila", "is_active": "FALSE", "department": "People", "col_x": ""}, {"signup_date": "45209", "company": " Umbrella", "unit": " LITER ", "amount": "$3,693.58", "city": "Jeddah", "is_active": "true", "department": "ENGINEERING", "col_x": ""}, {"signup_date": "2023-01-14", "company": "Cyberdyne", "unit": "l", "amount": "1.837,65", "city": "jeddah", "is_active": "true", "department": "PM", "col_x": ""}, {"signup_date": "15 Feb 2023", "company": "  Cyberdyne", "unit": "   liters", "amount": "6.214,56", "city": "dangila", "is_active": "Yes", "department": "dev", "col_x": ""}, {"signup_date": "27 Oct 2023", "company": "Cyberdyne", "unit": "liter", "amount": "1.367,09", "city": "Dangila", "is_active": "Y", "department": "Prod", "col_x": ""}, {"signup_date": "4/19/2023", "company": " Initech ", "unit": "   Lb", "amount": "$3,821.08", "city": "JEDDAH", "is_active": "No", "department": "It", "col_x": ""}, {"signup_date": "21 Oct 2023", "company": "Initech", "unit": "pound", "amount": "$3,019.97", "city": "Dangila", "is_active": "Yes", "department": "ENG.", "col_x": ""}, {"signup_date": "2023-06-24", "company": "Hooli", "unit": "L", "amount": "3.442,05", "city": "Dagila", "is_active": "0", "department": "it", "col_x": ""}, {"signup_date": "12/7/2023", "company": "Initech  ", "unit": "Lb", "amount": "None", "city": "Dangila", "is_active": "T", "department": "  engineering", "col_x": ""}, {"signup_date": "2023-12-09", "company": "Initech ", "unit": "L", "amount": "$3,090.31", "city": "JEDDAH", "is_active": "true", "department": "information  technology", "col_x": ""}, {"signup_date": "2023-01-02", "company": " Globex", "unit": "POUNDS", "amount": "$2,252.33", "city": "Jeeddah", "is_active": "false", "department": "PEOPLE", "col_x": ""}, {"signup_date": "", "company": "", "unit": "", "amount": "", "city": "", "is_active": "", "department": "", "col_x": ""}], "clean": [{"signup_date": "2023-02-04", "company": "Wonka Co", "unit": "L", "amount": 8349.94, "city": "Jeddah", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-06-13", "company": "Umbrella", "unit": "L", "amount": 6077.39, "city": "Jeddah", "is_active": true, "department": "Human Resources"}, {"signup_date": "2023-03-15", "company": "Stark Industries", "unit": "L", "amount": 3643.46, "city": "Dangila", "is_active": false, "department": "Product"}, {"signup_date": "2023-10-14", "company": "Wonka Co", "unit": "L", "amount": 1274.31, "city": "Jeddah", "is_active": true, "department": "Product"}, {"signup_date": "2023-07-09", "company": "Vehement", "unit": "lb", "amount": NaN, "city": "Dangila", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-05-06", "company": "Globex", "unit": "L", "amount": NaN, "city": "Dangila", "is_active": false, "department": "Product"}, {"signup_date": "2023-02-23", "company": "Umbrella", "unit": "L", "amount": 271.55, "city": "Jeddah", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-01-26", "company": "Stark Industries", "unit": "L", "amount": 2471.45, "city": "Dangila", "is_active": false, "department": "Information Technology"}, {"signup_date": "2023-09-09", "company": "Cyberdyne", "unit": "lb", "amount": 7936.83, "city": "Dangila", "is_active": true, "department": "Product"}, {"signup_date": "2023-05-07", "company": "Soylent Corp", "unit": "lb", "amount": 0.0, "city": "Dangila", "is_active": true, "department": "Product"}, {"signup_date": "2023-11-16", "company": "Cyberdyne", "unit": "lb", "amount": 2574.44, "city": "Dangila", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-02-18", "company": "Vehement", "unit": "L", "amount": 474.16, "city": "Jeddah", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-08-21", "company": "Umbrella", "unit": "L", "amount": 2336.69, "city": "Dangila", "is_active": true, "department": "Human Resources"}, {"signup_date": "2023-04-02", "company": "Globex", "unit": "L", "amount": 4786.71, "city": "Dangila", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-02-14", "company": "Soylent Corp", "unit": "L", "amount": 1273.15, "city": "Dangila", "is_active": true, "department": "Product"}, {"signup_date": "2023-08-14", "company": "Hooli", "unit": "L", "amount": NaN, "city": "Dangila", "is_active": true, "department": "Product"}, {"signup_date": "2023-12-03", "company": "Vehement", "unit": "L", "amount": 2203.29, "city": "Dangila", "is_active": false, "department": "Information Technology"}, {"signup_date": "2023-08-17", "company": "Initech", "unit": "L", "amount": 5104.6, "city": "Jeddah", "is_active": true, "department": "Product"}, {"signup_date": "2023-03-24", "company": "Acme Inc", "unit": "lb", "amount": 1662.1, "city": "Jeddah", "is_active": false, "department": "Product"}, {"signup_date": "2023-07-19", "company": "Globex", "unit": "L", "amount": 4948.05, "city": "Jeddah", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-08-05", "company": "Vehement", "unit": "lb", "amount": 7479.41, "city": "Dangila", "is_active": false, "department": "Product"}, {"signup_date": "2023-09-06", "company": "Globex", "unit": "lb", "amount": 7452.26, "city": "Jeddah", "is_active": true, "department": "Product"}, {"signup_date": "2023-09-18", "company": "Vehement", "unit": "lb", "amount": 1171.6, "city": "Dangila", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-04-05", "company": "Acme Inc", "unit": "L", "amount": NaN, "city": "Dangila", "is_active": true, "department": "Product"}, {"signup_date": "2023-12-08", "company": "Cyberdyne", "unit": "L", "amount": 4574.78, "city": "Jeddah", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-11-13", "company": "Vehement", "unit": "lb", "amount": 2951.11, "city": "Dangila", "is_active": true, "department": "Information Technology"}, {"signup_date": "2023-02-01", "company": "Globex", "unit": "lb", "amount": 6120.44, "city": "Dangila", "is_active": false, "department": "Product"}, {"signup_date": "2023-07-12", "company": "Umbrella", "unit": "L", "amount": 7608.34, "city": "Jeddah", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-11-06", "company": "Stark Industries", "unit": "L", "amount": 1841.05, "city": "Jeddah", "is_active": false, "department": "Engineering"}, {"signup_date": "2023-07-13", "company": "Globex", "unit": "L", "amount": 6810.11, "city": "Jeddah", "is_active": true, "department": "Information Technology"}, {"signup_date": "2023-06-12", "company": "Stark Industries", "unit": "lb", "amount": 5257.19, "city": "Dangila", "is_active": false, "department": "Product"}, {"signup_date": "2023-11-26", "company": "Initech", "unit": "L", "amount": 6676.63, "city": "Jeddah", "is_active": true, "department": "Information Technology"}, {"signup_date": "2023-03-19", "company": "Umbrella", "unit": "L", "amount": 8997.49, "city": "Dangila", "is_active": false, "department": "Engineering"}, {"signup_date": "2023-12-28", "company": "Initech", "unit": "L", "amount": 8394.45, "city": "Dangila", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-07-23", "company": "Stark Industries", "unit": "lb", "amount": 5546.38, "city": "Jeddah", "is_active": true, "department": "Information Technology"}, {"signup_date": "2023-01-15", "company": "Soylent Corp", "unit": "L", "amount": 3855.69, "city": "Dangila", "is_active": false, "department": "Human Resources"}, {"signup_date": "2023-10-10", "company": "Umbrella", "unit": "L", "amount": 3693.58, "city": "Jeddah", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-01-14", "company": "Cyberdyne", "unit": "L", "amount": 1837.65, "city": "Jeddah", "is_active": true, "department": "Product"}, {"signup_date": "2023-02-15", "company": "Cyberdyne", "unit": "L", "amount": 6214.56, "city": "Dangila", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-10-27", "company": "Cyberdyne", "unit": "L", "amount": 1367.09, "city": "Dangila", "is_active": true, "department": "Product"}, {"signup_date": "2023-04-19", "company": "Initech", "unit": "lb", "amount": 3821.08, "city": "Jeddah", "is_active": false, "department": "Information Technology"}, {"signup_date": "2023-10-21", "company": "Initech", "unit": "lb", "amount": 3019.97, "city": "Dangila", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-06-24", "company": "Hooli", "unit": "L", "amount": 3442.05, "city": "Dangila", "is_active": false, "department": "Information Technology"}, {"signup_date": "2023-12-07", "company": "Initech", "unit": "lb", "amount": NaN, "city": "Dangila", "is_active": true, "department": "Engineering"}, {"signup_date": "2023-12-09", "company": "Initech", "unit": "L", "amount": 3090.31, "city": "Jeddah", "is_active": true, "department": "Information Technology"}, {"signup_date": "2023-01-02", "company": "Globex", "unit": "lb", "amount": 2252.33, "city": "Jeddah", "is_active": false, "department": "Human Resources"}], "dirty_cols": ["signup_date", "company", "unit", "amount", "city", "is_active", "department", "col_x"], "clean_cols": ["signup_date", "company", "unit", "amount", "city", "is_active", "department"], "plan": {"dataset_summary": "47 rows × 8 columns. 7 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"litre": "L", "liter": "L", "pounds": "lb", "Liters": "L", "pound": "lb", "LB": "lb", "literrs": "L", "l": "L", "lbs": "lb", "LBS": "lb", "POUND": "lb", "Litre": "L", "liters": "L", "LITER": "L", "Lb": "lb", "POUNDS": "lb"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"jeddah": "Jeddah", "DANGILA": "Dangila", "JEDDAH": "Jeddah", "Dagnila": "Dangila", "dangila": "Dangila", "Dangia": "Dangila", "dangola": "Dangila", "Danglia": "Dangila", "JEDAH": "Jeddah", "Dagila": "Dangila", "Jeeddah": "Jeddah"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"human resources": "Human Resources", "pm": "Product", "Prod": "Product", "people": "Human Resources", "product": "Product", "HR": "Human Resources", "I.T.": "Information Technology", "Eg.": "Engineering", "peopl ops": "Human Resources", "PM": "Product", "Pod": "Product", "People": "Human Resources", "Hr": "Human Resources", "PRODUCT": "Product", "engineering": "Engineering", "IT..": "Information Technology", "PROD": "Product", "People Ops": "Human Resources", "I..T": "Information Technology", "Eng.": "Engineering", "R&D": "Engineering", "ENGINEERING": "Engineering", "dev": "Engineering", "It": "Information Technology", "ENG.": "Engineering", "it": "Information Technology", "information technology": "Information Technology", "PEOPLE": "Human Resources"}, "rationale": "Unified 28 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"phone": "(552)599-0901", "country": "Sweden", "rate": "42.0%", "industry": "Industrial "}, {"phone": "547-348-1986", "country": "kingdom  of sweden", "rate": "82.4%", "industry": "manufacturing"}, {"phone": "5553068142", "country": "CH ", "rate": "67.4%", "industry": "IT"}, {"phone": "(551)986-0183", "country": "swiss confederation", "rate": "87.1%", "industry": "  manufacturing"}, {"phone": "5867691166", "country": "Switzerland", "rate": "69.4%", "industry": "Software"}, {"phone": "590-621-1337", "country": " CH ", "rate": "61.4%", "industry": "tech"}, {"phone": "527.846.3303", "country": "CH", "rate": "31.2%", "industry": "tceh"}, {"phone": "543-486-3130", "country": "SWISS  CONFEDERATION", "rate": "94.4%", "industry": " mfg "}, {"phone": "5592269893", "country": "SWE", "rate": "27.7%", "industry": "Industrial"}, {"phone": "535-738-4776", "country": "ch", "rate": "44.2%", "industry": "Technology"}, {"phone": "(542)268-2437", "country": "SWE", "rate": "76.0%", "industry": "Mfg"}, {"phone": "5769251178", "country": "CHE", "rate": "63.2%", "industry": "it"}, {"phone": "5686982825", "country": "   ch", "rate": "14.6%", "industry": "industrial"}, {"phone": "570.169.9491", "country": "swe", "rate": "64.9%", "industry": "it"}, {"phone": "5683085875", "country": " SE", "rate": "1.3%", "industry": "Software"}, {"phone": "5686982825", "country": "   ch", "rate": "14.6%", "industry": "industrial"}, {"phone": "527.846.3303", "country": "CH", "rate": "31.2%", "industry": "tceh"}, {"phone": "", "country": "", "rate": "", "industry": ""}], "clean": [{"phone": "(552) 599-0901", "country": "Sweden", "rate": 0.42, "industry": "Manufacturing"}, {"phone": "(547) 348-1986", "country": "Sweden", "rate": 0.8240000000000001, "industry": "Manufacturing"}, {"phone": "(555) 306-8142", "country": "Switzerland", "rate": 0.674, "industry": "Technology"}, {"phone": "(551) 986-0183", "country": "Switzerland", "rate": 0.871, "industry": "Manufacturing"}, {"phone": "(586) 769-1166", "country": "Switzerland", "rate": 0.6940000000000001, "industry": "Technology"}, {"phone": "(590) 621-1337", "country": "Switzerland", "rate": 0.614, "industry": "Technology"}, {"phone": "(527) 846-3303", "country": "Switzerland", "rate": 0.312, "industry": "Technology"}, {"phone": "(543) 486-3130", "country": "Switzerland", "rate": 0.9440000000000001, "industry": "Manufacturing"}, {"phone": "(559) 226-9893", "country": "Sweden", "rate": 0.27699999999999997, "industry": "Manufacturing"}, {"phone": "(535) 738-4776", "country": "Switzerland", "rate": 0.442, "industry": "Technology"}, {"phone": "(542) 268-2437", "country": "Sweden", "rate": 0.76, "industry": "Manufacturing"}, {"phone": "(576) 925-1178", "country": "Switzerland", "rate": 0.632, "industry": "Technology"}, {"phone": "(568) 698-2825", "country": "Switzerland", "rate": 0.146, "industry": "Manufacturing"}, {"phone": "(570) 169-9491", "country": "Sweden", "rate": 0.649, "industry": "Technology"}, {"phone": "(568) 308-5875", "country": "Sweden", "rate": 0.013000000000000001, "industry": "Technology"}], "dirty_cols": ["phone", "country", "rate", "industry"], "clean_cols": ["phone", "country", "rate", "industry"], "plan": {"dataset_summary": "18 rows × 4 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kingdom of sweden": "Sweden", "CH": "Switzerland", "swiss confederation": "Switzerland", "SWISS CONFEDERATION": "Switzerland", "SWE": "Sweden", "ch": "Switzerland", "CHE": "Switzerland", "swe": "Sweden", "SE": "Sweden"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Industrial": "Manufacturing", "manufacturing": "Manufacturing", "IT": "Technology", "Software": "Technology", "tech": "Technology", "tceh": "Technology", "mfg": "Manufacturing", "Mfg": "Manufacturing", "it": "Technology", "industrial": "Manufacturing"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "T", "name": "Carol  Novak", "department": "People", "industry": "Technology   ", "rate": "93.4%", "unnamed": ""}, {"is_active": "N", "name": " Lena Petrov", "department": "Customer Support", "industry": "manufacturing", "rate": "85.2%", "unnamed": ""}, {"is_active": "0", "name": "David Lee", "department": "Human Resoucres", "industry": "Technology", "rate": "48.3%", "unnamed": ""}, {"is_active": "N", "name": "Ivan Cruz", "department": "Hr  ", "industry": "EDCUATION", "rate": "23.8%", "unnamed": ""}, {"is_active": "N", "name": "Grace Khan", "department": "Cust Spuport", "industry": "MFG", "rate": "58.0%", "unnamed": ""}, {"is_active": "Yes", "name": " Sara Novak", "department": "cs ", "industry": "INDUSTRIAL", "rate": "9.6%", "unnamed": ""}, {"is_active": "FALSE", "name": "Priya  Ali", "department": "Cust uSpport", "industry": " Education ", "rate": "3.8%", "unnamed": ""}, {"is_active": "Yes", "name": "Omar Reyes ", "department": "HR", "industry": " Industrial", "rate": "7.8%", "unnamed": ""}, {"is_active": "No", "name": "Alice Moore", "department": "People Ops", "industry": "INDUSTRIAL", "rate": "45.8%", "unnamed": ""}, {"is_active": "F", "name": "Heidi  Wong", "department": "HUMAN RESOURCES", "industry": "ed", "rate": "86.7%", "unnamed": ""}, {"is_active": "No", "name": "Sara Reyes", "department": "Customer Support", "industry": "SOFTWARE   ", "rate": "60.0%", "unnamed": ""}, {"is_active": "true", "name": "Grace Diaz", "department": "   People", "industry": "EDTECH", "rate": "15.9%", "unnamed": ""}, {"is_active": "1", "name": "Grace Fischer", "department": "customer support", "industry": "mfg", "rate": "93.9%", "unnamed": ""}, {"is_active": "T", "name": "Omar Khan", "department": "Cust Support", "industry": "Ed", "rate": "45.9%", "unnamed": ""}, {"is_active": "false", "name": "Lena  Adams", "department": "Support", "industry": "Tech", "rate": "40.2%", "unnamed": ""}, {"is_active": "N", "name": "Bob Khan", "department": " Human Resources", "industry": "tech", "rate": "96.7%", "unnamed": ""}, {"is_active": "FALSE", "name": "Eve Fischer ", "department": "Human Resources", "industry": "  Tech", "rate": "39.2%", "unnamed": ""}, {"is_active": "0", "name": "Eve Diaz", "department": "Support", "industry": "education   ", "rate": "24.2%", "unnamed": ""}, {"is_active": "false", "name": "Sara Reyes ", "department": "Human Resouces", "industry": "Teech", "rate": "24.6%", "unnamed": ""}, {"is_active": "No", "name": " Ivan Khan", "department": "   cs", "industry": "  Manufacturing", "rate": "67.7%", "unnamed": ""}, {"is_active": "TRUE", "name": "Bob Wong", "department": "Customer Support", "industry": "EDTECH", "rate": "9.2%", "unnamed": ""}], "clean": [{"is_active": true, "name": "Carol Novak", "department": "Human Resources", "industry": "Technology", "rate": 0.934}, {"is_active": false, "name": "Lena Petrov", "department": "Customer Support", "industry": "Manufacturing", "rate": 0.852}, {"is_active": false, "name": "David Lee", "department": "Human Resources", "industry": "Technology", "rate": 0.483}, {"is_active": false, "name": "Ivan Cruz", "department": "Human Resources", "industry": "Education", "rate": 0.23800000000000002}, {"is_active": false, "name": "Grace Khan", "department": "Customer Support", "industry": "Manufacturing", "rate": 0.58}, {"is_active": true, "name": "Sara Novak", "department": "Customer Support", "industry": "Manufacturing", "rate": 0.096}, {"is_active": false, "name": "Priya Ali", "department": "Customer Support", "industry": "Education", "rate": 0.038}, {"is_active": true, "name": "Omar Reyes", "department": "Human Resources", "industry": "Manufacturing", "rate": 0.078}, {"is_active": false, "name": "Alice Moore", "department": "Human Resources", "industry": "Manufacturing", "rate": 0.45799999999999996}, {"is_active": false, "name": "Heidi Wong", "department": "Human Resources", "industry": "Education", "rate": 0.867}, {"is_active": false, "name": "Sara Reyes", "department": "Customer Support", "industry": "Technology", "rate": 0.6}, {"is_active": true, "name": "Grace Diaz", "department": "Human Resources", "industry": "Education", "rate": 0.159}, {"is_active": true, "name": "Grace Fischer", "department": "Customer Support", "industry": "Manufacturing", "rate": 0.9390000000000001}, {"is_active": true, "name": "Omar Khan", "department": "Customer Support", "industry": "Education", "rate": 0.45899999999999996}, {"is_active": false, "name": "Lena Adams", "department": "Customer Support", "industry": "Technology", "rate": 0.402}, {"is_active": false, "name": "Bob Khan", "department": "Human Resources", "industry": "Technology", "rate": 0.9670000000000001}, {"is_active": false, "name": "Eve Fischer", "department": "Human Resources", "industry": "Technology", "rate": 0.392}, {"is_active": false, "name": "Eve Diaz", "department": "Customer Support", "industry": "Education", "rate": 0.242}, {"is_active": false, "name": "Sara Reyes", "department": "Human Resources", "industry": "Technology", "rate": 0.24600000000000002}, {"is_active": false, "name": "Ivan Khan", "department": "Customer Support", "industry": "Manufacturing", "rate": 0.677}, {"is_active": true, "name": "Bob Wong", "department": "Customer Support", "industry": "Education", "rate": 0.092}], "dirty_cols": ["is_active", "name", "department", "industry", "rate", "unnamed"], "clean_cols": ["is_active", "name", "department", "industry", "rate"], "plan": {"dataset_summary": "21 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"People": "Human Resources", "Human Resoucres": "Human Resources", "Hr": "Human Resources", "Cust Spuport": "Customer Support", "cs": "Customer Support", "Cust uSpport": "Customer Support", "HR": "Human Resources", "People Ops": "Human Resources", "HUMAN RESOURCES": "Human Resources", "customer support": "Customer Support", "Cust Support": "Customer Support", "Support": "Customer Support", "Human Resouces": "Human Resources"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"manufacturing": "Manufacturing", "EDCUATION": "Education", "MFG": "Manufacturing", "INDUSTRIAL": "Manufacturing", "Industrial": "Manufacturing", "ed": "Education", "SOFTWARE": "Technology", "EDTECH": "Education", "mfg": "Manufacturing", "Ed": "Education", "Tech": "Technology", "tech": "Technology", "education": "Education", "Teech": "Technology"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"job_title": "eng mgr", "country": "GIB", "unit": "Lb", "status": "closed lost ", "state": "WY   ", "email": "efglnep@mail.com"}, {"job_title": "Acconut Executive", "country": "Iot", "unit": "Lb", "status": "open", "state": " NM", "email": " CFGIDGB@TEST.ORG "}, {"job_title": "Eng Mgr", "country": "Gibraltar", "unit": " L ", "status": "lost", "state": "WY", "email": "GCHGG@CORP.IO"}, {"job_title": "Admin Asst", "country": "IOT", "unit": "pound", "status": " closed-won ", "state": "CT", "email": "OPBHG@EXAMPLE.COM"}, {"job_title": "ENGINEERING MGR", "country": "Io", "unit": "L", "status": "IN PROGRESS", "state": "CT", "email": "pemmm@test.org"}, {"job_title": "eng gr", "country": "  British Indian Ocean Territory", "unit": "Litre", "status": "Closd Won", "state": "CT", "email": " clhffn@test.org "}, {"job_title": " V.P.", "country": "british indian ocean territory   ", "unit": "LB", "status": "in progress", "state": "WY", "email": "   dpembbg@example.com"}, {"job_title": "Acct  Exec", "country": " IO ", "unit": "lb", "status": "  lost", "state": "Connecticut", "email": " LEBNPK@TEST.ORG "}, {"job_title": "ADMIN ASST", "country": "gibraltar", "unit": "litre", "status": "loht", "state": "connecticut", "email": "pjjbbj@test.org"}, {"job_title": "Engineening Manager", "country": "Io", "unit": "pound", "status": "closed won", "state": "Wy", "email": "kkkoof@example.com   "}, {"job_title": "Admin Asst", "country": "IO", "unit": "pounds", "status": "closed lost", "state": "WY", "email": "ifekcma@corp.io"}, {"job_title": "ENGINEERING MGR", "country": "Io", "unit": "L", "status": "IN PROGRESS", "state": "CT", "email": "pemmm@test.org"}, {"job_title": "Acconut Executive", "country": "Iot", "unit": "Lb", "status": "open", "state": " NM", "email": " CFGIDGB@TEST.ORG "}], "clean": [{"job_title": "Engineering Manager", "country": "Gibraltar", "unit": "lb", "status": "Lost", "state": "Wyoming", "email": "efglnep@mail.com"}, {"job_title": "Account Executive", "country": "British Indian Ocean Territory", "unit": "lb", "status": "Open", "state": "New Mexico", "email": "cfgidgb@test.org"}, {"job_title": "Engineering Manager", "country": "Gibraltar", "unit": "L", "status": "Lost", "state": "Wyoming", "email": "gchgg@corp.io"}, {"job_title": "Administrative Assistant", "country": "British Indian Ocean Territory", "unit": "lb", "status": "Won", "state": "Connecticut", "email": "opbhg@example.com"}, {"job_title": "Engineering Manager", "country": "British Indian Ocean Territory", "unit": "L", "status": "In Progress", "state": "Connecticut", "email": "pemmm@test.org"}, {"job_title": "Engineering Manager", "country": "British Indian Ocean Territory", "unit": "L", "status": "Won", "state": "Connecticut", "email": "clhffn@test.org"}, {"job_title": "Vice President", "country": "British Indian Ocean Territory", "unit": "lb", "status": "In Progress", "state": "Wyoming", "email": "dpembbg@example.com"}, {"job_title": "Account Executive", "country": "British Indian Ocean Territory", "unit": "lb", "status": "Lost", "state": "Connecticut", "email": "lebnpk@test.org"}, {"job_title": "Administrative Assistant", "country": "Gibraltar", "unit": "L", "status": "Lost", "state": "Connecticut", "email": "pjjbbj@test.org"}, {"job_title": "Engineering Manager", "country": "British Indian Ocean Territory", "unit": "lb", "status": "Won", "state": "Wyoming", "email": "kkkoof@example.com"}, {"job_title": "Administrative Assistant", "country": "British Indian Ocean Territory", "unit": "lb", "status": "Lost", "state": "Wyoming", "email": "ifekcma@corp.io"}], "dirty_cols": ["job_title", "country", "unit", "status", "state", "email"], "clean_cols": ["job_title", "country", "unit", "status", "state", "email"], "plan": {"dataset_summary": "13 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"eng mgr": "Engineering Manager", "Acconut Executive": "Account Executive", "Eng Mgr": "Engineering Manager", "Admin Asst": "Administrative Assistant", "ENGINEERING MGR": "Engineering Manager", "eng gr": "Engineering Manager", "V.P.": "Vice President", "Acct Exec": "Account Executive", "ADMIN ASST": "Administrative Assistant", "Engineening Manager": "Engineering Manager"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GIB": "Gibraltar", "Iot": "British Indian Ocean Territory", "IOT": "British Indian Ocean Territory", "Io": "British Indian Ocean Territory", "british indian ocean territory": "British Indian Ocean Territory", "IO": "British Indian Ocean Territory", "gibraltar": "Gibraltar"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Lb": "lb", "pound": "lb", "Litre": "L", "LB": "lb", "litre": "L", "pounds": "lb"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"closed lost": "Lost", "open": "Open", "lost": "Lost", "closed-won": "Won", "IN PROGRESS": "In Progress", "Closd Won": "Won", "in progress": "In Progress", "loht": "Lost", "closed won": "Won"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WY": "Wyoming", "NM": "New Mexico", "CT": "Connecticut", "connecticut": "Connecticut", "Wy": "Wyoming"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"state": "MISSISSIPPI", "phone": "5402787038", "email": " JGKIHA@TEST.ORG ", "signup_date": "5/23/2023", "status": "trial"}, {"state": "vermont", "phone": "(573)103-6854", "email": "BCNLI@MAIL.COM", "signup_date": "44997", "status": "Trial"}, {"state": "AK", "phone": "583-570-4099", "email": "FGCMA@EXAMPLE.COM", "signup_date": "2023-01-11", "status": "paused"}, {"state": "ak", "phone": "505-563-5978", "email": "LLFAH@TEST.ORG   ", "signup_date": "2023-10-22", "status": " paused "}, {"state": "UM", "phone": "525-848-3362", "email": "fneiij@example.com", "signup_date": "45094", "status": "cancelled"}, {"state": "Missisippi", "phone": "540-516-6859", "email": "MHFFFH@CORP.IO", "signup_date": "23 Jun 2023", "status": "   Active"}, {"state": "allaska", "phone": "509-624-7854", "email": "NJEJG@EXAMPLE.COM", "signup_date": "2023-08-19", "status": " free trial"}, {"state": "Mississippi", "phone": "5208762756", "email": "HOBBB@MAIL.COM  ", "signup_date": "2023-08-23", "status": "Trial"}, {"state": "VT", "phone": "503-936-5517", "email": "pofeg@corp.io", "signup_date": "2023-06-28", "status": "cancelled"}, {"state": " Mississippi ", "phone": "573.284.0545", "email": "  KLPENAN@CORP.IO", "signup_date": "45141", "status": "on  hold"}, {"state": "ak", "phone": "522.058.4977", "email": "NEJCFC@TEST.ORG", "signup_date": "2023-11-18", "status": "PAUSED"}, {"state": "VT", "phone": "(556)232-5907", "email": "OENDF@EXAMPLE.COM", "signup_date": "2023-03-23", "status": " Churned "}, {"state": "MS", "phone": "514.021.7746", "email": "FKCIPH@MAIL.COM", "signup_date": "45064", "status": "Trial"}, {"state": "Vermont", "phone": "5940049863", "email": "FCIKDNB@MAIL.COM", "signup_date": "21 Oct 2023", "status": "canclled "}, {"state": "mississippi", "phone": "5986566800", "email": "lcoco@example.com", "signup_date": "20 Mar 2023", "status": "trial"}, {"state": "um", "phone": "(511)026-0414", "email": "JDIMNFL@EXAMPLE.COM ", "signup_date": "5 Jul 2023", "status": "active"}, {"state": "mississippi", "phone": "(544)126-7039", "email": " ifnpjc@corp.io", "signup_date": "11/10/2023", "status": "churned"}, {"state": "Mississippi", "phone": "5080195216", "email": " gbdhm@test.org ", "signup_date": "2023-11-03", "status": " Churned "}, {"state": "   UNITED STATES MINOR OUTLYING ISLANDS", "phone": "555.546.6767", "email": "fbgc@mail.com", "signup_date": "8 Nov 2023", "status": "paused  "}, {"state": "VT", "phone": "556.616.9512", "email": "fhmffmf@example.com ", "signup_date": "26 Jul 2023", "status": "Active"}, {"state": "UM", "phone": "572-882-0660", "email": " bhml@mail.com ", "signup_date": "2023-08-28", "status": " ACTIVE "}, {"state": " MISSISSIPPI", "phone": "538.789.5376", "email": "LCHINNB@TEST.ORG", "signup_date": "2023-09-07", "status": "active"}, {"state": "MS   ", "phone": "541-972-8765", "email": "ocgmdm@mail.com", "signup_date": "12/12/2023", "status": "paused"}, {"state": "MS", "phone": "(510)875-6670", "email": "gccpai@test.org", "signup_date": "15 Nov 2023", "status": "free trial"}, {"state": "MS", "phone": "5852644834", "email": "DKPO@TEST.ORG", "signup_date": "24 Jun 2023", "status": "paused"}, {"state": "Mississippi", "phone": "5208762756", "email": "HOBBB@MAIL.COM  ", "signup_date": "2023-08-23", "status": "Trial"}, {"state": "", "phone": "", "email": "", "signup_date": "", "status": ""}], "clean": [{"state": "Mississippi", "phone": "(540) 278-7038", "email": "jgkiha@test.org", "signup_date": "2023-05-23", "status": "Trial"}, {"state": "Vermont", "phone": "(573) 103-6854", "email": "bcnli@mail.com", "signup_date": "2023-03-12", "status": "Trial"}, {"state": "Alaska", "phone": "(583) 570-4099", "email": "fgcma@example.com", "signup_date": "2023-01-11", "status": "Paused"}, {"state": "Alaska", "phone": "(505) 563-5978", "email": "llfah@test.org", "signup_date": "2023-10-22", "status": "Paused"}, {"state": "United States Minor Outlying Islands", "phone": "(525) 848-3362", "email": "fneiij@example.com", "signup_date": "2023-06-17", "status": "Churned"}, {"state": "Mississippi", "phone": "(540) 516-6859", "email": "mhfffh@corp.io", "signup_date": "2023-06-23", "status": "Active"}, {"state": "Alaska", "phone": "(509) 624-7854", "email": "njejg@example.com", "signup_date": "2023-08-19", "status": "Trial"}, {"state": "Mississippi", "phone": "(520) 876-2756", "email": "hobbb@mail.com", "signup_date": "2023-08-23", "status": "Trial"}, {"state": "Vermont", "phone": "(503) 936-5517", "email": "pofeg@corp.io", "signup_date": "2023-06-28", "status": "Churned"}, {"state": "Mississippi", "phone": "(573) 284-0545", "email": "klpenan@corp.io", "signup_date": "2023-08-03", "status": "Paused"}, {"state": "Alaska", "phone": "(522) 058-4977", "email": "nejcfc@test.org", "signup_date": "2023-11-18", "status": "Paused"}, {"state": "Vermont", "phone": "(556) 232-5907", "email": "oendf@example.com", "signup_date": "2023-03-23", "status": "Churned"}, {"state": "Mississippi", "phone": "(514) 021-7746", "email": "fkciph@mail.com", "signup_date": "2023-05-18", "status": "Trial"}, {"state": "Vermont", "phone": "(594) 004-9863", "email": "fcikdnb@mail.com", "signup_date": "2023-10-21", "status": "Churned"}, {"state": "Mississippi", "phone": "(598) 656-6800", "email": "lcoco@example.com", "signup_date": "2023-03-20", "status": "Trial"}, {"state": "United States Minor Outlying Islands", "phone": "(511) 026-0414", "email": "jdimnfl@example.com", "signup_date": "2023-07-05", "status": "Active"}, {"state": "Mississippi", "phone": "(544) 126-7039", "email": "ifnpjc@corp.io", "signup_date": "2023-11-10", "status": "Churned"}, {"state": "Mississippi", "phone": "(508) 019-5216", "email": "gbdhm@test.org", "signup_date": "2023-11-03", "status": "Churned"}, {"state": "United States Minor Outlying Islands", "phone": "(555) 546-6767", "email": "fbgc@mail.com", "signup_date": "2023-11-08", "status": "Paused"}, {"state": "Vermont", "phone": "(556) 616-9512", "email": "fhmffmf@example.com", "signup_date": "2023-07-26", "status": "Active"}, {"state": "United States Minor Outlying Islands", "phone": "(572) 882-0660", "email": "bhml@mail.com", "signup_date": "2023-08-28", "status": "Active"}, {"state": "Mississippi", "phone": "(538) 789-5376", "email": "lchinnb@test.org", "signup_date": "2023-09-07", "status": "Active"}, {"state": "Mississippi", "phone": "(541) 972-8765", "email": "ocgmdm@mail.com", "signup_date": "2023-12-12", "status": "Paused"}, {"state": "Mississippi", "phone": "(510) 875-6670", "email": "gccpai@test.org", "signup_date": "2023-11-15", "status": "Trial"}, {"state": "Mississippi", "phone": "(585) 264-4834", "email": "dkpo@test.org", "signup_date": "2023-06-24", "status": "Paused"}], "dirty_cols": ["state", "phone", "email", "signup_date", "status"], "clean_cols": ["state", "phone", "email", "signup_date", "status"], "plan": {"dataset_summary": "27 rows × 5 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MISSISSIPPI": "Mississippi", "vermont": "Vermont", "AK": "Alaska", "ak": "Alaska", "UM": "United States Minor Outlying Islands", "Missisippi": "Mississippi", "allaska": "Alaska", "VT": "Vermont", "MS": "Mississippi", "mississippi": "Mississippi", "um": "United States Minor Outlying Islands", "UNITED STATES MINOR OUTLYING ISLANDS": "United States Minor Outlying Islands"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"trial": "Trial", "paused": "Paused", "cancelled": "Churned", "free trial": "Trial", "on hold": "Paused", "PAUSED": "Paused", "canclled": "Churned", "active": "Active", "churned": "Churned", "ACTIVE": "Active"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "United Kingdom of Great Britain and Northern Ireland  ", "department": " PM ", "is_active": "false", "name": "Priya Cruz", "amount": "$5,719.50", "unit": "Litre", "phone": "573-836-3275", "state": "ny   ", "col_x": ""}, {"country": "LUX", "department": "  PROD", "is_active": "N", "name": "Mona Brandt   ", "amount": "715,31", "unit": "Mls", "phone": "(565)911-8079", "state": "Northern Mariana Islands", "col_x": ""}, {"country": "lca", "department": "fin", "is_active": "TRUE", "name": "Judy Park", "amount": "8.242,84", "unit": "liters", "phone": "500.453.0985", "state": "  Mp", "col_x": ""}, {"country": "saint lucia", "department": "FIN", "is_active": "N", "name": "Priya Petrov", "amount": "$5,795.46", "unit": " liter ", "phone": "(577)452-1450", "state": "MP", "col_x": ""}, {"country": "Luxembourg", "department": " pm ", "is_active": "0", "name": "  Grace Diaz", "amount": "N/A", "unit": "kilometers", "phone": "(509)269-2858", "state": "NY ", "col_x": ""}, {"country": "  Saint Lucia", "department": "Accounting", "is_active": "false", "name": "David Moore", "amount": "$2,808.63", "unit": "Kilometer", "phone": "(568)536-0595", "state": "Arkanss", "col_x": ""}, {"country": "luxembourg", "department": "FIN", "is_active": "N", "name": "Priya Johnson", "amount": "6.254,23", "unit": "liters", "phone": "547-002-5961", "state": "Arkansas", "col_x": ""}, {"country": "United Kingdom", "department": "product", "is_active": "1", "name": "   Grace Park", "amount": "3.784,41", "unit": " Mi", "phone": "585-438-5629", "state": "  AR", "col_x": ""}, {"country": "Cambodia ", "department": "FIN", "is_active": "0", "name": "   Judy Ali", "amount": "2.361,87", "unit": "KILOMETER", "phone": "570.310.0845", "state": "   MP", "col_x": ""}, {"country": "LU", "department": " prod ", "is_active": "N", "name": "Lena Adams", "amount": "$4,879.92", "unit": " Km ", "phone": "(500)749-2422", "state": "New  York", "col_x": ""}, {"country": "lca", "department": "Fin", "is_active": "0", "name": "Bob  Adams", "amount": "5.123,89", "unit": "kilomeser", "phone": "(558)062-4962", "state": "ar", "col_x": ""}, {"country": "GB", "department": "   Prod", "is_active": "1", "name": "Sara Adams   ", "amount": "5.020,49", "unit": "miles", "phone": "548-142-8557", "state": " MP", "col_x": ""}, {"country": "lca", "department": "Accounting", "is_active": "0", "name": "David Park", "amount": "2.103,61", "unit": "MLS", "phone": "5336249397", "state": "MP", "col_x": ""}, {"country": "Saint Lucia", "department": "accounting", "is_active": "1", "name": "Sara Brandt", "amount": "5.107,57", "unit": " L ", "phone": "5555638436", "state": " OK", "col_x": ""}, {"country": "UK", "department": "Fin  ", "is_active": "1", "name": "   Frank Fischer", "amount": "5.652,93", "unit": "kilometers", "phone": "540-003-4036", "state": "AR  ", "col_x": ""}, {"country": " cambodia ", "department": "PM", "is_active": "TRUE", "name": "Karl Brandt   ", "amount": "$1,318.76", "unit": "Mls", "phone": "(553)931-0146", "state": "AR", "col_x": ""}, {"country": "LUXEMBOURG", "department": "Fin", "is_active": "false", "name": "Priya Johnson", "amount": "5.968,67", "unit": "Kilometers", "phone": "527-921-1044", "state": "ny", "col_x": ""}, {"country": "KH", "department": "FINANCE", "is_active": "true", "name": "Bob Adams", "amount": "0", "unit": "ml", "phone": "5516061506", "state": "New  York", "col_x": ""}, {"country": "England", "department": "product   ", "is_active": "N", "name": "Carol  Ali", "amount": "455,76", "unit": "km", "phone": "592-973-4688", "state": "northern mariana islands", "col_x": ""}, {"country": "LC", "department": "Product", "is_active": "TRUE", "name": " Grace Park", "amount": "$5,847.63", "unit": "L", "phone": "5791493329", "state": "MP", "col_x": ""}, {"country": "   Luxembourg", "department": "Fin  ", "is_active": "Y", "name": "Heidi  Reyes", "amount": "#N/A", "unit": "mi", "phone": "522-468-2387", "state": "Nwe York", "col_x": ""}, {"country": "saint lucia", "department": "FINANCE", "is_active": "N", "name": "David Cruz", "amount": "$8,734.83", "unit": "Kilometers", "phone": "565-260-0598", "state": "NEW YORK", "col_x": ""}, {"country": "engkand", "department": "poduct", "is_active": "TRUE", "name": "  Ivan Ali", "amount": "2.624,15", "unit": "Mile", "phone": "506.285.2452", "state": "oklahoma", "col_x": ""}, {"country": "Saint Lucia", "department": "Accounting", "is_active": "FALSE", "name": "   Bob Lee", "amount": "$508.96", "unit": "   milliliter", "phone": "544-758-6717", "state": "ny", "col_x": ""}, {"country": "lc", "department": "PM", "is_active": "Y", "name": "   Karl Wong", "amount": "4.375,13", "unit": "Miles", "phone": "532-934-3484", "state": "AR", "col_x": ""}], "clean": [{"country": "United Kingdom", "department": "Product", "is_active": false, "name": "Priya Cruz", "amount": 5719.5, "unit": "L", "phone": "(573) 836-3275", "state": "New York"}, {"country": "Luxembourg", "department": "Product", "is_active": false, "name": "Mona Brandt", "amount": 715.31, "unit": "mL", "phone": "(565) 911-8079", "state": "Northern Mariana Islands"}, {"country": "Saint Lucia", "department": "Finance", "is_active": true, "name": "Judy Park", "amount": 8242.84, "unit": "L", "phone": "(500) 453-0985", "state": "Northern Mariana Islands"}, {"country": "Saint Lucia", "department": "Finance", "is_active": false, "name": "Priya Petrov", "amount": 5795.46, "unit": "L", "phone": "(577) 452-1450", "state": "Northern Mariana Islands"}, {"country": "Luxembourg", "department": "Product", "is_active": false, "name": "Grace Diaz", "amount": NaN, "unit": "km", "phone": "(509) 269-2858", "state": "New York"}, {"country": "Saint Lucia", "department": "Finance", "is_active": false, "name": "David Moore", "amount": 2808.63, "unit": "km", "phone": "(568) 536-0595", "state": "Arkansas"}, {"country": "Luxembourg", "department": "Finance", "is_active": false, "name": "Priya Johnson", "amount": 6254.23, "unit": "L", "phone": "(547) 002-5961", "state": "Arkansas"}, {"country": "United Kingdom", "department": "Product", "is_active": true, "name": "Grace Park", "amount": 3784.41, "unit": "mi", "phone": "(585) 438-5629", "state": "Arkansas"}, {"country": "Cambodia", "department": "Finance", "is_active": false, "name": "Judy Ali", "amount": 2361.87, "unit": "km", "phone": "(570) 310-0845", "state": "Northern Mariana Islands"}, {"country": "Luxembourg", "department": "Product", "is_active": false, "name": "Lena Adams", "amount": 4879.92, "unit": "km", "phone": "(500) 749-2422", "state": "New York"}, {"country": "Saint Lucia", "department": "Finance", "is_active": false, "name": "Bob Adams", "amount": 5123.89, "unit": "km", "phone": "(558) 062-4962", "state": "Arkansas"}, {"country": "United Kingdom", "department": "Product", "is_active": true, "name": "Sara Adams", "amount": 5020.49, "unit": "mi", "phone": "(548) 142-8557", "state": "Northern Mariana Islands"}, {"country": "Saint Lucia", "department": "Finance", "is_active": false, "name": "David Park", "amount": 2103.61, "unit": "mL", "phone": "(533) 624-9397", "state": "Northern Mariana Islands"}, {"country": "Saint Lucia", "department": "Finance", "is_active": true, "name": "Sara Brandt", "amount": 5107.57, "unit": "L", "phone": "(555) 563-8436", "state": "Oklahoma"}, {"country": "United Kingdom", "department": "Finance", "is_active": true, "name": "Frank Fischer", "amount": 5652.93, "unit": "km", "phone": "(540) 003-4036", "state": "Arkansas"}, {"country": "Cambodia", "department": "Product", "is_active": true, "name": "Karl Brandt", "amount": 1318.76, "unit": "mL", "phone": "(553) 931-0146", "state": "Arkansas"}, {"country": "Luxembourg", "department": "Finance", "is_active": false, "name": "Priya Johnson", "amount": 5968.67, "unit": "km", "phone": "(527) 921-1044", "state": "New York"}, {"country": "Cambodia", "department": "Finance", "is_active": true, "name": "Bob Adams", "amount": 0.0, "unit": "mL", "phone": "(551) 606-1506", "state": "New York"}, {"country": "United Kingdom", "department": "Product", "is_active": false, "name": "Carol Ali", "amount": 455.76, "unit": "km", "phone": "(592) 973-4688", "state": "Northern Mariana Islands"}, {"country": "Saint Lucia", "department": "Product", "is_active": true, "name": "Grace Park", "amount": 5847.63, "unit": "L", "phone": "(579) 149-3329", "state": "Northern Mariana Islands"}, {"country": "Luxembourg", "department": "Finance", "is_active": true, "name": "Heidi Reyes", "amount": NaN, "unit": "mi", "phone": "(522) 468-2387", "state": "New York"}, {"country": "Saint Lucia", "department": "Finance", "is_active": false, "name": "David Cruz", "amount": 8734.83, "unit": "km", "phone": "(565) 260-0598", "state": "New York"}, {"country": "United Kingdom", "department": "Product", "is_active": true, "name": "Ivan Ali", "amount": 2624.15, "unit": "mi", "phone": "(506) 285-2452", "state": "Oklahoma"}, {"country": "Saint Lucia", "department": "Finance", "is_active": false, "name": "Bob Lee", "amount": 508.96, "unit": "mL", "phone": "(544) 758-6717", "state": "New York"}, {"country": "Saint Lucia", "department": "Product", "is_active": true, "name": "Karl Wong", "amount": 4375.13, "unit": "mi", "phone": "(532) 934-3484", "state": "Arkansas"}], "dirty_cols": ["country", "department", "is_active", "name", "amount", "unit", "phone", "state", "col_x"], "clean_cols": ["country", "department", "is_active", "name", "amount", "unit", "phone", "state"], "plan": {"dataset_summary": "25 rows × 9 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"United Kingdom of Great Britain and Northern Ireland": "United Kingdom", "LUX": "Luxembourg", "lca": "Saint Lucia", "saint lucia": "Saint Lucia", "luxembourg": "Luxembourg", "LU": "Luxembourg", "GB": "United Kingdom", "UK": "United Kingdom", "cambodia": "Cambodia", "LUXEMBOURG": "Luxembourg", "KH": "Cambodia", "England": "United Kingdom", "LC": "Saint Lucia", "engkand": "United Kingdom", "lc": "Saint Lucia"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PM": "Product", "PROD": "Product", "fin": "Finance", "FIN": "Finance", "pm": "Product", "Accounting": "Finance", "product": "Product", "prod": "Product", "Fin": "Finance", "Prod": "Product", "accounting": "Finance", "FINANCE": "Finance", "poduct": "Product"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Litre": "L", "Mls": "mL", "liters": "L", "liter": "L", "kilometers": "km", "Kilometer": "km", "Mi": "mi", "KILOMETER": "km", "Km": "km", "kilomeser": "km", "miles": "mi", "MLS": "mL", "Kilometers": "km", "ml": "mL", "Mile": "mi", "milliliter": "mL", "Miles": "mi"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ny": "New York", "Mp": "Northern Mariana Islands", "MP": "Northern Mariana Islands", "NY": "New York", "Arkanss": "Arkansas", "AR": "Arkansas", "ar": "Arkansas", "OK": "Oklahoma", "northern mariana islands": "Northern Mariana Islands", "Nwe York": "New York", "NEW YORK": "New York", "oklahoma": "Oklahoma"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"phone": "507-119-7944", "industry": " IT ", "status": "WIP", "signup_date": "26 Mar 2023", "department": "IT ", "state": "   washington", "email": "mognnei@corp.io ", "city": "Svendborg", "extra": ""}, {"phone": "5803041953", "industry": " Software ", "status": "Closed Won", "signup_date": "7/10/2023", "department": "IT", "state": "MI ", "email": "mdofmc@test.org", "city": "Perugia", "extra": ""}, {"phone": "(520)953-1282", "industry": " TECH", "status": "Ongoing", "signup_date": "8 Apr 2023", "department": "information technology", "state": " Marbland ", "email": "   HBHGCF@EXAMPLE.COM", "city": "Vladimir", "extra": ""}, {"phone": "(572)772-8997", "industry": "  tech", "status": "Closed Won", "signup_date": "45086", "department": "I.T.", "state": " NM", "email": "lopckk@corp.io", "city": "Vladiiir  ", "extra": ""}, {"phone": "529-261-4172", "industry": "Oil & Gas", "status": "Close dWon", "signup_date": "3 Nov 2023", "department": "i.t.", "state": "Michigan", "email": "jpec@test.org", "city": "PERUGIA", "extra": ""}, {"phone": "5580877151", "industry": "IT   ", "status": "In Progress", "signup_date": "5 Jan 2023", "department": " i.t. ", "state": "MD", "email": "mmco@mail.com", "city": "svendborg", "extra": ""}, {"phone": "588-571-5037", "industry": "telecom", "status": "WON", "signup_date": "3 Mar 2023", "department": "mrketing", "state": "washington", "email": "JJGNGBP@MAIL.COM  ", "city": "svendborg", "extra": ""}, {"phone": "580-297-8156", "industry": " it ", "status": "in-progress", "signup_date": "2023-08-26", "department": "Marketing", "state": "NEW MEXICO", "email": " cblddoa@corp.io ", "city": "Maragogi", "extra": ""}, {"phone": "583-233-9336", "industry": "Oil & Gas", "status": "in progress", "signup_date": "45030", "department": "Mkt", "state": "md", "email": "KFMB@MAIL.COM", "city": "   Maragogi", "extra": ""}, {"phone": "(529)591-0073", "industry": "Comms", "status": "In Progress", "signup_date": "45022", "department": "I.T.", "state": "Michxgan", "email": "  IEHH@CORP.IO", "city": "VLADIMIR", "extra": ""}, {"phone": "5830951850", "industry": "Utilities", "status": " won", "signup_date": "2023-07-11", "department": "CS", "state": "ME", "email": "DIKKD@EXAMPLE.COM", "city": "Perugia", "extra": ""}, {"phone": "5453683767", "industry": "Utilities", "status": "Won", "signup_date": "7/10/2023", "department": "IT   ", "state": "Washington", "email": "   emanfo@mail.com", "city": "Svendbkrg", "extra": ""}, {"phone": "595-469-8399", "industry": "Telecom", "status": "ongoing", "signup_date": "45250", "department": "I.T.", "state": " Michigan ", "email": "MNEECJ@EXAMPLE.COM", "city": "perugia", "extra": ""}, {"phone": "(518)204-8394", "industry": "COMMS", "status": "in-progress", "signup_date": "2023-11-06", "department": "It  ", "state": "new mexico", "email": "  GBLLFHC@CORP.IO", "city": "Perugia", "extra": ""}, {"phone": "5393651904", "industry": "utilities", "status": "ONGOING", "signup_date": "8/14/2023", "department": "marketing", "state": "washington", "email": "engod@example.com", "city": "perugia", "extra": ""}, {"phone": "(526)854-7923", "industry": "eneergy", "status": "In Progress", "signup_date": "2023-09-20", "department": "informatin technology", "state": "Maine", "email": "kfdfp@corp.io", "city": "  Vladimir", "extra": ""}, {"phone": "561.254.3906", "industry": "Tech", "status": "ongoing", "signup_date": "4/18/2023", "department": "GROWTH", "state": "Wa", "email": "KBJLJ@MAIL.COM", "city": "PERUGIA", "extra": ""}, {"phone": "(522)204-0315", "industry": "SOFTWARE", "status": "closed-won  ", "signup_date": "2/1/2023", "department": " I.T. ", "state": "NM", "email": " EEBGN@CORP.IO", "city": "VLADIMIR", "extra": ""}, {"phone": "591-587-0423", "industry": "oil &  gas", "status": "in-progress", "signup_date": "15 Aug 2023", "department": "customer support", "state": "WASHINGTON", "email": "eilgi@example.com", "city": "Svendborg", "extra": ""}, {"phone": "5809906010", "industry": "Tech", "status": "WON", "signup_date": "12/1/2023", "department": "Mktg", "state": "MD", "email": "  emdigk@example.com", "city": "  SVENDBORG", "extra": ""}, {"phone": "5332830172", "industry": "IT", "status": "won", "signup_date": "12/6/2023", "department": " GROWTH ", "state": "Washington", "email": "ldhhih@example.com", "city": "maragogi", "extra": ""}, {"phone": "565-121-1182", "industry": "Technology", "status": " Won ", "signup_date": "8/19/2023", "department": "Mktg", "state": " Maine", "email": "coel@example.com", "city": "Svendborg", "extra": ""}, {"phone": "541.002.7365", "industry": "Telecom", "status": "WIP", "signup_date": "2023-08-07", "department": "information technology", "state": " maine ", "email": " PFJNDIM@CORP.IO ", "city": "maragogi", "extra": ""}, {"phone": "520.010.2988", "industry": "Software", "status": "in progress", "signup_date": "2023-07-04", "department": "cs", "state": " New Mexico", "email": "faenckl@mail.com   ", "city": "Maragogi   ", "extra": ""}, {"phone": "(581)644-8965", "industry": "Telecom", "status": "Won", "signup_date": "19 Mar 2023", "department": "MKT", "state": "Michigan", "email": "idodc@mail.com", "city": "Maragogi ", "extra": ""}, {"phone": "541.322.1623", "industry": "Telecommunications", "status": "  closedwon", "signup_date": "2023-06-10", "department": "marketing   ", "state": "NM", "email": "lhclh@example.com", "city": "Svendborg", "extra": ""}, {"phone": "(571)234-4866", "industry": "TECH", "status": "WON", "signup_date": "2023-04-12", "department": " MARKETING ", "state": "   Me", "email": "GCNH@CORP.IO", "city": "maragogi", "extra": ""}, {"phone": "(501)910-6116", "industry": "Technology", "status": "CLOSED  WON", "signup_date": "9 Jan 2023", "department": "support", "state": "Maine", "email": "PEFLEDJ@EXAMPLE.COM", "city": "Svendborg", "extra": ""}, {"phone": "584-203-8083", "industry": "Comms", "status": "Won", "signup_date": "45195", "department": "   Marketing", "state": "WASHINGTON", "email": "   MKLK@EXAMPLE.COM", "city": "perugia", "extra": ""}, {"phone": "501.697.5211", "industry": "Oil & Gas  ", "status": "WIP", "signup_date": "45131", "department": "MKTG", "state": "  MICHIGAN", "email": "  NIPHC@EXAMPLE.COM", "city": "   Vladimir", "extra": ""}, {"phone": "(519)419-1399", "industry": "Coms", "status": "in progress", "signup_date": "1/18/2023", "department": "SUPPORT", "state": "Washington", "email": "aifiom@example.com   ", "city": "Svendborg", "extra": ""}, {"phone": "(526)742-7541", "industry": "Comms ", "status": "WON", "signup_date": "2023-01-04", "department": "MKTG", "state": "Maine", "email": " FLFC@MAIL.COM", "city": "Vladimir", "extra": ""}, {"phone": "515.954.6012", "industry": " Telco", "status": "In Progress", "signup_date": "1/7/2023", "department": " CS ", "state": "   NM", "email": "jlmkgmd@test.org  ", "city": "perugia", "extra": ""}, {"phone": "558.934.3988", "industry": "TELECOMMUNICATIONS", "status": "WON", "signup_date": "2023-07-02", "department": "CS", "state": "new mexico", "email": "lgfg@example.com", "city": "Perugia", "extra": ""}, {"phone": "588-796-9927", "industry": "It", "status": "WIP", "signup_date": "2023-11-09", "department": "GROWTH", "state": "wa", "email": "fagpjpl@test.org", "city": "Vladimir", "extra": ""}, {"phone": "(519)461-7929", "industry": "tech", "status": "CLOSED WON  ", "signup_date": "7/16/2023", "department": "MARKETING", "state": "MD", "email": "HOIJH@CORP.IO", "city": "  Svendborg", "extra": ""}, {"phone": "(508)323-3719", "industry": " ENERGY ", "status": "won", "signup_date": "1/4/2023", "department": "CS", "state": "  new mexico", "email": "gheji@test.org ", "city": " perugia", "extra": ""}, {"phone": "548-207-0290", "industry": "utilities", "status": "in-progress", "signup_date": "5/27/2023", "department": "maketing", "state": "WA", "email": "mfodj@mail.com", "city": "  Maragogi", "extra": ""}, {"phone": "5101338255", "industry": "OIL & GAS", "status": " won", "signup_date": "8 Jan 2023", "department": "support", "state": "NM", "email": " pibmil@example.com", "city": "   Perugia", "extra": ""}, {"phone": "(579)685-6554", "industry": "TECHNOLOGY", "status": "Won", "signup_date": "1/4/2023", "department": "Mktg", "state": "New Mexico", "email": " bmef@mail.com", "city": "Svendborg", "extra": ""}, {"phone": "(556)583-8265", "industry": "tech", "status": "  IN PROGRESS", "signup_date": "45248", "department": "IT", "state": "New Mexico", "email": "INEKEC@CORP.IO  ", "city": " MARAGOGI", "extra": ""}, {"phone": "512-463-1047", "industry": "telco", "status": "won", "signup_date": "6/9/2023", "department": "Support", "state": "   Michigan", "email": "lifi@example.com", "city": "Maragogi", "extra": ""}, {"phone": "530-638-8266", "industry": "Telco", "status": "Ogoing", "signup_date": "9/11/2023", "department": "customer support", "state": "Maryland", "email": "BPKOMEP@EXAMPLE.COM", "city": "maragogi", "extra": ""}, {"phone": "5523396423", "industry": "Energy", "status": "won", "signup_date": "44980", "department": "Information Technology", "state": "MI", "email": "jcnchfd@mail.com", "city": "Vladimir", "extra": ""}, {"phone": "5010692939", "industry": "   Telecommunications", "status": "WON", "signup_date": "2023-12-06", "department": "growth   ", "state": "Maryland", "email": "nohgb@mail.com", "city": "Maragoi", "extra": ""}, {"phone": "(562)582-8466", "industry": "Telco   ", "status": "in progress", "signup_date": "3 Jan 2023", "department": " I.T.", "state": "Washington", "email": "  fmiipdf@test.org", "city": "Perugia", "extra": ""}, {"phone": "576-810-3178", "industry": "Uhilities", "status": " in-progress", "signup_date": "2023-09-15", "department": " IT ", "state": "NM", "email": "cfle@example.com", "city": "Vladimir   ", "extra": ""}, {"phone": "580.204.7903", "industry": "technology  ", "status": "  wip", "signup_date": "2023-08-06", "department": "customer support", "state": "   Wshington", "email": "nlfpoio@corp.io", "city": "Svendborg", "extra": ""}, {"phone": "574-963-2488", "industry": "utilities", "status": "ongoing", "signup_date": "2023-11-19", "department": "Mkt", "state": "  MI", "email": "mhea@test.org ", "city": "Perugia", "extra": ""}, {"phone": "537-613-6114", "industry": "oil & gas", "status": "In  Pogress", "signup_date": "44984", "department": "IT", "state": "Washington", "email": " GGNEHM@MAIL.COM ", "city": " Svendborg ", "extra": ""}, {"phone": "5941636846", "industry": "Telco", "status": "   WON", "signup_date": "11 Aug 2023", "department": "I.T.", "state": "Michigan", "email": "emja@corp.io", "city": "   maragogi", "extra": ""}, {"phone": "553-985-2823", "industry": "Technology", "status": "Won", "signup_date": "14 May 2023", "department": "IT", "state": "nm", "email": "fcipldo@example.com", "city": "Svendborg", "extra": ""}, {"phone": "5279663274", "industry": "   Energy", "status": "WIP", "signup_date": "45035", "department": "I.T.", "state": "new mexico   ", "email": "BANJDJM@EXAMPLE.COM", "city": "vladimir ", "extra": ""}, {"phone": "579.332.6176", "industry": "IT", "status": "Won", "signup_date": "45288", "department": "Information Technology", "state": "MD", "email": " GDCI@CORP.IO ", "city": "   maragogi", "extra": ""}, {"phone": "529-718-5296", "industry": " Telecom", "status": "in progress", "signup_date": "8 Sep 2023", "department": " support", "state": "Michigan", "email": "PAJBLN@MAIL.COM", "city": "Perugia", "extra": ""}, {"phone": "5163591571", "industry": "IT", "status": "WON", "signup_date": "45179", "department": "I.T.", "state": "ME", "email": "LKJP@CORP.IO", "city": "maragogi", "extra": ""}, {"phone": "537-613-6114", "industry": "oil & gas", "status": "In  Pogress", "signup_date": "44984", "department": "IT", "state": "Washington", "email": " GGNEHM@MAIL.COM ", "city": " Svendborg ", "extra": ""}, {"phone": "(579)685-6554", "industry": "TECHNOLOGY", "status": "Won", "signup_date": "1/4/2023", "department": "Mktg", "state": "New Mexico", "email": " bmef@mail.com", "city": "Svendborg", "extra": ""}, {"phone": "", "industry": "", "status": "", "signup_date": "", "department": "", "state": "", "email": "", "city": "", "extra": ""}], "clean": [{"phone": "(507) 119-7944", "industry": "Technology", "status": "In Progress", "signup_date": "2023-03-26", "department": "Information Technology", "state": "Washington", "email": "mognnei@corp.io", "city": "Svendborg"}, {"phone": "(580) 304-1953", "industry": "Technology", "status": "Won", "signup_date": "2023-07-10", "department": "Information Technology", "state": "Michigan", "email": "mdofmc@test.org", "city": "Perugia"}, {"phone": "(520) 953-1282", "industry": "Technology", "status": "In Progress", "signup_date": "2023-04-08", "department": "Information Technology", "state": "Maryland", "email": "hbhgcf@example.com", "city": "Vladimir"}, {"phone": "(572) 772-8997", "industry": "Technology", "status": "Won", "signup_date": "2023-06-09", "department": "Information Technology", "state": "New Mexico", "email": "lopckk@corp.io", "city": "Vladimir"}, {"phone": "(529) 261-4172", "industry": "Energy", "status": "Won", "signup_date": "2023-11-03", "department": "Information Technology", "state": "Michigan", "email": "jpec@test.org", "city": "Perugia"}, {"phone": "(558) 087-7151", "industry": "Technology", "status": "In Progress", "signup_date": "2023-01-05", "department": "Information Technology", "state": "Maryland", "email": "mmco@mail.com", "city": "Svendborg"}, {"phone": "(588) 571-5037", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-03-03", "department": "Marketing", "state": "Washington", "email": "jjgngbp@mail.com", "city": "Svendborg"}, {"phone": "(580) 297-8156", "industry": "Technology", "status": "In Progress", "signup_date": "2023-08-26", "department": "Marketing", "state": "New Mexico", "email": "cblddoa@corp.io", "city": "Maragogi"}, {"phone": "(583) 233-9336", "industry": "Energy", "status": "In Progress", "signup_date": "2023-04-14", "department": "Marketing", "state": "Maryland", "email": "kfmb@mail.com", "city": "Maragogi"}, {"phone": "(529) 591-0073", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-04-06", "department": "Information Technology", "state": "Michigan", "email": "iehh@corp.io", "city": "Vladimir"}, {"phone": "(583) 095-1850", "industry": "Energy", "status": "Won", "signup_date": "2023-07-11", "department": "Customer Support", "state": "Maine", "email": "dikkd@example.com", "city": "Perugia"}, {"phone": "(545) 368-3767", "industry": "Energy", "status": "Won", "signup_date": "2023-07-10", "department": "Information Technology", "state": "Washington", "email": "emanfo@mail.com", "city": "Svendborg"}, {"phone": "(595) 469-8399", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-11-20", "department": "Information Technology", "state": "Michigan", "email": "mneecj@example.com", "city": "Perugia"}, {"phone": "(518) 204-8394", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-11-06", "department": "Information Technology", "state": "New Mexico", "email": "gbllfhc@corp.io", "city": "Perugia"}, {"phone": "(539) 365-1904", "industry": "Energy", "status": "In Progress", "signup_date": "2023-08-14", "department": "Marketing", "state": "Washington", "email": "engod@example.com", "city": "Perugia"}, {"phone": "(526) 854-7923", "industry": "Energy", "status": "In Progress", "signup_date": "2023-09-20", "department": "Information Technology", "state": "Maine", "email": "kfdfp@corp.io", "city": "Vladimir"}, {"phone": "(561) 254-3906", "industry": "Technology", "status": "In Progress", "signup_date": "2023-04-18", "department": "Marketing", "state": "Washington", "email": "kbjlj@mail.com", "city": "Perugia"}, {"phone": "(522) 204-0315", "industry": "Technology", "status": "Won", "signup_date": "2023-02-01", "department": "Information Technology", "state": "New Mexico", "email": "eebgn@corp.io", "city": "Vladimir"}, {"phone": "(591) 587-0423", "industry": "Energy", "status": "In Progress", "signup_date": "2023-08-15", "department": "Customer Support", "state": "Washington", "email": "eilgi@example.com", "city": "Svendborg"}, {"phone": "(580) 990-6010", "industry": "Technology", "status": "Won", "signup_date": "2023-12-01", "department": "Marketing", "state": "Maryland", "email": "emdigk@example.com", "city": "Svendborg"}, {"phone": "(533) 283-0172", "industry": "Technology", "status": "Won", "signup_date": "2023-12-06", "department": "Marketing", "state": "Washington", "email": "ldhhih@example.com", "city": "Maragogi"}, {"phone": "(565) 121-1182", "industry": "Technology", "status": "Won", "signup_date": "2023-08-19", "department": "Marketing", "state": "Maine", "email": "coel@example.com", "city": "Svendborg"}, {"phone": "(541) 002-7365", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-08-07", "department": "Information Technology", "state": "Maine", "email": "pfjndim@corp.io", "city": "Maragogi"}, {"phone": "(520) 010-2988", "industry": "Technology", "status": "In Progress", "signup_date": "2023-07-04", "department": "Customer Support", "state": "New Mexico", "email": "faenckl@mail.com", "city": "Maragogi"}, {"phone": "(581) 644-8965", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-03-19", "department": "Marketing", "state": "Michigan", "email": "idodc@mail.com", "city": "Maragogi"}, {"phone": "(541) 322-1623", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-06-10", "department": "Marketing", "state": "New Mexico", "email": "lhclh@example.com", "city": "Svendborg"}, {"phone": "(571) 234-4866", "industry": "Technology", "status": "Won", "signup_date": "2023-04-12", "department": "Marketing", "state": "Maine", "email": "gcnh@corp.io", "city": "Maragogi"}, {"phone": "(501) 910-6116", "industry": "Technology", "status": "Won", "signup_date": "2023-01-09", "department": "Customer Support", "state": "Maine", "email": "pefledj@example.com", "city": "Svendborg"}, {"phone": "(584) 203-8083", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-09-26", "department": "Marketing", "state": "Washington", "email": "mklk@example.com", "city": "Perugia"}, {"phone": "(501) 697-5211", "industry": "Energy", "status": "In Progress", "signup_date": "2023-07-24", "department": "Marketing", "state": "Michigan", "email": "niphc@example.com", "city": "Vladimir"}, {"phone": "(519) 419-1399", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-01-18", "department": "Customer Support", "state": "Washington", "email": "aifiom@example.com", "city": "Svendborg"}, {"phone": "(526) 742-7541", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-01-04", "department": "Marketing", "state": "Maine", "email": "flfc@mail.com", "city": "Vladimir"}, {"phone": "(515) 954-6012", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-01-07", "department": "Customer Support", "state": "New Mexico", "email": "jlmkgmd@test.org", "city": "Perugia"}, {"phone": "(558) 934-3988", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-07-02", "department": "Customer Support", "state": "New Mexico", "email": "lgfg@example.com", "city": "Perugia"}, {"phone": "(588) 796-9927", "industry": "Technology", "status": "In Progress", "signup_date": "2023-11-09", "department": "Marketing", "state": "Washington", "email": "fagpjpl@test.org", "city": "Vladimir"}, {"phone": "(519) 461-7929", "industry": "Technology", "status": "Won", "signup_date": "2023-07-16", "department": "Marketing", "state": "Maryland", "email": "hoijh@corp.io", "city": "Svendborg"}, {"phone": "(508) 323-3719", "industry": "Energy", "status": "Won", "signup_date": "2023-01-04", "department": "Customer Support", "state": "New Mexico", "email": "gheji@test.org", "city": "Perugia"}, {"phone": "(548) 207-0290", "industry": "Energy", "status": "In Progress", "signup_date": "2023-05-27", "department": "Marketing", "state": "Washington", "email": "mfodj@mail.com", "city": "Maragogi"}, {"phone": "(510) 133-8255", "industry": "Energy", "status": "Won", "signup_date": "2023-01-08", "department": "Customer Support", "state": "New Mexico", "email": "pibmil@example.com", "city": "Perugia"}, {"phone": "(579) 685-6554", "industry": "Technology", "status": "Won", "signup_date": "2023-01-04", "department": "Marketing", "state": "New Mexico", "email": "bmef@mail.com", "city": "Svendborg"}, {"phone": "(556) 583-8265", "industry": "Technology", "status": "In Progress", "signup_date": "2023-11-18", "department": "Information Technology", "state": "New Mexico", "email": "inekec@corp.io", "city": "Maragogi"}, {"phone": "(512) 463-1047", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-06-09", "department": "Customer Support", "state": "Michigan", "email": "lifi@example.com", "city": "Maragogi"}, {"phone": "(530) 638-8266", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-09-11", "department": "Customer Support", "state": "Maryland", "email": "bpkomep@example.com", "city": "Maragogi"}, {"phone": "(552) 339-6423", "industry": "Energy", "status": "Won", "signup_date": "2023-02-23", "department": "Information Technology", "state": "Michigan", "email": "jcnchfd@mail.com", "city": "Vladimir"}, {"phone": "(501) 069-2939", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-12-06", "department": "Marketing", "state": "Maryland", "email": "nohgb@mail.com", "city": "Maragogi"}, {"phone": "(562) 582-8466", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-01-03", "department": "Information Technology", "state": "Washington", "email": "fmiipdf@test.org", "city": "Perugia"}, {"phone": "(576) 810-3178", "industry": "Energy", "status": "In Progress", "signup_date": "2023-09-15", "department": "Information Technology", "state": "New Mexico", "email": "cfle@example.com", "city": "Vladimir"}, {"phone": "(580) 204-7903", "industry": "Technology", "status": "In Progress", "signup_date": "2023-08-06", "department": "Customer Support", "state": "Washington", "email": "nlfpoio@corp.io", "city": "Svendborg"}, {"phone": "(574) 963-2488", "industry": "Energy", "status": "In Progress", "signup_date": "2023-11-19", "department": "Marketing", "state": "Michigan", "email": "mhea@test.org", "city": "Perugia"}, {"phone": "(537) 613-6114", "industry": "Energy", "status": "In Progress", "signup_date": "2023-02-27", "department": "Information Technology", "state": "Washington", "email": "ggnehm@mail.com", "city": "Svendborg"}, {"phone": "(594) 163-6846", "industry": "Telecommunications", "status": "Won", "signup_date": "2023-08-11", "department": "Information Technology", "state": "Michigan", "email": "emja@corp.io", "city": "Maragogi"}, {"phone": "(553) 985-2823", "industry": "Technology", "status": "Won", "signup_date": "2023-05-14", "department": "Information Technology", "state": "New Mexico", "email": "fcipldo@example.com", "city": "Svendborg"}, {"phone": "(527) 966-3274", "industry": "Energy", "status": "In Progress", "signup_date": "2023-04-19", "department": "Information Technology", "state": "New Mexico", "email": "banjdjm@example.com", "city": "Vladimir"}, {"phone": "(579) 332-6176", "industry": "Technology", "status": "Won", "signup_date": "2023-12-28", "department": "Information Technology", "state": "Maryland", "email": "gdci@corp.io", "city": "Maragogi"}, {"phone": "(529) 718-5296", "industry": "Telecommunications", "status": "In Progress", "signup_date": "2023-09-08", "department": "Customer Support", "state": "Michigan", "email": "pajbln@mail.com", "city": "Perugia"}, {"phone": "(516) 359-1571", "industry": "Technology", "status": "Won", "signup_date": "2023-09-10", "department": "Information Technology", "state": "Maine", "email": "lkjp@corp.io", "city": "Maragogi"}], "dirty_cols": ["phone", "industry", "status", "signup_date", "department", "state", "email", "city", "extra"], "clean_cols": ["phone", "industry", "status", "signup_date", "department", "state", "email", "city"], "plan": {"dataset_summary": "59 rows × 9 columns. 8 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Technology", "Software": "Technology", "TECH": "Technology", "tech": "Technology", "Oil & Gas": "Energy", "telecom": "Telecommunications", "it": "Technology", "Comms": "Telecommunications", "Utilities": "Energy", "Telecom": "Telecommunications", "COMMS": "Telecommunications", "utilities": "Energy", "eneergy": "Energy", "Tech": "Technology", "SOFTWARE": "Technology", "oil & gas": "Energy", "Coms": "Telecommunications", "Telco": "Telecommunications", "TELECOMMUNICATIONS": "Telecommunications", "It": "Technology", "ENERGY": "Energy", "OIL & GAS": "Energy", "TECHNOLOGY": "Technology", "telco": "Telecommunications", "Uhilities": "Energy", "technology": "Technology"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WIP": "In Progress", "Closed Won": "Won", "Ongoing": "In Progress", "Close dWon": "Won", "WON": "Won", "in-progress": "In Progress", "in progress": "In Progress", "won": "Won", "ongoing": "In Progress", "ONGOING": "In Progress", "closed-won": "Won", "closedwon": "Won", "CLOSED WON": "Won", "IN PROGRESS": "In Progress", "Ogoing": "In Progress", "wip": "In Progress", "In Pogress": "In Progress"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "information technology": "Information Technology", "I.T.": "Information Technology", "i.t.": "Information Technology", "mrketing": "Marketing", "Mkt": "Marketing", "CS": "Customer Support", "It": "Information Technology", "marketing": "Marketing", "informatin technology": "Information Technology", "GROWTH": "Marketing", "customer support": "Customer Support", "Mktg": "Marketing", "cs": "Customer Support", "MKT": "Marketing", "MARKETING": "Marketing", "support": "Customer Support", "MKTG": "Marketing", "SUPPORT": "Customer Support", "maketing": "Marketing", "Support": "Customer Support", "growth": "Marketing"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"washington": "Washington", "MI": "Michigan", "Marbland": "Maryland", "NM": "New Mexico", "MD": "Maryland", "NEW MEXICO": "New Mexico", "md": "Maryland", "Michxgan": "Michigan", "ME": "Maine", "new mexico": "New Mexico", "Wa": "Washington", "WASHINGTON": "Washington", "maine": "Maine", "Me": "Maine", "MICHIGAN": "Michigan", "wa": "Washington", "WA": "Washington", "Wshington": "Washington", "nm": "New Mexico"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Vladiiir": "Vladimir", "PERUGIA": "Perugia", "svendborg": "Svendborg", "VLADIMIR": "Vladimir", "Svendbkrg": "Svendborg", "perugia": "Perugia", "SVENDBORG": "Svendborg", "maragogi": "Maragogi", "MARAGOGI": "Maragogi", "Maragoi": "Maragogi", "vladimir": "Vladimir"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "AL", "is_active": "1", "currency": "Etb", "unit": "MI", "rate": "75.5%"}, {"state": "AL  ", "is_active": "false", "currency": "ETB", "unit": "Kg", "rate": "26.3%"}, {"state": "Alabama", "is_active": "N", "currency": "ETHIOPIAN BIRR", "unit": "mile", "rate": "50.2%"}, {"state": " DC ", "is_active": "N", "currency": "ETB", "unit": "miles", "rate": "37.1%"}, {"state": "Aalbama", "is_active": "1", "currency": "clp", "unit": "mi", "rate": "0.5%"}, {"state": "DC", "is_active": "T", "currency": " CLP", "unit": "kg", "rate": "62.6%"}, {"state": "al", "is_active": "true", "currency": " etb", "unit": "mi", "rate": "65.0%"}, {"state": "AL", "is_active": "1", "currency": "   ETHIOPIAN BIRR", "unit": "Mi", "rate": "84.7%"}, {"state": "District of Columbia", "is_active": "Y", "currency": "Ethiopian Birr", "unit": " Miles ", "rate": "51.4%"}, {"state": "Al", "is_active": "false", "currency": "ETB", "unit": "MILES", "rate": "98.5%"}, {"state": "DC", "is_active": "true", "currency": "etb", "unit": "mile", "rate": "93.1%"}, {"state": "district of columbia", "is_active": "F", "currency": "chilean peso", "unit": "mi", "rate": "76.2%"}, {"state": "Alabama", "is_active": "F", "currency": "Ethiopian Birr", "unit": "KG", "rate": "37.4%"}, {"state": "Alabamma", "is_active": "N", "currency": "clp", "unit": "mile", "rate": "82.3%"}, {"state": "DC", "is_active": "Yes", "currency": "ETB", "unit": "mi", "rate": "18.5%"}, {"state": "   dc", "is_active": "F", "currency": " ETHIOPIAN BIRR", "unit": "kilogram", "rate": "68.6%"}, {"state": "AL  ", "is_active": "false", "currency": "ETB", "unit": "Kg", "rate": "26.3%"}], "clean": [{"state": "Alabama", "is_active": true, "currency": "ETB", "unit": "mi", "rate": 0.755}, {"state": "Alabama", "is_active": false, "currency": "ETB", "unit": "kg", "rate": 0.263}, {"state": "Alabama", "is_active": false, "currency": "ETB", "unit": "mi", "rate": 0.502}, {"state": "District of Columbia", "is_active": false, "currency": "ETB", "unit": "mi", "rate": 0.371}, {"state": "Alabama", "is_active": true, "currency": "CLP", "unit": "mi", "rate": 0.005}, {"state": "District of Columbia", "is_active": true, "currency": "CLP", "unit": "kg", "rate": 0.626}, {"state": "Alabama", "is_active": true, "currency": "ETB", "unit": "mi", "rate": 0.65}, {"state": "Alabama", "is_active": true, "currency": "ETB", "unit": "mi", "rate": 0.847}, {"state": "District of Columbia", "is_active": true, "currency": "ETB", "unit": "mi", "rate": 0.514}, {"state": "Alabama", "is_active": false, "currency": "ETB", "unit": "mi", "rate": 0.985}, {"state": "District of Columbia", "is_active": true, "currency": "ETB", "unit": "mi", "rate": 0.9309999999999999}, {"state": "District of Columbia", "is_active": false, "currency": "CLP", "unit": "mi", "rate": 0.762}, {"state": "Alabama", "is_active": false, "currency": "ETB", "unit": "kg", "rate": 0.374}, {"state": "Alabama", "is_active": false, "currency": "CLP", "unit": "mi", "rate": 0.823}, {"state": "District of Columbia", "is_active": true, "currency": "ETB", "unit": "mi", "rate": 0.185}, {"state": "District of Columbia", "is_active": false, "currency": "ETB", "unit": "kg", "rate": 0.6859999999999999}], "dirty_cols": ["state", "is_active", "currency", "unit", "rate"], "clean_cols": ["state", "is_active", "currency", "unit", "rate"], "plan": {"dataset_summary": "17 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AL": "Alabama", "DC": "District of Columbia", "Aalbama": "Alabama", "al": "Alabama", "Al": "Alabama", "district of columbia": "District of Columbia", "Alabamma": "Alabama", "dc": "District of Columbia"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Etb": "ETB", "ETHIOPIAN BIRR": "ETB", "clp": "CLP", "etb": "ETB", "Ethiopian Birr": "ETB", "chilean peso": "CLP"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MI": "mi", "Kg": "kg", "mile": "mi", "miles": "mi", "Mi": "mi", "Miles": "mi", "MILES": "mi", "KG": "kg", "kilogram": "kg"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"city": " maebashi ", "department": "human resources", "job_title": "Sales Representative", "phone": "548.142.4782"}, {"city": " Tuomuwusitang ", "department": "Sales", "job_title": "Eng Mgr", "phone": "(582)251-8961"}, {"city": "Nuuk", "department": "Biz Dev", "job_title": "sales  rep", "phone": "(582)368-4000"}, {"city": "JejuC ity", "department": "hr", "job_title": "engineering manager", "phone": "5434753105"}, {"city": "Maebashi", "department": "Sales", "job_title": "Sales Rep  ", "phone": "5766751529"}, {"city": " Tuomuwusitang", "department": "Biz Dev", "job_title": "Sales Representative", "phone": "576-673-3672"}, {"city": "Arheilgen", "department": "people", "job_title": "Enginering Mgr", "phone": "(584)661-4657"}, {"city": "Meabashi", "department": "   people ops", "job_title": "eng  mgr", "phone": "(573)176-6818"}, {"city": "Tuomuwusitang", "department": "People", "job_title": "Engineering Mgr", "phone": "(521)459-9524"}, {"city": "Maebashi", "department": "SAES", "job_title": "Prod Manager", "phone": "590-204-4685"}, {"city": "Maecashi", "department": "HR", "job_title": "Prod  Manager", "phone": "(589)022-4942"}, {"city": "TUFMUWUSITANG", "department": "People", "job_title": "Eng  Mgr", "phone": "(509)207-4107"}, {"city": "Jeju City", "department": "People Ops ", "job_title": "Engineering Manager", "phone": "5774309669"}, {"city": " tumuwusitang ", "department": "People Ops", "job_title": "PROD MANAGER", "phone": "5504523840"}, {"city": "Nuuk   ", "department": "biz dev  ", "job_title": "Rep", "phone": "5215390236"}, {"city": "", "department": "", "job_title": "", "phone": ""}], "clean": [{"city": "Maebashi", "department": "Human Resources", "job_title": "Sales Representative", "phone": "(548) 142-4782"}, {"city": "Tuomuwusitang", "department": "Sales", "job_title": "Engineering Manager", "phone": "(582) 251-8961"}, {"city": "Nuuk", "department": "Sales", "job_title": "Sales Representative", "phone": "(582) 368-4000"}, {"city": "Jeju City", "department": "Human Resources", "job_title": "Engineering Manager", "phone": "(543) 475-3105"}, {"city": "Maebashi", "department": "Sales", "job_title": "Sales Representative", "phone": "(576) 675-1529"}, {"city": "Tuomuwusitang", "department": "Sales", "job_title": "Sales Representative", "phone": "(576) 673-3672"}, {"city": "Arheilgen", "department": "Human Resources", "job_title": "Engineering Manager", "phone": "(584) 661-4657"}, {"city": "Maebashi", "department": "Human Resources", "job_title": "Engineering Manager", "phone": "(573) 176-6818"}, {"city": "Tuomuwusitang", "department": "Human Resources", "job_title": "Engineering Manager", "phone": "(521) 459-9524"}, {"city": "Maebashi", "department": "Sales", "job_title": "Product Manager", "phone": "(590) 204-4685"}, {"city": "Maebashi", "department": "Human Resources", "job_title": "Product Manager", "phone": "(589) 022-4942"}, {"city": "Tuomuwusitang", "department": "Human Resources", "job_title": "Engineering Manager", "phone": "(509) 207-4107"}, {"city": "Jeju City", "department": "Human Resources", "job_title": "Engineering Manager", "phone": "(577) 430-9669"}, {"city": "Tuomuwusitang", "department": "Human Resources", "job_title": "Product Manager", "phone": "(550) 452-3840"}, {"city": "Nuuk", "department": "Sales", "job_title": "Sales Representative", "phone": "(521) 539-0236"}], "dirty_cols": ["city", "department", "job_title", "phone"], "clean_cols": ["city", "department", "job_title", "phone"], "plan": {"dataset_summary": "16 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"maebashi": "Maebashi", "JejuC ity": "Jeju City", "Meabashi": "Maebashi", "Maecashi": "Maebashi", "TUFMUWUSITANG": "Tuomuwusitang", "tumuwusitang": "Tuomuwusitang"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"human resources": "Human Resources", "Biz Dev": "Sales", "hr": "Human Resources", "people": "Human Resources", "people ops": "Human Resources", "People": "Human Resources", "SAES": "Sales", "HR": "Human Resources", "People Ops": "Human Resources", "biz dev": "Sales"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Eng Mgr": "Engineering Manager", "sales rep": "Sales Representative", "engineering manager": "Engineering Manager", "Sales Rep": "Sales Representative", "Enginering Mgr": "Engineering Manager", "eng mgr": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "Prod Manager": "Product Manager", "PROD MANAGER": "Product Manager", "Rep": "Sales Representative"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"state": "OH   ", "is_active": "1", "amount": "7.158,90", "signup_date": "2023-01-11", "status": "Overdue", "unit": "  inch", "rate": "28.5%", "email": "DDDEC@TEST.ORG", "unnamed": ""}, {"state": "DC", "is_active": "FALSE", "amount": "$8,920.83", "signup_date": "2/18/2023", "status": "Refnded", "unit": "Cm", "rate": "71.1%", "email": "emnp@mail.com   ", "unnamed": ""}, {"state": "WV", "is_active": "F", "amount": "$1,109.23", "signup_date": "45121", "status": "Pending", "unit": "INCHES", "rate": "44.5%", "email": "bbhj@test.org", "unnamed": ""}, {"state": "west virginia", "is_active": "N", "amount": "219,51", "signup_date": "12/9/2023", "status": " Pending ", "unit": "\"", "rate": "36.8%", "email": "POOL@MAIL.COM", "unnamed": ""}, {"state": "Maine", "is_active": "true", "amount": "None", "signup_date": "2023-03-16", "status": "paid", "unit": "centimeters", "rate": "0.7%", "email": "cnib@mail.com", "unnamed": ""}, {"state": "Ohio", "is_active": "false", "amount": "$5,590.84", "signup_date": "2023-09-02", "status": "  late", "unit": "cm", "rate": "97.7%", "email": " OLKKG@CORP.IO ", "unnamed": ""}, {"state": "MAINE", "is_active": "F", "amount": "6.874,13", "signup_date": "14 Mar 2023", "status": "unpaid", "unit": "   INCHES", "rate": "31.5%", "email": "LNMMA@MAIL.COM", "unnamed": ""}, {"state": "wv", "is_active": "N", "amount": "$4,467.84", "signup_date": "45164", "status": "unpaid", "unit": "centimeter", "rate": "28.7%", "email": "  FEGJIH@EXAMPLE.COM", "unnamed": ""}, {"state": "DC", "is_active": "N", "amount": "$817.17", "signup_date": "44963", "status": "PAUD", "unit": "centimeter", "rate": "91.7%", "email": "HMHJAB@MAIL.COM", "unnamed": ""}, {"state": "WV ", "is_active": "Y", "amount": "5.656,02", "signup_date": "2 Jun 2023", "status": "PAID", "unit": " inch ", "rate": "80.7%", "email": "dcholn@test.org", "unnamed": ""}, {"state": "District Of Columbia", "is_active": "No", "amount": "683,83", "signup_date": "45093", "status": "overdue", "unit": "in", "rate": "87.6%", "email": " OJKE@EXAMPLE.COM ", "unnamed": ""}, {"state": "District of Columbia", "is_active": "FALSE", "amount": "?", "signup_date": "2023-06-10", "status": "LATE", "unit": " Innch ", "rate": "74.8%", "email": " KFJF@MAIL.COM", "unnamed": ""}, {"state": "Ohio", "is_active": "T", "amount": "1.441,70", "signup_date": "45165", "status": "pending", "unit": "  INCHES", "rate": "99.8%", "email": "kmpl@corp.io   ", "unnamed": ""}, {"state": " NE", "is_active": "TRUE", "amount": "$4,395.61", "signup_date": "2023-08-23", "status": "Paid", "unit": "   \"", "rate": "95.3%", "email": "OGIH@TEST.ORG", "unnamed": ""}, {"state": "nebraska", "is_active": "1", "amount": "6.436,12", "signup_date": "12 Sep 2023", "status": "paid", "unit": "inch", "rate": "67.5%", "email": "lojhl@example.com", "unnamed": ""}, {"state": " Me", "is_active": "0", "amount": "TBD", "signup_date": "2023-03-18", "status": "paid", "unit": "centimeters", "rate": "56.4%", "email": "aakh@mail.com", "unnamed": ""}, {"state": "OH", "is_active": "1", "amount": "$4,992.18", "signup_date": "21 Feb 2023", "status": "overdue", "unit": "cm", "rate": "24.0%", "email": " gpmd@corp.io", "unnamed": ""}, {"state": " District of Columbia", "is_active": "false", "amount": "2.130,31", "signup_date": "26 Apr 2023", "status": "PAID", "unit": "\"", "rate": "5.8%", "email": " NBKMIB@EXAMPLE.COM ", "unnamed": ""}, {"state": " wv ", "is_active": "Yes", "amount": "?", "signup_date": "3/12/2023", "status": "due", "unit": "in", "rate": "92.9%", "email": "olbdci@corp.io", "unnamed": ""}, {"state": "nebraska", "is_active": "F", "amount": "5.749,22", "signup_date": "12 Jan 2023", "status": "overdue", "unit": " cm ", "rate": "7.1%", "email": "finbkp@corp.io", "unnamed": ""}, {"state": "west virginia", "is_active": "FALSE", "amount": "$4,165.94", "signup_date": "44986", "status": "overdue", "unit": "cenntimeters", "rate": "25.5%", "email": " KKJE@TEST.ORG ", "unnamed": ""}, {"state": "Ohio ", "is_active": "true", "amount": "$1,153.05", "signup_date": "1/27/2023", "status": "  unpaaid", "unit": " Cm ", "rate": "95.8%", "email": "pefjemc@corp.io", "unnamed": ""}, {"state": "Ne   ", "is_active": "1", "amount": "$3,895.11", "signup_date": "45030", "status": " late ", "unit": "CM", "rate": "26.1%", "email": " GKKJA@MAIL.COM ", "unnamed": ""}, {"state": "district of columbia   ", "is_active": "T", "amount": "$6,539.12", "signup_date": "45001", "status": "Paid", "unit": "inch", "rate": "23.8%", "email": "jgfg@test.org", "unnamed": ""}, {"state": "   Ohio", "is_active": "F", "amount": "5.740,14", "signup_date": "45187", "status": "PENDING", "unit": "inches   ", "rate": "37.4%", "email": " ggbmgen@test.org", "unnamed": ""}, {"state": "WETS VIRGINIA", "is_active": "TRUE", "amount": "TBD", "signup_date": "2023-10-28", "status": "late", "unit": " in", "rate": "46.6%", "email": "   OJPGE@MAIL.COM", "unnamed": ""}, {"state": "WEST  VIRGINIA", "is_active": "Yes", "amount": "4.796,16", "signup_date": "9/28/2023", "status": "  due", "unit": "inches", "rate": "4.3%", "email": "foenlkj@test.org", "unnamed": ""}, {"state": "District of  Columbia", "is_active": "Yes", "amount": "4.738,63", "signup_date": "8/4/2023", "status": "due", "unit": "\"", "rate": "45.3%", "email": "algmab@corp.io", "unnamed": ""}, {"state": "west virginia", "is_active": "1", "amount": "4.185,25", "signup_date": "8 Oct 2023", "status": "paid", "unit": "inches", "rate": "8.2%", "email": "HAPJ@MAIL.COM", "unnamed": ""}, {"state": "ME", "is_active": "TRUE", "amount": "#N/A", "signup_date": "17 Nov 2023", "status": "PAID", "unit": "centimeters", "rate": "73.9%", "email": " PMOJ@EXAMPLE.COM ", "unnamed": ""}, {"state": "District of Columbia", "is_active": "true", "amount": "$7,607.40", "signup_date": "10/24/2023", "status": "paid", "unit": "cm", "rate": "61.9%", "email": " APMKNA@TEST.ORG", "unnamed": ""}, {"state": "ohio", "is_active": "No", "amount": "$4,614.33", "signup_date": "2023-12-16", "status": "late", "unit": "inch", "rate": "29.2%", "email": "DFMOO@MAIL.COM", "unnamed": ""}, {"state": "wv", "is_active": "T", "amount": "8.173,59", "signup_date": "2023-02-01", "status": " overdue ", "unit": " centimeters ", "rate": "85.5%", "email": "cihoeh@example.com", "unnamed": ""}, {"state": "", "is_active": "", "amount": "", "signup_date": "", "status": "", "unit": "", "rate": "", "email": "", "unnamed": ""}], "clean": [{"state": "Ohio", "is_active": true, "amount": 7158.9, "signup_date": "2023-01-11", "status": "Overdue", "unit": "in", "rate": 0.285, "email": "dddec@test.org"}, {"state": "District of Columbia", "is_active": false, "amount": 8920.83, "signup_date": "2023-02-18", "status": "Refunded", "unit": "cm", "rate": 0.711, "email": "emnp@mail.com"}, {"state": "West Virginia", "is_active": false, "amount": 1109.23, "signup_date": "2023-07-14", "status": "Pending", "unit": "in", "rate": 0.445, "email": "bbhj@test.org"}, {"state": "West Virginia", "is_active": false, "amount": 219.51, "signup_date": "2023-12-09", "status": "Pending", "unit": "in", "rate": 0.368, "email": "pool@mail.com"}, {"state": "Maine", "is_active": true, "amount": NaN, "signup_date": "2023-03-16", "status": "Paid", "unit": "cm", "rate": 0.006999999999999999, "email": "cnib@mail.com"}, {"state": "Ohio", "is_active": false, "amount": 5590.84, "signup_date": "2023-09-02", "status": "Overdue", "unit": "cm", "rate": 0.977, "email": "olkkg@corp.io"}, {"state": "Maine", "is_active": false, "amount": 6874.13, "signup_date": "2023-03-14", "status": "Pending", "unit": "in", "rate": 0.315, "email": "lnmma@mail.com"}, {"state": "West Virginia", "is_active": false, "amount": 4467.84, "signup_date": "2023-08-26", "status": "Pending", "unit": "cm", "rate": 0.287, "email": "fegjih@example.com"}, {"state": "District of Columbia", "is_active": false, "amount": 817.17, "signup_date": "2023-02-06", "status": "Paid", "unit": "cm", "rate": 0.917, "email": "hmhjab@mail.com"}, {"state": "West Virginia", "is_active": true, "amount": 5656.02, "signup_date": "2023-06-02", "status": "Paid", "unit": "in", "rate": 0.807, "email": "dcholn@test.org"}, {"state": "District of Columbia", "is_active": false, "amount": 683.83, "signup_date": "2023-06-16", "status": "Overdue", "unit": "in", "rate": 0.8759999999999999, "email": "ojke@example.com"}, {"state": "District of Columbia", "is_active": false, "amount": NaN, "signup_date": "2023-06-10", "status": "Overdue", "unit": "in", "rate": 0.748, "email": "kfjf@mail.com"}, {"state": "Ohio", "is_active": true, "amount": 1441.7, "signup_date": "2023-08-27", "status": "Pending", "unit": "in", "rate": 0.998, "email": "kmpl@corp.io"}, {"state": "Nebraska", "is_active": true, "amount": 4395.61, "signup_date": "2023-08-23", "status": "Paid", "unit": "in", "rate": 0.953, "email": "ogih@test.org"}, {"state": "Nebraska", "is_active": true, "amount": 6436.12, "signup_date": "2023-09-12", "status": "Paid", "unit": "in", "rate": 0.675, "email": "lojhl@example.com"}, {"state": "Maine", "is_active": false, "amount": NaN, "signup_date": "2023-03-18", "status": "Paid", "unit": "cm", "rate": 0.564, "email": "aakh@mail.com"}, {"state": "Ohio", "is_active": true, "amount": 4992.18, "signup_date": "2023-02-21", "status": "Overdue", "unit": "cm", "rate": 0.24, "email": "gpmd@corp.io"}, {"state": "District of Columbia", "is_active": false, "amount": 2130.31, "signup_date": "2023-04-26", "status": "Paid", "unit": "in", "rate": 0.057999999999999996, "email": "nbkmib@example.com"}, {"state": "West Virginia", "is_active": true, "amount": NaN, "signup_date": "2023-03-12", "status": "Pending", "unit": "in", "rate": 0.929, "email": "olbdci@corp.io"}, {"state": "Nebraska", "is_active": false, "amount": 5749.22, "signup_date": "2023-01-12", "status": "Overdue", "unit": "cm", "rate": 0.071, "email": "finbkp@corp.io"}, {"state": "West Virginia", "is_active": false, "amount": 4165.94, "signup_date": "2023-03-01", "status": "Overdue", "unit": "cm", "rate": 0.255, "email": "kkje@test.org"}, {"state": "Ohio", "is_active": true, "amount": 1153.05, "signup_date": "2023-01-27", "status": "Pending", "unit": "cm", "rate": 0.958, "email": "pefjemc@corp.io"}, {"state": "Nebraska", "is_active": true, "amount": 3895.11, "signup_date": "2023-04-14", "status": "Overdue", "unit": "cm", "rate": 0.261, "email": "gkkja@mail.com"}, {"state": "District of Columbia", "is_active": true, "amount": 6539.12, "signup_date": "2023-03-16", "status": "Paid", "unit": "in", "rate": 0.23800000000000002, "email": "jgfg@test.org"}, {"state": "Ohio", "is_active": false, "amount": 5740.14, "signup_date": "2023-09-18", "status": "Pending", "unit": "in", "rate": 0.374, "email": "ggbmgen@test.org"}, {"state": "West Virginia", "is_active": true, "amount": NaN, "signup_date": "2023-10-28", "status": "Overdue", "unit": "in", "rate": 0.466, "email": "ojpge@mail.com"}, {"state": "West Virginia", "is_active": true, "amount": 4796.16, "signup_date": "2023-09-28", "status": "Pending", "unit": "in", "rate": 0.043, "email": "foenlkj@test.org"}, {"state": "District of Columbia", "is_active": true, "amount": 4738.63, "signup_date": "2023-08-04", "status": "Pending", "unit": "in", "rate": 0.45299999999999996, "email": "algmab@corp.io"}, {"state": "West Virginia", "is_active": true, "amount": 4185.25, "signup_date": "2023-10-08", "status": "Paid", "unit": "in", "rate": 0.08199999999999999, "email": "hapj@mail.com"}, {"state": "Maine", "is_active": true, "amount": NaN, "signup_date": "2023-11-17", "status": "Paid", "unit": "cm", "rate": 0.7390000000000001, "email": "pmoj@example.com"}, {"state": "District of Columbia", "is_active": true, "amount": 7607.4, "signup_date": "2023-10-24", "status": "Paid", "unit": "cm", "rate": 0.619, "email": "apmkna@test.org"}, {"state": "Ohio", "is_active": false, "amount": 4614.33, "signup_date": "2023-12-16", "status": "Overdue", "unit": "in", "rate": 0.292, "email": "dfmoo@mail.com"}, {"state": "West Virginia", "is_active": true, "amount": 8173.59, "signup_date": "2023-02-01", "status": "Overdue", "unit": "cm", "rate": 0.855, "email": "cihoeh@example.com"}], "dirty_cols": ["state", "is_active", "amount", "signup_date", "status", "unit", "rate", "email", "unnamed"], "clean_cols": ["state", "is_active", "amount", "signup_date", "status", "unit", "rate", "email"], "plan": {"dataset_summary": "34 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OH": "Ohio", "DC": "District of Columbia", "WV": "West Virginia", "west virginia": "West Virginia", "MAINE": "Maine", "wv": "West Virginia", "District Of Columbia": "District of Columbia", "NE": "Nebraska", "nebraska": "Nebraska", "Me": "Maine", "Ne": "Nebraska", "district of columbia": "District of Columbia", "WETS VIRGINIA": "West Virginia", "WEST VIRGINIA": "West Virginia", "ME": "Maine", "ohio": "Ohio"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Refnded": "Refunded", "paid": "Paid", "late": "Overdue", "unpaid": "Pending", "PAUD": "Paid", "PAID": "Paid", "overdue": "Overdue", "LATE": "Overdue", "pending": "Pending", "due": "Pending", "unpaaid": "Pending", "PENDING": "Pending"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"inch": "in", "Cm": "cm", "INCHES": "in", "\"": "in", "centimeters": "cm", "centimeter": "cm", "Innch": "in", "cenntimeters": "cm", "CM": "cm", "inches": "in"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"city": "Draguignan", "amount": "6.786,37", "currency": "guana dollar", "rate": "27.9%", "status": "overdue", "name": "Karl Brandt", "col_x": ""}, {"city": "Kalisz  ", "amount": "$4,644.75", "currency": "Guyana Dollar  ", "rate": "31.4%", "status": "overdue", "name": "Grace  Ali", "col_x": ""}, {"city": "NOVA ANDRADINA   ", "amount": "7.520,42", "currency": "Pula", "rate": "55.4%", "status": "Overdue", "name": "Carol Smith", "col_x": ""}, {"city": "Nova Andradina", "amount": "--", "currency": "Guyana Dollar", "rate": "39.8%", "status": "Refunded", "name": "   Bob Wong", "col_x": ""}, {"city": "draguignan", "amount": "$1,341.92", "currency": "Taka", "rate": "99.2%", "status": "refund", "name": "Judy Brandt", "col_x": ""}, {"city": "Draguignan", "amount": "$7,704.42", "currency": " GYD ", "rate": "79.7%", "status": "paid  ", "name": "Ivan Reyes", "col_x": ""}, {"city": "KALISZ", "amount": "2.905,41", "currency": "bwp", "rate": "10.6%", "status": "Refunded", "name": "   Mona Lee", "col_x": ""}, {"city": "Kalisz", "amount": "7.958,06", "currency": " Pula", "rate": "76.0%", "status": "refunded", "name": "Bob Lee ", "col_x": ""}, {"city": "nova  andradina", "amount": "4.369,69", "currency": "BWP", "rate": "20.3%", "status": "PAID", "name": "Eve Moore", "col_x": ""}, {"city": "DRAGUIGNAN", "amount": "2.624,09", "currency": "BWP", "rate": "84.2%", "status": "REFUND", "name": "Mona Reyes   ", "col_x": ""}, {"city": "draguignan", "amount": "2.020,41", "currency": "bdt", "rate": "95.4%", "status": "overdqe", "name": "Lena Diaz  ", "col_x": ""}, {"city": "nova andradina", "amount": "3.811,43", "currency": "Pula", "rate": "23.6%", "status": "PAID", "name": "David Park", "col_x": ""}, {"city": "Draguignan", "amount": "810,80", "currency": "TAKA", "rate": "15.4%", "status": "refund", "name": "Judy Petrov   ", "col_x": ""}, {"city": "draguignan", "amount": "6.167,92", "currency": "Pula", "rate": "83.5%", "status": "overdue", "name": "Judy Brandt", "col_x": ""}, {"city": "kailsz", "amount": "7.418,94", "currency": "Pula", "rate": "4.3%", "status": "REFUNDED", "name": "Eve Moore", "col_x": ""}, {"city": "DRAGUIGNAN", "amount": "2.158,94", "currency": "Bwp   ", "rate": "19.0%", "status": "overdue", "name": "Omar  Khan", "col_x": ""}, {"city": "  Draguignan", "amount": "None", "currency": "gyd", "rate": "43.3%", "status": "paid", "name": "Lena  Adams", "col_x": ""}, {"city": "KALISZ", "amount": "6.248,57", "currency": "GYD", "rate": "75.3%", "status": "paid", "name": "Frank  Moore", "col_x": ""}, {"city": " Kalisz ", "amount": "3.937,51", "currency": "Tqka", "rate": "17.7%", "status": "refund", "name": "Karl Fischer  ", "col_x": ""}, {"city": "   Nova Andrdaina", "amount": "-", "currency": "Guyana  Dollar", "rate": "91.0%", "status": "refunded", "name": "  Sara Petrov", "col_x": ""}, {"city": "KAALISZ", "amount": "#N/A", "currency": "guyana dollar ", "rate": "29.3%", "status": "Paid", "name": " Eve Novak", "col_x": ""}, {"city": "Kalisz", "amount": "2.994,06", "currency": "pula", "rate": "34.5%", "status": "OVERDUE", "name": "Bob Lee", "col_x": ""}, {"city": "nova andradina", "amount": "1.241,71", "currency": "TAKA  ", "rate": "42.2%", "status": "late", "name": " Ivan Wong", "col_x": ""}, {"city": " Draguignan ", "amount": "$2,091.17", "currency": "BDT", "rate": "3.8%", "status": "Paid", "name": "Heidi Cruz", "col_x": ""}, {"city": "  draguignan", "amount": "6.055,61", "currency": "GYD ", "rate": "54.1%", "status": "overdue", "name": "Alice  Diaz", "col_x": ""}, {"city": "draguignan", "amount": "1.763,35", "currency": "Guyana Dollar", "rate": "69.4%", "status": "REFUNDED", "name": "Lena Wong ", "col_x": ""}, {"city": "nxva andradina", "amount": "4.756,66", "currency": "BWP", "rate": "38.7%", "status": "Paid", "name": "Eve Cruz", "col_x": ""}, {"city": "Draguignan", "amount": "6.710,47", "currency": "BWP", "rate": "60.1%", "status": "paid", "name": " Frank Johnson", "col_x": ""}, {"city": "  kalpsz", "amount": "?", "currency": "BDT ", "rate": "82.1%", "status": " Refunded", "name": "  Omar Smith", "col_x": ""}, {"city": "  Draguignan", "amount": "6.322,55", "currency": "taak", "rate": "50.5%", "status": "paid", "name": "Eve Reyes", "col_x": ""}, {"city": "KALISZ", "amount": "na", "currency": "TAKA ", "rate": "3.9%", "status": "overdue   ", "name": "Lena  Fischer", "col_x": ""}, {"city": "  Nova Andradina", "amount": "8.139,49", "currency": "BDT", "rate": "97.0%", "status": "RRFUND", "name": "Judy Diaz", "col_x": ""}, {"city": "kallisz", "amount": "-", "currency": "Bdt", "rate": "25.5%", "status": "  paid", "name": "Karl Cruz ", "col_x": ""}, {"city": "Kalisz", "amount": "$3,191.88", "currency": "  BWP", "rate": "12.6%", "status": "LATE", "name": " Judy Lee", "col_x": ""}, {"city": "Draguignan", "amount": "861,03", "currency": "   BDT", "rate": "59.4%", "status": "paid", "name": "Priya Moore", "col_x": ""}, {"city": "Draguignan", "amount": "4.243,59", "currency": "pulla", "rate": "3.7%", "status": "Paid", "name": " Carol Park", "col_x": ""}, {"city": "KALISZ", "amount": "$5,470.97", "currency": "Pula   ", "rate": "43.3%", "status": "refund", "name": "Judy Fischer", "col_x": ""}, {"city": "KALISZ", "amount": "$459.30", "currency": " pula ", "rate": "5.9%", "status": "PAID", "name": "   Lena Fischer", "col_x": ""}, {"city": "   Draguignan", "amount": "$5,177.47", "currency": "GYD", "rate": "31.3%", "status": "refunded", "name": "Mona  Fischer", "col_x": ""}, {"city": "Nova Andradina", "amount": "0", "currency": "BWP", "rate": "86.3%", "status": "  late", "name": "Mona Reyes", "col_x": ""}, {"city": "NOVA ANDRADINA", "amount": "1.119,78", "currency": "Guyana Dollmr", "rate": "39.0%", "status": "Paid", "name": "Judy Moore", "col_x": ""}, {"city": "Nova Andradina", "amount": "--", "currency": "Guyana Dollar", "rate": "39.8%", "status": "Refunded", "name": "   Bob Wong", "col_x": ""}, {"city": "DRAGUIGNAN", "amount": "2.158,94", "currency": "Bwp   ", "rate": "19.0%", "status": "overdue", "name": "Omar  Khan", "col_x": ""}], "clean": [{"city": "Draguignan", "amount": 6786.37, "currency": "GYD", "rate": 0.27899999999999997, "status": "Overdue", "name": "Karl Brandt"}, {"city": "Kalisz", "amount": 4644.75, "currency": "GYD", "rate": 0.314, "status": "Overdue", "name": "Grace Ali"}, {"city": "Nova Andradina", "amount": 7520.42, "currency": "BWP", "rate": 0.5539999999999999, "status": "Overdue", "name": "Carol Smith"}, {"city": "Nova Andradina", "amount": NaN, "currency": "GYD", "rate": 0.39799999999999996, "status": "Refunded", "name": "Bob Wong"}, {"city": "Draguignan", "amount": 1341.92, "currency": "BDT", "rate": 0.992, "status": "Refunded", "name": "Judy Brandt"}, {"city": "Draguignan", "amount": 7704.42, "currency": "GYD", "rate": 0.797, "status": "Paid", "name": "Ivan Reyes"}, {"city": "Kalisz", "amount": 2905.41, "currency": "BWP", "rate": 0.106, "status": "Refunded", "name": "Mona Lee"}, {"city": "Kalisz", "amount": 7958.06, "currency": "BWP", "rate": 0.76, "status": "Refunded", "name": "Bob Lee"}, {"city": "Nova Andradina", "amount": 4369.69, "currency": "BWP", "rate": 0.203, "status": "Paid", "name": "Eve Moore"}, {"city": "Draguignan", "amount": 2624.09, "currency": "BWP", "rate": 0.8420000000000001, "status": "Refunded", "name": "Mona Reyes"}, {"city": "Draguignan", "amount": 2020.41, "currency": "BDT", "rate": 0.9540000000000001, "status": "Overdue", "name": "Lena Diaz"}, {"city": "Nova Andradina", "amount": 3811.43, "currency": "BWP", "rate": 0.23600000000000002, "status": "Paid", "name": "David Park"}, {"city": "Draguignan", "amount": 810.8, "currency": "BDT", "rate": 0.154, "status": "Refunded", "name": "Judy Petrov"}, {"city": "Draguignan", "amount": 6167.92, "currency": "BWP", "rate": 0.835, "status": "Overdue", "name": "Judy Brandt"}, {"city": "Kalisz", "amount": 7418.94, "currency": "BWP", "rate": 0.043, "status": "Refunded", "name": "Eve Moore"}, {"city": "Draguignan", "amount": 2158.94, "currency": "BWP", "rate": 0.19, "status": "Overdue", "name": "Omar Khan"}, {"city": "Draguignan", "amount": NaN, "currency": "GYD", "rate": 0.433, "status": "Paid", "name": "Lena Adams"}, {"city": "Kalisz", "amount": 6248.57, "currency": "GYD", "rate": 0.753, "status": "Paid", "name": "Frank Moore"}, {"city": "Kalisz", "amount": 3937.51, "currency": "BDT", "rate": 0.177, "status": "Refunded", "name": "Karl Fischer"}, {"city": "Nova Andradina", "amount": NaN, "currency": "GYD", "rate": 0.91, "status": "Refunded", "name": "Sara Petrov"}, {"city": "Kalisz", "amount": NaN, "currency": "GYD", "rate": 0.293, "status": "Paid", "name": "Eve Novak"}, {"city": "Kalisz", "amount": 2994.06, "currency": "BWP", "rate": 0.345, "status": "Overdue", "name": "Bob Lee"}, {"city": "Nova Andradina", "amount": 1241.71, "currency": "BDT", "rate": 0.42200000000000004, "status": "Overdue", "name": "Ivan Wong"}, {"city": "Draguignan", "amount": 2091.17, "currency": "BDT", "rate": 0.038, "status": "Paid", "name": "Heidi Cruz"}, {"city": "Draguignan", "amount": 6055.61, "currency": "GYD", "rate": 0.541, "status": "Overdue", "name": "Alice Diaz"}, {"city": "Draguignan", "amount": 1763.35, "currency": "GYD", "rate": 0.6940000000000001, "status": "Refunded", "name": "Lena Wong"}, {"city": "Nova Andradina", "amount": 4756.66, "currency": "BWP", "rate": 0.387, "status": "Paid", "name": "Eve Cruz"}, {"city": "Draguignan", "amount": 6710.47, "currency": "BWP", "rate": 0.601, "status": "Paid", "name": "Frank Johnson"}, {"city": "Kalisz", "amount": NaN, "currency": "BDT", "rate": 0.821, "status": "Refunded", "name": "Omar Smith"}, {"city": "Draguignan", "amount": 6322.55, "currency": "BDT", "rate": 0.505, "status": "Paid", "name": "Eve Reyes"}, {"city": "Kalisz", "amount": NaN, "currency": "BDT", "rate": 0.039, "status": "Overdue", "name": "Lena Fischer"}, {"city": "Nova Andradina", "amount": 8139.49, "currency": "BDT", "rate": 0.97, "status": "Refunded", "name": "Judy Diaz"}, {"city": "Kalisz", "amount": NaN, "currency": "BDT", "rate": 0.255, "status": "Paid", "name": "Karl Cruz"}, {"city": "Kalisz", "amount": 3191.88, "currency": "BWP", "rate": 0.126, "status": "Overdue", "name": "Judy Lee"}, {"city": "Draguignan", "amount": 861.03, "currency": "BDT", "rate": 0.594, "status": "Paid", "name": "Priya Moore"}, {"city": "Draguignan", "amount": 4243.59, "currency": "BWP", "rate": 0.037000000000000005, "status": "Paid", "name": "Carol Park"}, {"city": "Kalisz", "amount": 5470.97, "currency": "BWP", "rate": 0.433, "status": "Refunded", "name": "Judy Fischer"}, {"city": "Kalisz", "amount": 459.3, "currency": "BWP", "rate": 0.059000000000000004, "status": "Paid", "name": "Lena Fischer"}, {"city": "Draguignan", "amount": 5177.47, "currency": "GYD", "rate": 0.313, "status": "Refunded", "name": "Mona Fischer"}, {"city": "Nova Andradina", "amount": 0.0, "currency": "BWP", "rate": 0.863, "status": "Overdue", "name": "Mona Reyes"}, {"city": "Nova Andradina", "amount": 1119.78, "currency": "GYD", "rate": 0.39, "status": "Paid", "name": "Judy Moore"}], "dirty_cols": ["city", "amount", "currency", "rate", "status", "name", "col_x"], "clean_cols": ["city", "amount", "currency", "rate", "status", "name"], "plan": {"dataset_summary": "43 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NOVA ANDRADINA": "Nova Andradina", "draguignan": "Draguignan", "KALISZ": "Kalisz", "nova andradina": "Nova Andradina", "DRAGUIGNAN": "Draguignan", "kailsz": "Kalisz", "Nova Andrdaina": "Nova Andradina", "KAALISZ": "Kalisz", "nxva andradina": "Nova Andradina", "kalpsz": "Kalisz", "kallisz": "Kalisz"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"guana dollar": "GYD", "Guyana Dollar": "GYD", "Pula": "BWP", "Taka": "BDT", "bwp": "BWP", "bdt": "BDT", "TAKA": "BDT", "Bwp": "BWP", "gyd": "GYD", "Tqka": "BDT", "guyana dollar": "GYD", "pula": "BWP", "taak": "BDT", "Bdt": "BDT", "pulla": "BWP", "Guyana Dollmr": "GYD"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"overdue": "Overdue", "refund": "Refunded", "paid": "Paid", "refunded": "Refunded", "PAID": "Paid", "REFUND": "Refunded", "overdqe": "Overdue", "REFUNDED": "Refunded", "OVERDUE": "Overdue", "late": "Overdue", "RRFUND": "Refunded", "LATE": "Overdue"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"phone": "504-879-1726", "amount": "$97.04", "currency": " taka ", "name": "Karl Novak"}, {"phone": "(515)492-6623", "amount": "$5,732.60", "currency": " Taka ", "name": "Eve Moore   "}, {"phone": "515-843-1229", "amount": "8.214,17", "currency": "Aed", "name": "Sara Khan"}, {"phone": "(509)648-6909", "amount": "3.707,38", "currency": "BDT", "name": "  Carol Smith"}, {"phone": "5880372779", "amount": "$236.54", "currency": "Kwnza", "name": " Mona Park"}, {"phone": "503-680-0473", "amount": "4.934,19", "currency": "   HNL", "name": "Sara Lee"}, {"phone": "553-246-1702", "amount": "8.311,57", "currency": "aed", "name": "David  Khan"}, {"phone": "512.592.1577", "amount": "570,80", "currency": "   AED", "name": "Alice  Wong"}, {"phone": "548-340-3500", "amount": "$6,185.39", "currency": "aed", "name": "Grace Lee  "}, {"phone": "(507)424-4610", "amount": "$7,652.35", "currency": "AED   ", "name": "Omar Lee   "}, {"phone": "542-761-6751", "amount": "--", "currency": "BDT", "name": "Priya Petrov"}, {"phone": "526-074-5755", "amount": "0", "currency": "AED  ", "name": "Eve Johnson"}, {"phone": "568.105.4888", "amount": "None", "currency": "hnl", "name": "Karl Cruz"}, {"phone": "5924544444", "amount": "$6,829.29", "currency": "BDT ", "name": "Frank Reyes"}, {"phone": "(547)571-0119", "amount": "$3,827.00", "currency": "uae dirham", "name": "Judy Moore"}, {"phone": "5000188148", "amount": "$871.51", "currency": "aoa", "name": "Grace Brandt"}, {"phone": "5897969003", "amount": "3.192,18", "currency": "AOA", "name": "Judy Ali   "}, {"phone": "(533)388-3967", "amount": "$2,055.63", "currency": "taka", "name": "  Ivan Park"}, {"phone": "573.315.0350", "amount": "$7,116.22", "currency": "  uae dirham", "name": "Carol Khan"}], "clean": [{"phone": "(504) 879-1726", "amount": 97.04, "currency": "BDT", "name": "Karl Novak"}, {"phone": "(515) 492-6623", "amount": 5732.6, "currency": "BDT", "name": "Eve Moore"}, {"phone": "(515) 843-1229", "amount": 8214.17, "currency": "AED", "name": "Sara Khan"}, {"phone": "(509) 648-6909", "amount": 3707.38, "currency": "BDT", "name": "Carol Smith"}, {"phone": "(588) 037-2779", "amount": 236.54, "currency": "AOA", "name": "Mona Park"}, {"phone": "(503) 680-0473", "amount": 4934.19, "currency": "HNL", "name": "Sara Lee"}, {"phone": "(553) 246-1702", "amount": 8311.57, "currency": "AED", "name": "David Khan"}, {"phone": "(512) 592-1577", "amount": 570.8, "currency": "AED", "name": "Alice Wong"}, {"phone": "(548) 340-3500", "amount": 6185.39, "currency": "AED", "name": "Grace Lee"}, {"phone": "(507) 424-4610", "amount": 7652.35, "currency": "AED", "name": "Omar Lee"}, {"phone": "(542) 761-6751", "amount": NaN, "currency": "BDT", "name": "Priya Petrov"}, {"phone": "(526) 074-5755", "amount": 0.0, "currency": "AED", "name": "Eve Johnson"}, {"phone": "(568) 105-4888", "amount": NaN, "currency": "HNL", "name": "Karl Cruz"}, {"phone": "(592) 454-4444", "amount": 6829.29, "currency": "BDT", "name": "Frank Reyes"}, {"phone": "(547) 571-0119", "amount": 3827.0, "currency": "AED", "name": "Judy Moore"}, {"phone": "(500) 018-8148", "amount": 871.51, "currency": "AOA", "name": "Grace Brandt"}, {"phone": "(589) 796-9003", "amount": 3192.18, "currency": "AOA", "name": "Judy Ali"}, {"phone": "(533) 388-3967", "amount": 2055.63, "currency": "BDT", "name": "Ivan Park"}, {"phone": "(573) 315-0350", "amount": 7116.22, "currency": "AED", "name": "Carol Khan"}], "dirty_cols": ["phone", "amount", "currency", "name"], "clean_cols": ["phone", "amount", "currency", "name"], "plan": {"dataset_summary": "19 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"taka": "BDT", "Taka": "BDT", "Aed": "AED", "Kwnza": "AOA", "aed": "AED", "hnl": "HNL", "uae dirham": "AED", "aoa": "AOA"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"is_active": "0", "email": "APBDP@TEST.ORG", "department": "Ops", "country": "Vc", "unit": "kg", "notes2": ""}, {"is_active": "false", "email": "  pifb@example.com", "department": "ops", "country": "AUSTRIA", "unit": "in", "notes2": ""}, {"is_active": "T", "email": " LOGLJD@CORP.IO ", "department": "Operatios", "country": "TKM", "unit": "   kilometer", "notes2": ""}, {"is_active": "TRUE", "email": "   LOJNN@CORP.IO", "department": "OPERATIONS", "country": "   vct", "unit": "Kg", "notes2": ""}, {"is_active": "No", "email": "iogec@mail.com", "department": "Sales", "country": "Nc", "unit": "inches  ", "notes2": ""}, {"is_active": "1", "email": "cdig@test.org", "department": "sales", "country": "   Vc", "unit": "km", "notes2": ""}, {"is_active": "F", "email": "OJIOAML@CORP.IO", "department": "operations", "country": "Curaçao", "unit": "\"", "notes2": ""}, {"is_active": "true", "email": "  fgmec@mail.com", "department": "OPERATIONS", "country": "tm", "unit": " KG", "notes2": ""}, {"is_active": "true", "email": " FPKCL@MAIL.COM ", "department": "sales", "country": "Curaçao", "unit": "KG", "notes2": ""}, {"is_active": "FALSE", "email": "JHDN@TEST.ORG", "department": "Ops", "country": "cw", "unit": "kilometers", "notes2": ""}, {"is_active": "Y", "email": "GMLDDEJ@MAIL.COM", "department": "Biz Dev", "country": "republic of austria", "unit": "kilogram", "notes2": ""}, {"is_active": "0", "email": "fbod@mail.com", "department": "operations", "country": "vc", "unit": "kilogram", "notes2": ""}, {"is_active": "N", "email": " OKIGL@CORP.IO", "department": "Biz Dev", "country": "nc", "unit": "km", "notes2": ""}, {"is_active": "true", "email": "PKDNB@TEST.ORG", "department": "oprations", "country": "Republic of Austria", "unit": " inches", "notes2": ""}, {"is_active": "TRUE", "email": "dfmmd@corp.io  ", "department": "SALLES", "country": " tkm ", "unit": "km", "notes2": ""}, {"is_active": "0", "email": "ilmlgf@example.com", "department": "Sales", "country": "Curaçao", "unit": "  kg", "notes2": ""}, {"is_active": "Yes", "email": "  annan@mail.com", "department": "Sales", "country": "cuw", "unit": "kilograms  ", "notes2": ""}, {"is_active": "true", "email": "PFONKGP@TEST.ORG", "department": " Operations", "country": "TKM", "unit": "\"", "notes2": ""}, {"is_active": "FALSE", "email": "DMDBBF@MAIL.COM  ", "department": "operations", "country": "saint vincent and the grenadines", "unit": "INCH ", "notes2": ""}, {"is_active": "N", "email": "   ABCLO@TEST.ORG", "department": "OPERATIONS", "country": "NC", "unit": "inch  ", "notes2": ""}, {"is_active": "0", "email": "EBAOCGC@CORP.IO   ", "department": "sales", "country": "Turkmenistan", "unit": "kilometer", "notes2": ""}, {"is_active": "Yes", "email": "LCDAI@CORP.IO", "department": "Operatins", "country": "Turkmentstan", "unit": "Kg", "notes2": ""}, {"is_active": "true", "email": "   pnoha@mail.com", "department": "Ops", "country": " Austria ", "unit": "kilogram", "notes2": ""}, {"is_active": "No", "email": " liife@example.com", "department": "Biz Dev", "country": "Turkmenistan", "unit": "kilometr", "notes2": ""}, {"is_active": "TRUE", "email": "nafchgn@mail.com", "department": "Sales", "country": "  nc", "unit": " inches", "notes2": ""}, {"is_active": "false", "email": "lcnde@mail.com", "department": "Ops", "country": "CURAÇAO", "unit": "kg", "notes2": ""}, {"is_active": "false", "email": "PMOIHA@TEST.ORG", "department": "   Biz Dev", "country": " curaçao ", "unit": "KG", "notes2": ""}, {"is_active": "TRUE", "email": "   BKLIDFM@EXAMPLE.COM", "department": "Operations", "country": "Curaçao", "unit": "kilograms", "notes2": ""}, {"is_active": "TRUE", "email": " gpgfhd@test.org", "department": "bhz dev", "country": "CW", "unit": "km", "notes2": ""}, {"is_active": "FALSE", "email": "MHJKG@EXAMPLE.COM", "department": "Biz  Dev", "country": "NCL", "unit": "kg", "notes2": ""}, {"is_active": "Y", "email": "  egjgim@corp.io", "department": "Ops", "country": "ncl  ", "unit": "km", "notes2": ""}, {"is_active": "Y", "email": "ggodh@example.com", "department": "sales", "country": " austria", "unit": "kilometers", "notes2": ""}, {"is_active": "true", "email": "NMKMB@CORP.IO", "department": "Operations", "country": "TURKMENISTAN", "unit": "\"", "notes2": ""}, {"is_active": "FALSE", "email": "OBBNACG@TEST.ORG", "department": "ops ", "country": "VCT", "unit": "KM", "notes2": ""}, {"is_active": "0", "email": "JKDHJMI@TEST.ORG", "department": "Biz Dev", "country": " at ", "unit": "kg", "notes2": ""}, {"is_active": "Yes", "email": "ODKB@TEST.ORG", "department": "Operations", "country": "CUW", "unit": "\"   ", "notes2": ""}, {"is_active": "T", "email": "dolkpf@mail.com", "department": "biz dev", "country": "AT", "unit": "  kilogram", "notes2": ""}, {"is_active": "true", "email": "NNHHII@CORP.IO", "department": "BizDev", "country": "   AT", "unit": "kilometer", "notes2": ""}, {"is_active": "N", "email": " GICOP@MAIL.COM", "department": "Ops", "country": "cw ", "unit": "inches", "notes2": ""}, {"is_active": "N", "email": " OKIGL@CORP.IO", "department": "Biz Dev", "country": "nc", "unit": "km", "notes2": ""}], "clean": [{"is_active": false, "email": "apbdp@test.org", "department": "Operations", "country": "Saint Vincent and the Grenadines", "unit": "kg"}, {"is_active": false, "email": "pifb@example.com", "department": "Operations", "country": "Austria", "unit": "in"}, {"is_active": true, "email": "logljd@corp.io", "department": "Operations", "country": "Turkmenistan", "unit": "km"}, {"is_active": true, "email": "lojnn@corp.io", "department": "Operations", "country": "Saint Vincent and the Grenadines", "unit": "kg"}, {"is_active": false, "email": "iogec@mail.com", "department": "Sales", "country": "New Caledonia", "unit": "in"}, {"is_active": true, "email": "cdig@test.org", "department": "Sales", "country": "Saint Vincent and the Grenadines", "unit": "km"}, {"is_active": false, "email": "ojioaml@corp.io", "department": "Operations", "country": "Curaçao", "unit": "in"}, {"is_active": true, "email": "fgmec@mail.com", "department": "Operations", "country": "Turkmenistan", "unit": "kg"}, {"is_active": true, "email": "fpkcl@mail.com", "department": "Sales", "country": "Curaçao", "unit": "kg"}, {"is_active": false, "email": "jhdn@test.org", "department": "Operations", "country": "Curaçao", "unit": "km"}, {"is_active": true, "email": "gmlddej@mail.com", "department": "Sales", "country": "Austria", "unit": "kg"}, {"is_active": false, "email": "fbod@mail.com", "department": "Operations", "country": "Saint Vincent and the Grenadines", "unit": "kg"}, {"is_active": false, "email": "okigl@corp.io", "department": "Sales", "country": "New Caledonia", "unit": "km"}, {"is_active": true, "email": "pkdnb@test.org", "department": "Operations", "country": "Austria", "unit": "in"}, {"is_active": true, "email": "dfmmd@corp.io", "department": "Sales", "country": "Turkmenistan", "unit": "km"}, {"is_active": false, "email": "ilmlgf@example.com", "department": "Sales", "country": "Curaçao", "unit": "kg"}, {"is_active": true, "email": "annan@mail.com", "department": "Sales", "country": "Curaçao", "unit": "kg"}, {"is_active": true, "email": "pfonkgp@test.org", "department": "Operations", "country": "Turkmenistan", "unit": "in"}, {"is_active": false, "email": "dmdbbf@mail.com", "department": "Operations", "country": "Saint Vincent and the Grenadines", "unit": "in"}, {"is_active": false, "email": "abclo@test.org", "department": "Operations", "country": "New Caledonia", "unit": "in"}, {"is_active": false, "email": "ebaocgc@corp.io", "department": "Sales", "country": "Turkmenistan", "unit": "km"}, {"is_active": true, "email": "lcdai@corp.io", "department": "Operations", "country": "Turkmenistan", "unit": "kg"}, {"is_active": true, "email": "pnoha@mail.com", "department": "Operations", "country": "Austria", "unit": "kg"}, {"is_active": false, "email": "liife@example.com", "department": "Sales", "country": "Turkmenistan", "unit": "km"}, {"is_active": true, "email": "nafchgn@mail.com", "department": "Sales", "country": "New Caledonia", "unit": "in"}, {"is_active": false, "email": "lcnde@mail.com", "department": "Operations", "country": "Curaçao", "unit": "kg"}, {"is_active": false, "email": "pmoiha@test.org", "department": "Sales", "country": "Curaçao", "unit": "kg"}, {"is_active": true, "email": "bklidfm@example.com", "department": "Operations", "country": "Curaçao", "unit": "kg"}, {"is_active": true, "email": "gpgfhd@test.org", "department": "Sales", "country": "Curaçao", "unit": "km"}, {"is_active": false, "email": "mhjkg@example.com", "department": "Sales", "country": "New Caledonia", "unit": "kg"}, {"is_active": true, "email": "egjgim@corp.io", "department": "Operations", "country": "New Caledonia", "unit": "km"}, {"is_active": true, "email": "ggodh@example.com", "department": "Sales", "country": "Austria", "unit": "km"}, {"is_active": true, "email": "nmkmb@corp.io", "department": "Operations", "country": "Turkmenistan", "unit": "in"}, {"is_active": false, "email": "obbnacg@test.org", "department": "Operations", "country": "Saint Vincent and the Grenadines", "unit": "km"}, {"is_active": false, "email": "jkdhjmi@test.org", "department": "Sales", "country": "Austria", "unit": "kg"}, {"is_active": true, "email": "odkb@test.org", "department": "Operations", "country": "Curaçao", "unit": "in"}, {"is_active": true, "email": "dolkpf@mail.com", "department": "Sales", "country": "Austria", "unit": "kg"}, {"is_active": true, "email": "nnhhii@corp.io", "department": "Sales", "country": "Austria", "unit": "km"}, {"is_active": false, "email": "gicop@mail.com", "department": "Operations", "country": "Curaçao", "unit": "in"}], "dirty_cols": ["is_active", "email", "department", "country", "unit", "notes2"], "clean_cols": ["is_active", "email", "department", "country", "unit"], "plan": {"dataset_summary": "40 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ops": "Operations", "ops": "Operations", "Operatios": "Operations", "OPERATIONS": "Operations", "sales": "Sales", "operations": "Operations", "Biz Dev": "Sales", "oprations": "Operations", "SALLES": "Sales", "Operatins": "Operations", "bhz dev": "Sales", "biz dev": "Sales", "BizDev": "Sales"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Vc": "Saint Vincent and the Grenadines", "AUSTRIA": "Austria", "TKM": "Turkmenistan", "vct": "Saint Vincent and the Grenadines", "Nc": "New Caledonia", "tm": "Turkmenistan", "cw": "Curaçao", "republic of austria": "Austria", "vc": "Saint Vincent and the Grenadines", "nc": "New Caledonia", "Republic of Austria": "Austria", "tkm": "Turkmenistan", "cuw": "Curaçao", "saint vincent and the grenadines": "Saint Vincent and the Grenadines", "NC": "New Caledonia", "Turkmentstan": "Turkmenistan", "CURAÇAO": "Curaçao", "curaçao": "Curaçao", "CW": "Curaçao", "NCL": "New Caledonia", "ncl": "New Caledonia", "austria": "Austria", "TURKMENISTAN": "Turkmenistan", "VCT": "Saint Vincent and the Grenadines", "at": "Austria", "CUW": "Curaçao", "AT": "Austria"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilometer": "km", "Kg": "kg", "inches": "in", "\"": "in", "KG": "kg", "kilometers": "km", "kilogram": "kg", "kilograms": "kg", "INCH": "in", "inch": "in", "kilometr": "km", "KM": "km"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "81.7%", "company": "Acme  Inc", "state": "MA", "extra": ""}, {"rate": "13.6%", "company": "Acme  Inc", "state": "massachusetts", "extra": ""}, {"rate": "75.1%", "company": " Soylent Corp", "state": "Hi", "extra": ""}, {"rate": "87.6%", "company": "Hooli", "state": "   Ma", "extra": ""}, {"rate": "4.7%", "company": "Acme Inc", "state": "Maryland", "extra": ""}, {"rate": "28.2%", "company": "Umbrella", "state": "HI", "extra": ""}, {"rate": "72.5%", "company": " Cyberdyne", "state": " Maryland", "extra": ""}, {"rate": "67.1%", "company": "  Globex", "state": "hi", "extra": ""}, {"rate": "53.8%", "company": "Initech ", "state": " MD ", "extra": ""}, {"rate": "61.9%", "company": "Umbrella", "state": "Md", "extra": ""}, {"rate": "60.5%", "company": "Hooli", "state": "  MA", "extra": ""}, {"rate": "22.9%", "company": "Soylent Corp", "state": "Massachusetts", "extra": ""}, {"rate": "41.3%", "company": "Globex", "state": "MD", "extra": ""}, {"rate": "65.1%", "company": "Wonka Co", "state": "HAWAII", "extra": ""}, {"rate": "73.9%", "company": "Wonka Co", "state": "MASSACHUSETTS", "extra": ""}, {"rate": "68.8%", "company": " Vehement ", "state": "  Maryland", "extra": ""}, {"rate": "65.3%", "company": "Soylent Corp", "state": "MA", "extra": ""}, {"rate": "53.8%", "company": "Initech ", "state": " MD ", "extra": ""}], "clean": [{"rate": 0.8170000000000001, "company": "Acme Inc", "state": "Massachusetts"}, {"rate": 0.136, "company": "Acme Inc", "state": "Massachusetts"}, {"rate": 0.7509999999999999, "company": "Soylent Corp", "state": "Hawaii"}, {"rate": 0.8759999999999999, "company": "Hooli", "state": "Massachusetts"}, {"rate": 0.047, "company": "Acme Inc", "state": "Maryland"}, {"rate": 0.282, "company": "Umbrella", "state": "Hawaii"}, {"rate": 0.725, "company": "Cyberdyne", "state": "Maryland"}, {"rate": 0.6709999999999999, "company": "Globex", "state": "Hawaii"}, {"rate": 0.5379999999999999, "company": "Initech", "state": "Maryland"}, {"rate": 0.619, "company": "Umbrella", "state": "Maryland"}, {"rate": 0.605, "company": "Hooli", "state": "Massachusetts"}, {"rate": 0.22899999999999998, "company": "Soylent Corp", "state": "Massachusetts"}, {"rate": 0.413, "company": "Globex", "state": "Maryland"}, {"rate": 0.6509999999999999, "company": "Wonka Co", "state": "Hawaii"}, {"rate": 0.7390000000000001, "company": "Wonka Co", "state": "Massachusetts"}, {"rate": 0.688, "company": "Vehement", "state": "Maryland"}, {"rate": 0.653, "company": "Soylent Corp", "state": "Massachusetts"}], "dirty_cols": ["rate", "company", "state", "extra"], "clean_cols": ["rate", "company", "state"], "plan": {"dataset_summary": "18 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MA": "Massachusetts", "massachusetts": "Massachusetts", "Hi": "Hawaii", "Ma": "Massachusetts", "HI": "Hawaii", "hi": "Hawaii", "MD": "Maryland", "Md": "Maryland", "HAWAII": "Hawaii", "MASSACHUSETTS": "Massachusetts"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "puerto cabello", "industry": "finance", "unit": "POUNDS   ", "phone": "5023973319", "email": "hofopdn@corp.io"}, {"city": "Atotonilco el Alto", "industry": "  Finance", "unit": " KM ", "phone": "550.353.3120", "email": "PNOJBBC@MAIL.COM   "}, {"city": " alanta ", "industry": "Telecommunications", "unit": " litre ", "phone": "532.838.9915", "email": "  nbgeh@test.org"}, {"city": "Atotonilco el Alto", "industry": " Telecmo ", "unit": "lbs", "phone": "592.375.0814", "email": "HBGCC@TEST.ORG"}, {"city": "Pekin", "industry": "Telecom", "unit": "in   ", "phone": "5556976191", "email": "jjbci@mail.com"}, {"city": "Atlanta", "industry": "telco", "unit": "km", "phone": "5542363892", "email": " CIAK@TEST.ORG "}, {"city": "puerto cabello", "industry": " Finance ", "unit": "INCHES", "phone": "(504)443-3345", "email": "abjema@mail.com"}, {"city": "puerto cabello", "industry": "Telecommunications", "unit": "kilometers", "phone": "590.313.1654", "email": "NJGGEAB@CORP.IO   "}, {"city": "Pekin", "industry": " Banking", "unit": "Lb", "phone": "(556)950-3372", "email": "  ekgfel@example.com"}, {"city": "   Atlanta", "industry": "banking", "unit": "inch", "phone": "(559)839-0841", "email": "JMJJ@EXAMPLE.COM"}, {"city": "Atotonilco el Alto", "industry": " Banking ", "unit": "killometers", "phone": "525.964.5795", "email": "ILJLDA@CORP.IO"}, {"city": "atotonilco el alto", "industry": "   Telco", "unit": "kilometers", "phone": "506-580-3980", "email": "  kpjna@test.org"}, {"city": "Atotonilco el Alto", "industry": "  Banking", "unit": " L ", "phone": "544-103-4070", "email": "lpdgpg@test.org"}, {"city": "Puerto Cabello", "industry": "Finace", "unit": "LITER", "phone": "5849506257", "email": "DOIDGCH@MAIL.COM"}, {"city": "PUERTO  CABELLO", "industry": "TELCO", "unit": "\"", "phone": "585-635-9157", "email": "cpgkpc@example.com"}, {"city": "pekin", "industry": "finance", "unit": "Pound", "phone": "5302611406", "email": "phimel@example.com"}, {"city": "pekin", "industry": "telco ", "unit": "\"", "phone": "511.530.6721", "email": "idool@mail.com"}, {"city": "Puerto aCbello", "industry": "Fintehc   ", "unit": "\"", "phone": "548.886.5798", "email": "gjfcn@example.com"}, {"city": "pekin", "industry": "   teleom", "unit": "l", "phone": "(550)172-5068", "email": "GFJDKFO@CORP.IO"}, {"city": "Pekiin", "industry": "Tjlecom", "unit": " kilometer ", "phone": "561.620.6505", "email": " kijmh@test.org "}, {"city": "Atotonilco el Alto", "industry": "telecoommunications ", "unit": "\"", "phone": "5715309752", "email": "FCLLCPB@MAIL.COM"}, {"city": "ATOTONILCO EL ALTO ", "industry": "FinServ", "unit": "LITRE", "phone": "5207982943", "email": "bmieg@test.org"}, {"city": "Puero Cabello", "industry": "Telecom", "unit": "liyers", "phone": "5707490788", "email": "LKNMP@TEST.ORG  "}, {"city": "puerto cabello", "industry": "Comms", "unit": "Km", "phone": "5121980453", "email": "GLHL@MAIL.COM"}, {"city": "Puerto Cabello", "industry": "COMMS", "unit": "l", "phone": "509.972.0718", "email": "OGDCP@EXAMPLE.COM"}, {"city": "pekin", "industry": "Telecommunications", "unit": "pound", "phone": "(568)520-0951", "email": " ieddjol@corp.io"}, {"city": " atlanta ", "industry": "FnServ", "unit": "l", "phone": "562-530-1786", "email": "ipegb@example.com"}, {"city": "atlanta", "industry": "Comms", "unit": "LBS", "phone": "541.740.4177", "email": "OAMBO@MAIL.COM"}, {"city": "puerto cabello   ", "industry": "telecom", "unit": "lter", "phone": "5651163026", "email": "jamkc@corp.io"}, {"city": "atotonilco el alto   ", "industry": "   telco", "unit": "inch", "phone": "560-448-2898", "email": "BOMDJ@EXAMPLE.COM"}, {"city": "  Attotonilco el Alto", "industry": "FINANCE", "unit": " Lb ", "phone": "(590)427-7056", "email": "okcehnm@corp.io"}, {"city": "ATLANTA", "industry": "Banking   ", "unit": "poud  ", "phone": "(560)522-9733", "email": "  glpocih@mail.com"}, {"city": "Atotonilco el Alto  ", "industry": " FINANCE ", "unit": " l ", "phone": "561.440.4045", "email": "dajndf@corp.io"}, {"city": "Puerto Cabello", "industry": "   Comms", "unit": "kilometers", "phone": "513-015-5159", "email": " nffnhg@corp.io"}, {"city": "Puero Cabello", "industry": "Telecom", "unit": "liyers", "phone": "5707490788", "email": "LKNMP@TEST.ORG  "}], "clean": [{"city": "Puerto Cabello", "industry": "Financial Services", "unit": "lb", "phone": "(502) 397-3319", "email": "hofopdn@corp.io"}, {"city": "Atotonilco el Alto", "industry": "Financial Services", "unit": "km", "phone": "(550) 353-3120", "email": "pnojbbc@mail.com"}, {"city": "Atlanta", "industry": "Telecommunications", "unit": "L", "phone": "(532) 838-9915", "email": "nbgeh@test.org"}, {"city": "Atotonilco el Alto", "industry": "Telecommunications", "unit": "lb", "phone": "(592) 375-0814", "email": "hbgcc@test.org"}, {"city": "Pekin", "industry": "Telecommunications", "unit": "in", "phone": "(555) 697-6191", "email": "jjbci@mail.com"}, {"city": "Atlanta", "industry": "Telecommunications", "unit": "km", "phone": "(554) 236-3892", "email": "ciak@test.org"}, {"city": "Puerto Cabello", "industry": "Financial Services", "unit": "in", "phone": "(504) 443-3345", "email": "abjema@mail.com"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "km", "phone": "(590) 313-1654", "email": "njggeab@corp.io"}, {"city": "Pekin", "industry": "Financial Services", "unit": "lb", "phone": "(556) 950-3372", "email": "ekgfel@example.com"}, {"city": "Atlanta", "industry": "Financial Services", "unit": "in", "phone": "(559) 839-0841", "email": "jmjj@example.com"}, {"city": "Atotonilco el Alto", "industry": "Financial Services", "unit": "km", "phone": "(525) 964-5795", "email": "iljlda@corp.io"}, {"city": "Atotonilco el Alto", "industry": "Telecommunications", "unit": "km", "phone": "(506) 580-3980", "email": "kpjna@test.org"}, {"city": "Atotonilco el Alto", "industry": "Financial Services", "unit": "L", "phone": "(544) 103-4070", "email": "lpdgpg@test.org"}, {"city": "Puerto Cabello", "industry": "Financial Services", "unit": "L", "phone": "(584) 950-6257", "email": "doidgch@mail.com"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "in", "phone": "(585) 635-9157", "email": "cpgkpc@example.com"}, {"city": "Pekin", "industry": "Financial Services", "unit": "lb", "phone": "(530) 261-1406", "email": "phimel@example.com"}, {"city": "Pekin", "industry": "Telecommunications", "unit": "in", "phone": "(511) 530-6721", "email": "idool@mail.com"}, {"city": "Puerto Cabello", "industry": "Financial Services", "unit": "in", "phone": "(548) 886-5798", "email": "gjfcn@example.com"}, {"city": "Pekin", "industry": "Telecommunications", "unit": "L", "phone": "(550) 172-5068", "email": "gfjdkfo@corp.io"}, {"city": "Pekin", "industry": "Telecommunications", "unit": "km", "phone": "(561) 620-6505", "email": "kijmh@test.org"}, {"city": "Atotonilco el Alto", "industry": "Telecommunications", "unit": "in", "phone": "(571) 530-9752", "email": "fcllcpb@mail.com"}, {"city": "Atotonilco el Alto", "industry": "Financial Services", "unit": "L", "phone": "(520) 798-2943", "email": "bmieg@test.org"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "L", "phone": "(570) 749-0788", "email": "lknmp@test.org"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "km", "phone": "(512) 198-0453", "email": "glhl@mail.com"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "L", "phone": "(509) 972-0718", "email": "ogdcp@example.com"}, {"city": "Pekin", "industry": "Telecommunications", "unit": "lb", "phone": "(568) 520-0951", "email": "ieddjol@corp.io"}, {"city": "Atlanta", "industry": "Financial Services", "unit": "L", "phone": "(562) 530-1786", "email": "ipegb@example.com"}, {"city": "Atlanta", "industry": "Telecommunications", "unit": "lb", "phone": "(541) 740-4177", "email": "oambo@mail.com"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "L", "phone": "(565) 116-3026", "email": "jamkc@corp.io"}, {"city": "Atotonilco el Alto", "industry": "Telecommunications", "unit": "in", "phone": "(560) 448-2898", "email": "bomdj@example.com"}, {"city": "Atotonilco el Alto", "industry": "Financial Services", "unit": "lb", "phone": "(590) 427-7056", "email": "okcehnm@corp.io"}, {"city": "Atlanta", "industry": "Financial Services", "unit": "lb", "phone": "(560) 522-9733", "email": "glpocih@mail.com"}, {"city": "Atotonilco el Alto", "industry": "Financial Services", "unit": "L", "phone": "(561) 440-4045", "email": "dajndf@corp.io"}, {"city": "Puerto Cabello", "industry": "Telecommunications", "unit": "km", "phone": "(513) 015-5159", "email": "nffnhg@corp.io"}], "dirty_cols": ["city", "industry", "unit", "phone", "email"], "clean_cols": ["city", "industry", "unit", "phone", "email"], "plan": {"dataset_summary": "35 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"puerto cabello": "Puerto Cabello", "alanta": "Atlanta", "atotonilco el alto": "Atotonilco el Alto", "PUERTO CABELLO": "Puerto Cabello", "pekin": "Pekin", "Puerto aCbello": "Puerto Cabello", "Pekiin": "Pekin", "ATOTONILCO EL ALTO": "Atotonilco el Alto", "Puero Cabello": "Puerto Cabello", "atlanta": "Atlanta", "Attotonilco el Alto": "Atotonilco el Alto", "ATLANTA": "Atlanta"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"finance": "Financial Services", "Finance": "Financial Services", "Telecmo": "Telecommunications", "Telecom": "Telecommunications", "telco": "Telecommunications", "Banking": "Financial Services", "banking": "Financial Services", "Telco": "Telecommunications", "Finace": "Financial Services", "TELCO": "Telecommunications", "Fintehc": "Financial Services", "teleom": "Telecommunications", "Tjlecom": "Telecommunications", "telecoommunications": "Telecommunications", "FinServ": "Financial Services", "Comms": "Telecommunications", "COMMS": "Telecommunications", "FnServ": "Financial Services", "telecom": "Telecommunications", "FINANCE": "Financial Services"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"POUNDS": "lb", "KM": "km", "litre": "L", "lbs": "lb", "INCHES": "in", "kilometers": "km", "Lb": "lb", "inch": "in", "killometers": "km", "LITER": "L", "\"": "in", "Pound": "lb", "l": "L", "kilometer": "km", "LITRE": "L", "liyers": "L", "Km": "km", "pound": "lb", "LBS": "lb", "lter": "L", "poud": "lb"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"signup_date": "44996", "name": "Lena Lee", "rate": "17.6%", "unit": " miles "}, {"signup_date": "2023-09-20", "name": "Mona Brandt", "rate": "24.8%", "unit": " kilometer "}, {"signup_date": "44977", "name": "  Priya Moore", "rate": "19.5%", "unit": "Inch "}, {"signup_date": "2023-11-12", "name": "   Priya Khan", "rate": "56.6%", "unit": "L"}, {"signup_date": "45144", "name": "  Grace Novak", "rate": "87.9%", "unit": "  l"}, {"signup_date": "2023-05-27", "name": "Sara  Ali", "rate": "81.7%", "unit": "\""}, {"signup_date": "10 Jan 2023", "name": "   Mona Park", "rate": "11.2%", "unit": "\""}, {"signup_date": "2023-12-18", "name": "Lena Fischer", "rate": "13.2%", "unit": "KILOMETER"}, {"signup_date": "8 Nov 2023", "name": "Mona Fischer ", "rate": "41.6%", "unit": "miles"}, {"signup_date": "45079", "name": "Priya Petrov", "rate": "56.1%", "unit": "\""}, {"signup_date": "2023-10-16", "name": "   Lena Smith", "rate": "78.6%", "unit": "Lietrs"}, {"signup_date": "2023-12-23", "name": "Heidi Wong", "rate": "22.8%", "unit": "inches"}, {"signup_date": "9 Sep 2023", "name": "Grace Lee", "rate": "28.2%", "unit": "inches"}, {"signup_date": "17 Dec 2023", "name": "Frank Cruz", "rate": "97.4%", "unit": "Kilometer"}], "clean": [{"signup_date": "2023-03-11", "name": "Lena Lee", "rate": 0.17600000000000002, "unit": "mi"}, {"signup_date": "2023-09-20", "name": "Mona Brandt", "rate": 0.248, "unit": "km"}, {"signup_date": "2023-02-20", "name": "Priya Moore", "rate": 0.195, "unit": "in"}, {"signup_date": "2023-11-12", "name": "Priya Khan", "rate": 0.5660000000000001, "unit": "L"}, {"signup_date": "2023-08-06", "name": "Grace Novak", "rate": 0.879, "unit": "L"}, {"signup_date": "2023-05-27", "name": "Sara Ali", "rate": 0.8170000000000001, "unit": "in"}, {"signup_date": "2023-01-10", "name": "Mona Park", "rate": 0.11199999999999999, "unit": "in"}, {"signup_date": "2023-12-18", "name": "Lena Fischer", "rate": 0.132, "unit": "km"}, {"signup_date": "2023-11-08", "name": "Mona Fischer", "rate": 0.41600000000000004, "unit": "mi"}, {"signup_date": "2023-06-02", "name": "Priya Petrov", "rate": 0.561, "unit": "in"}, {"signup_date": "2023-10-16", "name": "Lena Smith", "rate": 0.7859999999999999, "unit": "L"}, {"signup_date": "2023-12-23", "name": "Heidi Wong", "rate": 0.228, "unit": "in"}, {"signup_date": "2023-09-09", "name": "Grace Lee", "rate": 0.282, "unit": "in"}, {"signup_date": "2023-12-17", "name": "Frank Cruz", "rate": 0.9740000000000001, "unit": "km"}], "dirty_cols": ["signup_date", "name", "rate", "unit"], "clean_cols": ["signup_date", "name", "rate", "unit"], "plan": {"dataset_summary": "14 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miles": "mi", "kilometer": "km", "Inch": "in", "l": "L", "\"": "in", "KILOMETER": "km", "Lietrs": "L", "inches": "in", "Kilometer": "km"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": "GROWTH", "signup_date": "2023-08-15", "name": " Omar Lee", "phone": "510.366.2571", "unnamed": ""}, {"department": "Marketing", "signup_date": "2023-01-24", "name": "Alice  Diaz", "phone": "(505)797-8517", "unnamed": ""}, {"department": " operations ", "signup_date": "2/12/2023", "name": "Heidi Lee", "phone": "565.294.7412", "unnamed": ""}, {"department": "Growth", "signup_date": "9 Aug 2023", "name": "   Heidi Reyes", "phone": "(533)754-0149", "unnamed": ""}, {"department": "   Growth", "signup_date": "45262", "name": "Ivan Khan", "phone": "5606262365", "unnamed": ""}, {"department": "MKTG ", "signup_date": "44934", "name": "  David Smith", "phone": "5874686670", "unnamed": ""}, {"department": "operations", "signup_date": "11/26/2023", "name": "David Adams", "phone": "5965995292", "unnamed": ""}, {"department": "gowth", "signup_date": "2023-10-04", "name": "Eve Ali", "phone": "516-160-1885", "unnamed": ""}, {"department": "Legal & Compliance", "signup_date": "19 Aug 2023", "name": "Karl Fischer", "phone": "599-261-5905", "unnamed": ""}, {"department": "legal", "signup_date": "15 Dec 2023", "name": "   Heidi Petrov", "phone": "520.885.6106", "unnamed": ""}, {"department": "Legal", "signup_date": "2023-08-01", "name": "Heidi Novak ", "phone": "(545)012-9571", "unnamed": ""}, {"department": "legal & compliance", "signup_date": "8 May 2023", "name": "Mona  Lee", "phone": "558.490.7720", "unnamed": ""}, {"department": "Growwth", "signup_date": "5/25/2023", "name": "Carol  Cruz", "phone": "519-947-0531", "unnamed": ""}, {"department": "Ops", "signup_date": "1/26/2023", "name": "Judy Johnson", "phone": "(565)938-2215", "unnamed": ""}, {"department": "legal", "signup_date": "8/21/2023", "name": "Omar Lee  ", "phone": "502-147-1169", "unnamed": ""}, {"department": "legal", "signup_date": "44959", "name": "   Alice Smith", "phone": "5988494757", "unnamed": ""}, {"department": " OPS ", "signup_date": "9/22/2023", "name": "Bob Moore", "phone": "5904799635", "unnamed": ""}, {"department": "Marketing", "signup_date": "4/6/2023", "name": "   Priya Reyes", "phone": "593-180-2200", "unnamed": ""}, {"department": "Marketing", "signup_date": "2023-04-27", "name": "Judy Cruz", "phone": "5659898310", "unnamed": ""}, {"department": "leegal", "signup_date": "2023-03-03", "name": " David Petrov", "phone": "540.121.4785", "unnamed": ""}, {"department": "ops  ", "signup_date": "2023-07-10", "name": "David Smith", "phone": "598.436.9639", "unnamed": ""}, {"department": "Operatins", "signup_date": "24 Jun 2023", "name": "Ivan Cruz", "phone": "5048028817", "unnamed": ""}, {"department": "Legal", "signup_date": "21 Jan 2023", "name": "Frank  Fischer", "phone": "(528)575-1937", "unnamed": ""}, {"department": "operations", "signup_date": "2023-11-21", "name": " Carol Lee", "phone": "(560)406-7713", "unnamed": ""}, {"department": " operations ", "signup_date": "2/12/2023", "name": "Heidi Lee", "phone": "565.294.7412", "unnamed": ""}, {"department": "Operatins", "signup_date": "24 Jun 2023", "name": "Ivan Cruz", "phone": "5048028817", "unnamed": ""}], "clean": [{"department": "Marketing", "signup_date": "2023-08-15", "name": "Omar Lee", "phone": "(510) 366-2571"}, {"department": "Marketing", "signup_date": "2023-01-24", "name": "Alice Diaz", "phone": "(505) 797-8517"}, {"department": "Operations", "signup_date": "2023-02-12", "name": "Heidi Lee", "phone": "(565) 294-7412"}, {"department": "Marketing", "signup_date": "2023-08-09", "name": "Heidi Reyes", "phone": "(533) 754-0149"}, {"department": "Marketing", "signup_date": "2023-12-02", "name": "Ivan Khan", "phone": "(560) 626-2365"}, {"department": "Marketing", "signup_date": "2023-01-08", "name": "David Smith", "phone": "(587) 468-6670"}, {"department": "Operations", "signup_date": "2023-11-26", "name": "David Adams", "phone": "(596) 599-5292"}, {"department": "Marketing", "signup_date": "2023-10-04", "name": "Eve Ali", "phone": "(516) 160-1885"}, {"department": "Legal", "signup_date": "2023-08-19", "name": "Karl Fischer", "phone": "(599) 261-5905"}, {"department": "Legal", "signup_date": "2023-12-15", "name": "Heidi Petrov", "phone": "(520) 885-6106"}, {"department": "Legal", "signup_date": "2023-08-01", "name": "Heidi Novak", "phone": "(545) 012-9571"}, {"department": "Legal", "signup_date": "2023-05-08", "name": "Mona Lee", "phone": "(558) 490-7720"}, {"department": "Marketing", "signup_date": "2023-05-25", "name": "Carol Cruz", "phone": "(519) 947-0531"}, {"department": "Operations", "signup_date": "2023-01-26", "name": "Judy Johnson", "phone": "(565) 938-2215"}, {"department": "Legal", "signup_date": "2023-08-21", "name": "Omar Lee", "phone": "(502) 147-1169"}, {"department": "Legal", "signup_date": "2023-02-02", "name": "Alice Smith", "phone": "(598) 849-4757"}, {"department": "Operations", "signup_date": "2023-09-22", "name": "Bob Moore", "phone": "(590) 479-9635"}, {"department": "Marketing", "signup_date": "2023-04-06", "name": "Priya Reyes", "phone": "(593) 180-2200"}, {"department": "Marketing", "signup_date": "2023-04-27", "name": "Judy Cruz", "phone": "(565) 989-8310"}, {"department": "Legal", "signup_date": "2023-03-03", "name": "David Petrov", "phone": "(540) 121-4785"}, {"department": "Operations", "signup_date": "2023-07-10", "name": "David Smith", "phone": "(598) 436-9639"}, {"department": "Operations", "signup_date": "2023-06-24", "name": "Ivan Cruz", "phone": "(504) 802-8817"}, {"department": "Legal", "signup_date": "2023-01-21", "name": "Frank Fischer", "phone": "(528) 575-1937"}, {"department": "Operations", "signup_date": "2023-11-21", "name": "Carol Lee", "phone": "(560) 406-7713"}], "dirty_cols": ["department", "signup_date", "name", "phone", "unnamed"], "clean_cols": ["department", "signup_date", "name", "phone"], "plan": {"dataset_summary": "26 rows × 5 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GROWTH": "Marketing", "operations": "Operations", "Growth": "Marketing", "MKTG": "Marketing", "gowth": "Marketing", "Legal & Compliance": "Legal", "legal": "Legal", "legal & compliance": "Legal", "Growwth": "Marketing", "Ops": "Operations", "OPS": "Operations", "leegal": "Legal", "ops": "Operations", "Operatins": "Operations"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"name": "Eve  Adams", "city": "Parnaraa", "department": "People Ops", "currency": "Guinean  Franc", "extra": ""}, {"name": "Frank Petrov   ", "city": "Pameungpeuk", "department": "People Osp", "currency": "Yuan Renminbi   ", "extra": ""}, {"name": "Grace  Wong", "city": "Pameungpuk", "department": "ops", "currency": "Argentine Peso", "extra": ""}, {"name": "   Carol Moore", "city": "De Meern", "department": "People Ops", "currency": "gnf", "extra": ""}, {"name": "Bob Ali", "city": " Pameungpeuk ", "department": "ops", "currency": "ARS  ", "extra": ""}, {"name": "Carol Cruz", "city": "   parnarama", "department": "hr", "currency": "gnf", "extra": ""}, {"name": "Eve Brandt", "city": "  Pameungpeuk", "department": "HR", "currency": "  CNY", "extra": ""}, {"name": "   Priya Smith", "city": "Sngai Udang  ", "department": "People Ops", "currency": "Guinean Franc", "extra": ""}, {"name": "Ivan Adams", "city": "Fuorigrotta", "department": "  Ops", "currency": "cny", "extra": ""}, {"name": "Alice Brandt", "city": "Sungai  Udang", "department": " people ops", "currency": "Argentine  Peso", "extra": ""}, {"name": "Carol Cruz", "city": "   parnarama", "department": "hr", "currency": "gnf", "extra": ""}, {"name": "Eve  Adams", "city": "Parnaraa", "department": "People Ops", "currency": "Guinean  Franc", "extra": ""}, {"name": "", "city": "", "department": "", "currency": "", "extra": ""}], "clean": [{"name": "Eve Adams", "city": "Parnarama", "department": "Human Resources", "currency": "GNF"}, {"name": "Frank Petrov", "city": "Pameungpeuk", "department": "Human Resources", "currency": "CNY"}, {"name": "Grace Wong", "city": "Pameungpeuk", "department": "Operations", "currency": "ARS"}, {"name": "Carol Moore", "city": "De Meern", "department": "Human Resources", "currency": "GNF"}, {"name": "Bob Ali", "city": "Pameungpeuk", "department": "Operations", "currency": "ARS"}, {"name": "Carol Cruz", "city": "Parnarama", "department": "Human Resources", "currency": "GNF"}, {"name": "Eve Brandt", "city": "Pameungpeuk", "department": "Human Resources", "currency": "CNY"}, {"name": "Priya Smith", "city": "Sungai Udang", "department": "Human Resources", "currency": "GNF"}, {"name": "Ivan Adams", "city": "Fuorigrotta", "department": "Operations", "currency": "CNY"}, {"name": "Alice Brandt", "city": "Sungai Udang", "department": "Human Resources", "currency": "ARS"}], "dirty_cols": ["name", "city", "department", "currency", "extra"], "clean_cols": ["name", "city", "department", "currency"], "plan": {"dataset_summary": "13 rows × 5 columns. 4 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Parnaraa": "Parnarama", "Pameungpuk": "Pameungpeuk", "parnarama": "Parnarama", "Sngai Udang": "Sungai Udang"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"People Ops": "Human Resources", "People Osp": "Human Resources", "ops": "Operations", "hr": "Human Resources", "HR": "Human Resources", "Ops": "Operations", "people ops": "Human Resources"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Guinean Franc": "GNF", "Yuan Renminbi": "CNY", "Argentine Peso": "ARS", "gnf": "GNF", "cny": "CNY"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "MICHIGAN", "signup_date": "28 Aug 2023", "company": "Hooli", "is_active": "0", "country": "Fiji", "email": "njfnngk@test.org ", "phone": "524-448-0883", "currency": "aed"}, {"state": "MT", "signup_date": "2023-03-11", "company": "Soylent Corp", "is_active": "true", "country": "   kena", "email": "mbca@test.org", "phone": "560.352.0681", "currency": "COLOMBIAN PESO"}, {"state": "MI", "signup_date": "6 Jan 2023", "company": " Hooli", "is_active": "FALSE", "country": "republic of kenya", "email": "aoack@example.com", "phone": "(535)885-1146", "currency": "CHIEAN PESO"}, {"state": "Norht Dakota", "signup_date": "9/1/2023", "company": "Hooli", "is_active": "Yes", "country": "Fici", "email": "cojng@test.org", "phone": "590.679.2050", "currency": "CLP"}, {"state": "mt", "signup_date": "27 Jul 2023", "company": "Initech ", "is_active": "T", "country": "fji ", "email": "HNJNME@TEST.ORG", "phone": "594.778.9388", "currency": "dominican peso"}, {"state": "mt", "signup_date": "3/13/2023", "company": "Stark Industries", "is_active": "true", "country": "Fiji", "email": "fkpm@example.com", "phone": "512-845-9591", "currency": "Dominican Peso"}, {"state": "mt", "signup_date": "2023-02-03", "company": "Wonka Co", "is_active": "0", "country": "republic of the congo", "email": "  hdcna@mail.com", "phone": "595.865.3428", "currency": "Dominican Peso  "}, {"state": "north  dakota", "signup_date": "2023-05-23", "company": "  Acme Inc", "is_active": "F", "country": "Congo", "email": "  PKCGIML@TEST.ORG", "phone": "523-609-5325", "currency": "  chilean peso"}, {"state": "north dakota", "signup_date": "2023-01-03", "company": " Initech ", "is_active": "0", "country": "FJI   ", "email": "  jgidpdh@test.org", "phone": "5885196921", "currency": "Dominican Peso"}, {"state": "nd", "signup_date": "4/25/2023", "company": "   Globex", "is_active": "No", "country": "COG", "email": "IGPHK@TEST.ORG", "phone": "545-717-0674", "currency": "CLP"}, {"state": "Arkansas", "signup_date": "2023-01-15", "company": " Wonka Co", "is_active": "F", "country": "KEN", "email": "ebmdml@corp.io  ", "phone": "(516)454-8761", "currency": "CLP"}, {"state": "MONTANA", "signup_date": "2023-06-19", "company": " Globex ", "is_active": "No", "country": "Republic Of Sierra Leone", "email": "ggjnkd@mail.com", "phone": "5164348675", "currency": "dop"}, {"state": "ND", "signup_date": "2023-10-17", "company": "Vehement", "is_active": "1", "country": "Republic of Kenya", "email": "BELHGAJ@MAIL.COM", "phone": "592.362.4667", "currency": "dominican peso"}, {"state": "Micigan", "signup_date": "45233", "company": "Hooli", "is_active": "0", "country": "Congo", "email": " GCAOO@MAIL.COM ", "phone": "505-453-5048", "currency": "   Clp"}, {"state": "mt", "signup_date": "21 Mar 2023", "company": " Cyberdyne ", "is_active": "false", "country": "sl  ", "email": "llaba@corp.io   ", "phone": "5378782878", "currency": "DOP"}, {"state": "MI", "signup_date": "2/11/2023", "company": "Initech", "is_active": "1", "country": "cg", "email": "FNEPPB@EXAMPLE.COM", "phone": "5274770678", "currency": "COP"}, {"state": "montana", "signup_date": "11/12/2023", "company": " Cyberdyne", "is_active": "Yes", "country": "  republic of sierra leone", "email": "ADCJ@CORP.IO", "phone": "(596)798-1726", "currency": "AED"}, {"state": "MI", "signup_date": "45172", "company": "Umbrella  ", "is_active": "1", "country": "SL", "email": "BDMFGH@CORP.IO", "phone": "524.506.1982", "currency": "Colombian Peso"}, {"state": " ND ", "signup_date": "2023-04-23", "company": "   Soylent Corp", "is_active": "true", "country": "SLE", "email": "lolgmh@example.com", "phone": "(502)098-2346", "currency": "DOP"}, {"state": "montana", "signup_date": "15 Aug 2023", "company": "Stark Industries", "is_active": "false", "country": " Kenya", "email": " LCGGD@TEST.ORG ", "phone": "5688518432", "currency": "   cop"}, {"state": "mt  ", "signup_date": "44997", "company": "Stark Industries  ", "is_active": "1", "country": "Republic Of Kenya", "email": "INEFHI@CORP.IO", "phone": "593-770-0411", "currency": "clp"}, {"state": "MT", "signup_date": "9 Sep 2023", "company": "Acme Inc   ", "is_active": "T", "country": " Sl", "email": "  OICEK@EXAMPLE.COM", "phone": "5660880679", "currency": "Dominian Peso"}, {"state": "Arkansas", "signup_date": "1/26/2023", "company": "Hooli", "is_active": "F", "country": "congo", "email": "LOPPLAL@TEST.ORG  ", "phone": "5260218887", "currency": "CLP"}, {"state": "north dakota", "signup_date": "2023-08-11", "company": "Hooli ", "is_active": "FALSE", "country": "Republic of Kenya", "email": " hggi@test.org", "phone": "557.164.0073", "currency": "aed"}, {"state": "MI", "signup_date": "21 Mar 2023", "company": "Cyberdyne   ", "is_active": "1", "country": "Repuboic of Sierra Leone", "email": "OAMBGD@MAIL.COM  ", "phone": "5630868132", "currency": "Aed"}, {"state": "Montana", "signup_date": "5/19/2023", "company": "Vehement", "is_active": "No", "country": "  Republic Of Sierra Leone", "email": "BALID@MAIL.COM   ", "phone": "538.256.7956", "currency": "dominican  peso"}, {"state": " ND ", "signup_date": "2023-01-14", "company": "Cyberdyne", "is_active": "1", "country": " ken ", "email": "GAKKBE@MAIL.COM ", "phone": "(503)923-2744", "currency": "Dominican Peso"}, {"state": "Moneana", "signup_date": "27 Sep 2023", "company": "Vehement", "is_active": "1", "country": "Republic of Kenya", "email": "adcnop@example.com", "phone": "590.999.3833", "currency": "uae dirham"}, {"state": "Arkansas", "signup_date": "2023-09-18", "company": "Acme  Inc", "is_active": "T", "country": " fj ", "email": " cdpi@test.org", "phone": "527.049.1243", "currency": "Dop"}, {"state": "MICHIGAN   ", "signup_date": "45042", "company": "  Acme Inc", "is_active": "1", "country": "Kenya", "email": "mcheoji@corp.io", "phone": "5992385282", "currency": "uae  dirham"}, {"state": "nd", "signup_date": "2023-11-08", "company": " Vehement ", "is_active": "T", "country": "Republic of Fiji", "email": "ldknc@example.com", "phone": "536.693.4217", "currency": "Cop"}, {"state": "AR", "signup_date": "1 Jan 2023", "company": "Acme Inc", "is_active": "N", "country": "kenya   ", "email": "MIONBBC@TEST.ORG", "phone": "5312949998", "currency": " dop "}, {"state": " ar ", "signup_date": "45280", "company": "Wonka Co", "is_active": "Y", "country": "Sierra Leone   ", "email": "fmmkbpo@example.com", "phone": "596.004.7508", "currency": " AED "}, {"state": "ARKANSAS", "signup_date": "1/1/2023", "company": " Hooli", "is_active": "FALSE", "country": "Sierra Leone", "email": " nnak@mail.com", "phone": "5252229210", "currency": "clp"}, {"state": "MI", "signup_date": "2023-04-24", "company": "Stark Industries", "is_active": "No", "country": "ken", "email": "   cadod@example.com", "phone": "564-195-5235", "currency": "DOMINICAN PESO"}, {"state": "MI", "signup_date": "2023-02-26", "company": " Globex ", "is_active": "true", "country": "Fiji", "email": "JENELP@TEST.ORG", "phone": "589.985.5441", "currency": "COP"}, {"state": "Ar", "signup_date": "17 Apr 2023", "company": "Globex", "is_active": "Yes", "country": "republic of sierraa leone", "email": "   dgflgh@test.org", "phone": "(502)299-7439", "currency": "Dominican  Peso"}, {"state": "AR", "signup_date": "21 Jul 2023", "company": "Globex", "is_active": "No", "country": "Republic of Kenya", "email": "pdlm@mail.com", "phone": "589-220-7681", "currency": "DOP"}, {"state": "ND", "signup_date": "2/27/2023", "company": "Globex", "is_active": "Y", "country": "KEN", "email": "MMKEFEF@MAIL.COM", "phone": "546-570-8583", "currency": "clp"}, {"state": "MI", "signup_date": "2/11/2023", "company": "Initech", "is_active": "1", "country": "cg", "email": "FNEPPB@EXAMPLE.COM", "phone": "5274770678", "currency": "COP"}, {"state": "Moneana", "signup_date": "27 Sep 2023", "company": "Vehement", "is_active": "1", "country": "Republic of Kenya", "email": "adcnop@example.com", "phone": "590.999.3833", "currency": "uae dirham"}], "clean": [{"state": "Michigan", "signup_date": "2023-08-28", "company": "Hooli", "is_active": false, "country": "Fiji", "email": "njfnngk@test.org", "phone": "(524) 448-0883", "currency": "AED"}, {"state": "Montana", "signup_date": "2023-03-11", "company": "Soylent Corp", "is_active": true, "country": "Kenya", "email": "mbca@test.org", "phone": "(560) 352-0681", "currency": "COP"}, {"state": "Michigan", "signup_date": "2023-01-06", "company": "Hooli", "is_active": false, "country": "Kenya", "email": "aoack@example.com", "phone": "(535) 885-1146", "currency": "CLP"}, {"state": "North Dakota", "signup_date": "2023-09-01", "company": "Hooli", "is_active": true, "country": "Fiji", "email": "cojng@test.org", "phone": "(590) 679-2050", "currency": "CLP"}, {"state": "Montana", "signup_date": "2023-07-27", "company": "Initech", "is_active": true, "country": "Fiji", "email": "hnjnme@test.org", "phone": "(594) 778-9388", "currency": "DOP"}, {"state": "Montana", "signup_date": "2023-03-13", "company": "Stark Industries", "is_active": true, "country": "Fiji", "email": "fkpm@example.com", "phone": "(512) 845-9591", "currency": "DOP"}, {"state": "Montana", "signup_date": "2023-02-03", "company": "Wonka Co", "is_active": false, "country": "Congo", "email": "hdcna@mail.com", "phone": "(595) 865-3428", "currency": "DOP"}, {"state": "North Dakota", "signup_date": "2023-05-23", "company": "Acme Inc", "is_active": false, "country": "Congo", "email": "pkcgiml@test.org", "phone": "(523) 609-5325", "currency": "CLP"}, {"state": "North Dakota", "signup_date": "2023-01-03", "company": "Initech", "is_active": false, "country": "Fiji", "email": "jgidpdh@test.org", "phone": "(588) 519-6921", "currency": "DOP"}, {"state": "North Dakota", "signup_date": "2023-04-25", "company": "Globex", "is_active": false, "country": "Congo", "email": "igphk@test.org", "phone": "(545) 717-0674", "currency": "CLP"}, {"state": "Arkansas", "signup_date": "2023-01-15", "company": "Wonka Co", "is_active": false, "country": "Kenya", "email": "ebmdml@corp.io", "phone": "(516) 454-8761", "currency": "CLP"}, {"state": "Montana", "signup_date": "2023-06-19", "company": "Globex", "is_active": false, "country": "Sierra Leone", "email": "ggjnkd@mail.com", "phone": "(516) 434-8675", "currency": "DOP"}, {"state": "North Dakota", "signup_date": "2023-10-17", "company": "Vehement", "is_active": true, "country": "Kenya", "email": "belhgaj@mail.com", "phone": "(592) 362-4667", "currency": "DOP"}, {"state": "Michigan", "signup_date": "2023-11-03", "company": "Hooli", "is_active": false, "country": "Congo", "email": "gcaoo@mail.com", "phone": "(505) 453-5048", "currency": "CLP"}, {"state": "Montana", "signup_date": "2023-03-21", "company": "Cyberdyne", "is_active": false, "country": "Sierra Leone", "email": "llaba@corp.io", "phone": "(537) 878-2878", "currency": "DOP"}, {"state": "Michigan", "signup_date": "2023-02-11", "company": "Initech", "is_active": true, "country": "Congo", "email": "fneppb@example.com", "phone": "(527) 477-0678", "currency": "COP"}, {"state": "Montana", "signup_date": "2023-11-12", "company": "Cyberdyne", "is_active": true, "country": "Sierra Leone", "email": "adcj@corp.io", "phone": "(596) 798-1726", "currency": "AED"}, {"state": "Michigan", "signup_date": "2023-09-03", "company": "Umbrella", "is_active": true, "country": "Sierra Leone", "email": "bdmfgh@corp.io", "phone": "(524) 506-1982", "currency": "COP"}, {"state": "North Dakota", "signup_date": "2023-04-23", "company": "Soylent Corp", "is_active": true, "country": "Sierra Leone", "email": "lolgmh@example.com", "phone": "(502) 098-2346", "currency": "DOP"}, {"state": "Montana", "signup_date": "2023-08-15", "company": "Stark Industries", "is_active": false, "country": "Kenya", "email": "lcggd@test.org", "phone": "(568) 851-8432", "currency": "COP"}, {"state": "Montana", "signup_date": "2023-03-12", "company": "Stark Industries", "is_active": true, "country": "Kenya", "email": "inefhi@corp.io", "phone": "(593) 770-0411", "currency": "CLP"}, {"state": "Montana", "signup_date": "2023-09-09", "company": "Acme Inc", "is_active": true, "country": "Sierra Leone", "email": "oicek@example.com", "phone": "(566) 088-0679", "currency": "DOP"}, {"state": "Arkansas", "signup_date": "2023-01-26", "company": "Hooli", "is_active": false, "country": "Congo", "email": "lopplal@test.org", "phone": "(526) 021-8887", "currency": "CLP"}, {"state": "North Dakota", "signup_date": "2023-08-11", "company": "Hooli", "is_active": false, "country": "Kenya", "email": "hggi@test.org", "phone": "(557) 164-0073", "currency": "AED"}, {"state": "Michigan", "signup_date": "2023-03-21", "company": "Cyberdyne", "is_active": true, "country": "Sierra Leone", "email": "oambgd@mail.com", "phone": "(563) 086-8132", "currency": "AED"}, {"state": "Montana", "signup_date": "2023-05-19", "company": "Vehement", "is_active": false, "country": "Sierra Leone", "email": "balid@mail.com", "phone": "(538) 256-7956", "currency": "DOP"}, {"state": "North Dakota", "signup_date": "2023-01-14", "company": "Cyberdyne", "is_active": true, "country": "Kenya", "email": "gakkbe@mail.com", "phone": "(503) 923-2744", "currency": "DOP"}, {"state": "Montana", "signup_date": "2023-09-27", "company": "Vehement", "is_active": true, "country": "Kenya", "email": "adcnop@example.com", "phone": "(590) 999-3833", "currency": "AED"}, {"state": "Arkansas", "signup_date": "2023-09-18", "company": "Acme Inc", "is_active": true, "country": "Fiji", "email": "cdpi@test.org", "phone": "(527) 049-1243", "currency": "DOP"}, {"state": "Michigan", "signup_date": "2023-04-26", "company": "Acme Inc", "is_active": true, "country": "Kenya", "email": "mcheoji@corp.io", "phone": "(599) 238-5282", "currency": "AED"}, {"state": "North Dakota", "signup_date": "2023-11-08", "company": "Vehement", "is_active": true, "country": "Fiji", "email": "ldknc@example.com", "phone": "(536) 693-4217", "currency": "COP"}, {"state": "Arkansas", "signup_date": "2023-01-01", "company": "Acme Inc", "is_active": false, "country": "Kenya", "email": "mionbbc@test.org", "phone": "(531) 294-9998", "currency": "DOP"}, {"state": "Arkansas", "signup_date": "2023-12-20", "company": "Wonka Co", "is_active": true, "country": "Sierra Leone", "email": "fmmkbpo@example.com", "phone": "(596) 004-7508", "currency": "AED"}, {"state": "Arkansas", "signup_date": "2023-01-01", "company": "Hooli", "is_active": false, "country": "Sierra Leone", "email": "nnak@mail.com", "phone": "(525) 222-9210", "currency": "CLP"}, {"state": "Michigan", "signup_date": "2023-04-24", "company": "Stark Industries", "is_active": false, "country": "Kenya", "email": "cadod@example.com", "phone": "(564) 195-5235", "currency": "DOP"}, {"state": "Michigan", "signup_date": "2023-02-26", "company": "Globex", "is_active": true, "country": "Fiji", "email": "jenelp@test.org", "phone": "(589) 985-5441", "currency": "COP"}, {"state": "Arkansas", "signup_date": "2023-04-17", "company": "Globex", "is_active": true, "country": "Sierra Leone", "email": "dgflgh@test.org", "phone": "(502) 299-7439", "currency": "DOP"}, {"state": "Arkansas", "signup_date": "2023-07-21", "company": "Globex", "is_active": false, "country": "Kenya", "email": "pdlm@mail.com", "phone": "(589) 220-7681", "currency": "DOP"}, {"state": "North Dakota", "signup_date": "2023-02-27", "company": "Globex", "is_active": true, "country": "Kenya", "email": "mmkefef@mail.com", "phone": "(546) 570-8583", "currency": "CLP"}], "dirty_cols": ["state", "signup_date", "company", "is_active", "country", "email", "phone", "currency"], "clean_cols": ["state", "signup_date", "company", "is_active", "country", "email", "phone", "currency"], "plan": {"dataset_summary": "41 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MICHIGAN": "Michigan", "MT": "Montana", "MI": "Michigan", "Norht Dakota": "North Dakota", "mt": "Montana", "north dakota": "North Dakota", "nd": "North Dakota", "MONTANA": "Montana", "ND": "North Dakota", "Micigan": "Michigan", "montana": "Montana", "Moneana": "Montana", "AR": "Arkansas", "ar": "Arkansas", "ARKANSAS": "Arkansas", "Ar": "Arkansas"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kena": "Kenya", "republic of kenya": "Kenya", "Fici": "Fiji", "fji": "Fiji", "republic of the congo": "Congo", "FJI": "Fiji", "COG": "Congo", "KEN": "Kenya", "Republic Of Sierra Leone": "Sierra Leone", "Republic of Kenya": "Kenya", "sl": "Sierra Leone", "cg": "Congo", "republic of sierra leone": "Sierra Leone", "SL": "Sierra Leone", "SLE": "Sierra Leone", "Republic Of Kenya": "Kenya", "Sl": "Sierra Leone", "congo": "Congo", "Repuboic of Sierra Leone": "Sierra Leone", "ken": "Kenya", "fj": "Fiji", "Republic of Fiji": "Fiji", "kenya": "Kenya", "republic of sierraa leone": "Sierra Leone"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"aed": "AED", "COLOMBIAN PESO": "COP", "CHIEAN PESO": "CLP", "dominican peso": "DOP", "Dominican Peso": "DOP", "chilean peso": "CLP", "dop": "DOP", "Clp": "CLP", "Colombian Peso": "COP", "cop": "COP", "clp": "CLP", "Dominian Peso": "DOP", "Aed": "AED", "uae dirham": "AED", "Dop": "DOP", "Cop": "COP", "DOMINICAN PESO": "DOP"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "David Khan", "rate": "20.4%", "phone": "560.402.8168", "extra": ""}, {"name": "Heidi Petrov ", "rate": "80.6%", "phone": "561-457-3868", "extra": ""}, {"name": "Ivan Wong", "rate": "80.4%", "phone": "569.450.7547", "extra": ""}, {"name": "   Bob Reyes", "rate": "5.3%", "phone": "(546)709-0025", "extra": ""}, {"name": "Lena Smith", "rate": "24.6%", "phone": "5486292974", "extra": ""}, {"name": "Carol Johnson", "rate": "72.8%", "phone": "5278086274", "extra": ""}, {"name": "Karl Park   ", "rate": "8.5%", "phone": "5578941174", "extra": ""}, {"name": "Heidi Khan", "rate": "82.6%", "phone": "543.114.5734", "extra": ""}, {"name": "Carol Cruz", "rate": "89.1%", "phone": "5924559858", "extra": ""}, {"name": "Lena Petrov", "rate": "50.0%", "phone": "519-968-5625", "extra": ""}, {"name": "Bob  Brandt", "rate": "38.7%", "phone": "(572)012-6763", "extra": ""}, {"name": "David Khan", "rate": "20.4%", "phone": "560.402.8168", "extra": ""}], "clean": [{"name": "David Khan", "rate": 0.204, "phone": "(560) 402-8168"}, {"name": "Heidi Petrov", "rate": 0.8059999999999999, "phone": "(561) 457-3868"}, {"name": "Ivan Wong", "rate": 0.804, "phone": "(569) 450-7547"}, {"name": "Bob Reyes", "rate": 0.053, "phone": "(546) 709-0025"}, {"name": "Lena Smith", "rate": 0.24600000000000002, "phone": "(548) 629-2974"}, {"name": "Carol Johnson", "rate": 0.728, "phone": "(527) 808-6274"}, {"name": "Karl Park", "rate": 0.085, "phone": "(557) 894-1174"}, {"name": "Heidi Khan", "rate": 0.826, "phone": "(543) 114-5734"}, {"name": "Carol Cruz", "rate": 0.8909999999999999, "phone": "(592) 455-9858"}, {"name": "Lena Petrov", "rate": 0.5, "phone": "(519) 968-5625"}, {"name": "Bob Brandt", "rate": 0.387, "phone": "(572) 012-6763"}], "dirty_cols": ["name", "rate", "phone", "extra"], "clean_cols": ["name", "rate", "phone"], "plan": {"dataset_summary": "12 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"company": "   Wonka Co", "email": "pienea@corp.io", "country": "EASTERN REPUBLIC OF URUGUYA", "status": "canceled", "currency": " htg ", "city": " Guxi ", "is_active": "FALSE", "unit": "mL  "}, {"company": "Hooli", "email": "LODNFNH@EXAMPLE.COM", "country": "Zw", "status": "active", "currency": "Bob", "city": "Kom Ombo", "is_active": "TRUE", "unit": "km"}, {"company": "Wonka  Co", "email": "afib@corp.io ", "country": "ury", "status": "churned", "currency": "HTG", "city": "kom ombo", "is_active": "1", "unit": "mls"}, {"company": "Hooli   ", "email": "ABIIJ@TEST.ORG", "country": "ZW", "status": "on hold", "currency": "GOURDE", "city": "Rees", "is_active": "Y", "unit": " IN"}, {"company": "Wonka Co", "email": " HCKKC@TEST.ORG", "country": "nc", "status": "active", "currency": "Gourde  ", "city": "Rees", "is_active": "0", "unit": "in"}, {"company": "Initech", "email": "bjefhlc@corp.io", "country": "URY", "status": "canceled", "currency": "Boliviano", "city": "rees   ", "is_active": "N", "unit": "inch"}, {"company": "Soylent  Corp", "email": "iobdphf@corp.io", "country": "ZWE", "status": "On  Hold", "currency": "htg  ", "city": "GUXI", "is_active": "FALSE", "unit": "ml"}, {"company": "Acme Inc", "email": "pamle@example.com", "country": "nc", "status": "   paused", "currency": "bob", "city": "Guxi", "is_active": "Y", "unit": "mL"}, {"company": "Initech", "email": "JHLN@MAIL.COM", "country": "  zwe", "status": "  active", "currency": "Gourde", "city": "Chorley", "is_active": "Yes", "unit": "\""}, {"company": "Acme Inc", "email": "kjcai@corp.io", "country": "URY", "status": "trial", "currency": " BOB", "city": "Guxi ", "is_active": "FALSE", "unit": "MILLLIITER   "}, {"company": "Initech", "email": " CBLBP@EXAMPLE.COM", "country": "Zwe", "status": "pauhed   ", "currency": " boliviano ", "city": "GUXI", "is_active": "FALSE", "unit": " km"}, {"company": "Hooli", "email": "elhggc@example.com", "country": "uy", "status": "On Hold", "currency": " boliviano", "city": "Gui", "is_active": "T", "unit": "In"}, {"company": "  Cyberdyne", "email": "  DNLMFL@CORP.IO", "country": "ury", "status": "PAUSED", "currency": "Gourde", "city": "chorley", "is_active": "Yes", "unit": "kilometrs"}, {"company": "Soylent Corp", "email": " acbafpc@example.com ", "country": "zwe", "status": "Paused  ", "currency": "  Htg", "city": "Guxi", "is_active": "FALSE", "unit": "kilometers"}, {"company": "Wonka Co", "email": " DKAEB@TEST.ORG ", "country": "NCL", "status": "PAUSED", "currency": "Gourde", "city": "chorley", "is_active": "T", "unit": "\""}, {"company": "Acme Inc", "email": "CGDGLLN@TEST.ORG", "country": "New Caledonnia", "status": "FREE TRIAL", "currency": "Gourde", "city": "Chorley", "is_active": "false", "unit": "in"}, {"company": "Cyberdyne", "email": "dofnifm@test.org", "country": "NCL", "status": "TRIIAL", "currency": " bob ", "city": "Kom Ombo", "is_active": "F", "unit": " in "}, {"company": "Vehement", "email": "BNCMEEG@MAIL.COM", "country": "Eastern Republic Of Urusuay", "status": "Churned", "currency": "htg", "city": "Chorley", "is_active": "Yes", "unit": "ml  "}, {"company": "Hooli", "email": " eakkak@example.com", "country": "Eastern Republic of Uruguay", "status": " Trihl ", "currency": "Gourde", "city": "rees", "is_active": "1", "unit": "kilometers   "}, {"company": "Globex", "email": "mjkcnm@corp.io", "country": "zimbabwe", "status": "ACTVIE", "currency": "GOURDE", "city": "Chorley", "is_active": "Y", "unit": "ml"}, {"company": "Umbrella", "email": "PCLFIL@EXAMPLE.COM", "country": "ZWE", "status": "Paused", "currency": "BOB", "city": "CHORLEY", "is_active": "T", "unit": " inches "}, {"company": "Globex  ", "email": "fgelhdm@mail.com", "country": "new caledonia  ", "status": "canceled", "currency": "BOB", "city": " guxi ", "is_active": "false", "unit": "mL"}, {"company": "   Acme Inc", "email": "KOGKNHK@EXAMPLE.COM", "country": "zwe  ", "status": "Churned ", "currency": "BOB", "city": "Guxi", "is_active": "Y", "unit": "INCHES"}, {"company": "Hooli  ", "email": "HGJMGH@CORP.IO", "country": "New Caledonia", "status": "   TRIAL", "currency": "BOB", "city": "Choorley", "is_active": "1", "unit": "In  "}, {"company": "Acme Inc   ", "email": "PKIJFPA@EXAMPLE.COM  ", "country": "uy", "status": "canceled", "currency": " HTG", "city": "kom ombo", "is_active": "FALSE", "unit": "kilometers"}, {"company": "Hooli  ", "email": "BPDFKH@EXAMPLE.COM", "country": "EASTERN REPUBLIC OF URUUGUAY", "status": "free trial", "currency": "HTG", "city": " KOM OMBO", "is_active": "false", "unit": "milliliter"}, {"company": " Cyberdyne ", "email": "andkd@mail.com", "country": "NEW CALEDONIA", "status": "Churned", "currency": " GOURDE", "city": "Guxi ", "is_active": "F", "unit": "\""}, {"company": "Hooli", "email": "  AFJC@MAIL.COM", "country": "   EASTERN REPUBLIC OF URUGUAY", "status": "on hold", "currency": "Gourde  ", "city": "Chorley", "is_active": "F", "unit": "ml"}, {"company": "Soylent Corp", "email": "CCKAEDO@MAIL.COM", "country": " NC ", "status": "   paused", "currency": "boliviano", "city": "Guxi", "is_active": "N", "unit": "kilometers"}, {"company": "Stark  Industries", "email": "DJOOLAH@MAIL.COM   ", "country": " Zw", "status": "paused", "currency": "Boliviano ", "city": "kom ombo", "is_active": "1", "unit": "mls"}, {"company": "Globex   ", "email": "  jecjo@mail.com", "country": "ncl", "status": " cancelled", "currency": "bob", "city": "Rees", "is_active": "1", "unit": "in"}, {"company": "Umbrella", "email": "  aafgkk@mail.com", "country": "ury", "status": "TRIAL", "currency": " Boliviano ", "city": "Kom Ombo", "is_active": "0", "unit": "in"}, {"company": "Acme  Inc", "email": "ecpkok@mail.com", "country": "ncl   ", "status": "on hold", "currency": "Gourde", "city": "kom ombo", "is_active": "F", "unit": "inch"}, {"company": "Wonka Co ", "email": "cgdnfab@test.org", "country": "nc", "status": "churned", "currency": "Boliviano", "city": "chorley", "is_active": "N", "unit": "inches"}, {"company": "Cyberdyne", "email": " bmjpn@example.com ", "country": "Eastern Republic Of Uruguay", "status": " Churned", "currency": " Boliviano", "city": "rees", "is_active": "No", "unit": "mL"}, {"company": "Soylent Corp", "email": "ckpjff@example.com ", "country": "Eastern Republic of Uruguay", "status": "On Hold", "currency": "BOB", "city": "guxi", "is_active": "true", "unit": "milliliter"}, {"company": " Wonka Co", "email": "PHKDA@EXAMPLE.COM", "country": "nc", "status": "Paused", "currency": "Gourde", "city": "chorley", "is_active": "N", "unit": "Inch"}, {"company": "  Stark Industries", "email": "FKBJ@TEST.ORG", "country": "ZWE", "status": "ACTIVE   ", "currency": "gourde", "city": "GUXI", "is_active": "F", "unit": "kilometer"}, {"company": " Cyberdyne ", "email": "BCEBDJ@CORP.IO", "country": "New Caledonia", "status": " churned", "currency": "   bob", "city": "Chorley", "is_active": "Y", "unit": "ML"}, {"company": "Soylent Corp", "email": " ehba@example.com ", "country": "Eastern Republic of Uruguay ", "status": "canceled", "currency": "htg", "city": "Gxui", "is_active": "F", "unit": "IN"}, {"company": "  Globex", "email": "aclggnj@mail.com", "country": "eastern republic of uruguay", "status": "pased", "currency": "htg", "city": "Rees", "is_active": "0", "unit": "millilter"}, {"company": "Stark Industries", "email": "HJKHC@EXAMPLE.COM", "country": "uruguay", "status": "ATIVE", "currency": "Gourde", "city": "Kom  Ombo", "is_active": "Y", "unit": "km"}, {"company": "Initech", "email": " OMALDDG@EXAMPLE.COM", "country": "   New Caledonia", "status": "PAUSED", "currency": "  Bob", "city": " Chorley ", "is_active": "0", "unit": "icnh"}, {"company": "  Globex", "email": " hekj@mail.com ", "country": "ZW", "status": "trial", "currency": "BOLIVIANO", "city": "Chorley", "is_active": "true", "unit": "ml"}, {"company": "   Hooli", "email": "hchb@corp.io ", "country": "zw", "status": " Paused ", "currency": "boliviano", "city": "  Kom Ombo", "is_active": "0", "unit": "ML"}, {"company": "Acme Inc", "email": "lignk@mail.com", "country": "ury", "status": "free  trial", "currency": "HTG", "city": "KOM OMBO", "is_active": "1", "unit": " \" "}, {"company": "  Stark Industries", "email": "   kegegi@corp.io", "country": "Eastern Republoc of Uruguay", "status": "trial ", "currency": "GOURDE", "city": "  guhi", "is_active": "Yes", "unit": "   km"}, {"company": "Globex", "email": "enjn@corp.io  ", "country": " nc ", "status": "trial", "currency": "gourde", "city": "Kom Ombo", "is_active": "Yes", "unit": "ml"}, {"company": "Soylent Corp", "email": "NOBMB@EXAMPLE.COM", "country": "ZW", "status": "cancelled", "currency": "BOB   ", "city": "Kom Ombo", "is_active": "FALSE", "unit": "kilometer  "}, {"company": "Wonka Co  ", "email": " NANOEBG@CORP.IO ", "country": " ury ", "status": " Active", "currency": "boliviano", "city": "Guxi", "is_active": "false", "unit": "\""}, {"company": "Hooli", "email": " JDCJO@EXAMPLE.COM", "country": "New Caledonia   ", "status": "cancelled", "currency": "HTG", "city": "Chorley", "is_active": "1", "unit": "IN"}, {"company": "Cyberdyne", "email": "mpdhhc@mail.com", "country": "  Eastern Republic Of Uruguay", "status": "trial", "currency": "bob", "city": "Rees", "is_active": "true", "unit": " Km "}, {"company": "Globex", "email": "DIIEP@EXAMPLE.COM", "country": "NC ", "status": " TIAL", "currency": "bob", "city": "kom ombo", "is_active": "Yes", "unit": "mls"}, {"company": "Soylent Corp", "email": " CANH@TEST.ORG ", "country": " ZWE ", "status": "Churned", "currency": "Boliviano", "city": "kom ombo", "is_active": "N", "unit": "inches"}, {"company": "Umbrella", "email": "MBHFA@CORP.IO", "country": "uruguay", "status": "active", "currency": "Boliviano", "city": "Chorley", "is_active": "false", "unit": " ml "}, {"company": "Initech", "email": "CGFHBN@TEST.ORG", "country": "ncl", "status": " active ", "currency": "GOURDE", "city": "Kom Ombo", "is_active": "F", "unit": "inches"}, {"company": "  Wonka Co", "email": "NNMK@TEST.ORG", "country": "zimbabwe", "status": "TRIAL", "currency": "   BOLIVIANO", "city": "guxi", "is_active": "1", "unit": "ml"}, {"company": "Globex", "email": "   DAMMB@TEST.ORG", "country": "Eastern Republic of Uruguay", "status": "Churnud", "currency": "Boliviano", "city": "rees", "is_active": "false", "unit": "\""}, {"company": "   Stark Industries", "email": "MEHG@EXAMPLE.COM", "country": "uruguay", "status": "TRIAL", "currency": "Gourde", "city": "Komm Ombo", "is_active": "Yes", "unit": " kilometers "}, {"company": "Stark Industries   ", "email": "eibhe@corp.io", "country": "UY", "status": "paused", "currency": "Gourde  ", "city": "Kom Ombo", "is_active": "true", "unit": "km "}, {"company": "", "email": "", "country": "", "status": "", "currency": "", "city": "", "is_active": "", "unit": ""}], "clean": [{"company": "Wonka Co", "email": "pienea@corp.io", "country": "Uruguay", "status": "Churned", "currency": "HTG", "city": "Guxi", "is_active": false, "unit": "mL"}, {"company": "Hooli", "email": "lodnfnh@example.com", "country": "Zimbabwe", "status": "Active", "currency": "BOB", "city": "Kom Ombo", "is_active": true, "unit": "km"}, {"company": "Wonka Co", "email": "afib@corp.io", "country": "Uruguay", "status": "Churned", "currency": "HTG", "city": "Kom Ombo", "is_active": true, "unit": "mL"}, {"company": "Hooli", "email": "abiij@test.org", "country": "Zimbabwe", "status": "Paused", "currency": "HTG", "city": "Rees", "is_active": true, "unit": "in"}, {"company": "Wonka Co", "email": "hckkc@test.org", "country": "New Caledonia", "status": "Active", "currency": "HTG", "city": "Rees", "is_active": false, "unit": "in"}, {"company": "Initech", "email": "bjefhlc@corp.io", "country": "Uruguay", "status": "Churned", "currency": "BOB", "city": "Rees", "is_active": false, "unit": "in"}, {"company": "Soylent Corp", "email": "iobdphf@corp.io", "country": "Zimbabwe", "status": "Paused", "currency": "HTG", "city": "Guxi", "is_active": false, "unit": "mL"}, {"company": "Acme Inc", "email": "pamle@example.com", "country": "New Caledonia", "status": "Paused", "currency": "BOB", "city": "Guxi", "is_active": true, "unit": "mL"}, {"company": "Initech", "email": "jhln@mail.com", "country": "Zimbabwe", "status": "Active", "currency": "HTG", "city": "Chorley", "is_active": true, "unit": "in"}, {"company": "Acme Inc", "email": "kjcai@corp.io", "country": "Uruguay", "status": "Trial", "currency": "BOB", "city": "Guxi", "is_active": false, "unit": "mL"}, {"company": "Initech", "email": "cblbp@example.com", "country": "Zimbabwe", "status": "Paused", "currency": "BOB", "city": "Guxi", "is_active": false, "unit": "km"}, {"company": "Hooli", "email": "elhggc@example.com", "country": "Uruguay", "status": "Paused", "currency": "BOB", "city": "Guxi", "is_active": true, "unit": "in"}, {"company": "Cyberdyne", "email": "dnlmfl@corp.io", "country": "Uruguay", "status": "Paused", "currency": "HTG", "city": "Chorley", "is_active": true, "unit": "km"}, {"company": "Soylent Corp", "email": "acbafpc@example.com", "country": "Zimbabwe", "status": "Paused", "currency": "HTG", "city": "Guxi", "is_active": false, "unit": "km"}, {"company": "Wonka Co", "email": "dkaeb@test.org", "country": "New Caledonia", "status": "Paused", "currency": "HTG", "city": "Chorley", "is_active": true, "unit": "in"}, {"company": "Acme Inc", "email": "cgdglln@test.org", "country": "New Caledonia", "status": "Trial", "currency": "HTG", "city": "Chorley", "is_active": false, "unit": "in"}, {"company": "Cyberdyne", "email": "dofnifm@test.org", "country": "New Caledonia", "status": "Trial", "currency": "BOB", "city": "Kom Ombo", "is_active": false, "unit": "in"}, {"company": "Vehement", "email": "bncmeeg@mail.com", "country": "Uruguay", "status": "Churned", "currency": "HTG", "city": "Chorley", "is_active": true, "unit": "mL"}, {"company": "Hooli", "email": "eakkak@example.com", "country": "Uruguay", "status": "Trial", "currency": "HTG", "city": "Rees", "is_active": true, "unit": "km"}, {"company": "Globex", "email": "mjkcnm@corp.io", "country": "Zimbabwe", "status": "Active", "currency": "HTG", "city": "Chorley", "is_active": true, "unit": "mL"}, {"company": "Umbrella", "email": "pclfil@example.com", "country": "Zimbabwe", "status": "Paused", "currency": "BOB", "city": "Chorley", "is_active": true, "unit": "in"}, {"company": "Globex", "email": "fgelhdm@mail.com", "country": "New Caledonia", "status": "Churned", "currency": "BOB", "city": "Guxi", "is_active": false, "unit": "mL"}, {"company": "Acme Inc", "email": "kogknhk@example.com", "country": "Zimbabwe", "status": "Churned", "currency": "BOB", "city": "Guxi", "is_active": true, "unit": "in"}, {"company": "Hooli", "email": "hgjmgh@corp.io", "country": "New Caledonia", "status": "Trial", "currency": "BOB", "city": "Chorley", "is_active": true, "unit": "in"}, {"company": "Acme Inc", "email": "pkijfpa@example.com", "country": "Uruguay", "status": "Churned", "currency": "HTG", "city": "Kom Ombo", "is_active": false, "unit": "km"}, {"company": "Hooli", "email": "bpdfkh@example.com", "country": "Uruguay", "status": "Trial", "currency": "HTG", "city": "Kom Ombo", "is_active": false, "unit": "mL"}, {"company": "Cyberdyne", "email": "andkd@mail.com", "country": "New Caledonia", "status": "Churned", "currency": "HTG", "city": "Guxi", "is_active": false, "unit": "in"}, {"company": "Hooli", "email": "afjc@mail.com", "country": "Uruguay", "status": "Paused", "currency": "HTG", "city": "Chorley", "is_active": false, "unit": "mL"}, {"company": "Soylent Corp", "email": "cckaedo@mail.com", "country": "New Caledonia", "status": "Paused", "currency": "BOB", "city": "Guxi", "is_active": false, "unit": "km"}, {"company": "Stark Industries", "email": "djoolah@mail.com", "country": "Zimbabwe", "status": "Paused", "currency": "BOB", "city": "Kom Ombo", "is_active": true, "unit": "mL"}, {"company": "Globex", "email": "jecjo@mail.com", "country": "New Caledonia", "status": "Churned", "currency": "BOB", "city": "Rees", "is_active": true, "unit": "in"}, {"company": "Umbrella", "email": "aafgkk@mail.com", "country": "Uruguay", "status": "Trial", "currency": "BOB", "city": "Kom Ombo", "is_active": false, "unit": "in"}, {"company": "Acme Inc", "email": "ecpkok@mail.com", "country": "New Caledonia", "status": "Paused", "currency": "HTG", "city": "Kom Ombo", "is_active": false, "unit": "in"}, {"company": "Wonka Co", "email": "cgdnfab@test.org", "country": "New Caledonia", "status": "Churned", "currency": "BOB", "city": "Chorley", "is_active": false, "unit": "in"}, {"company": "Cyberdyne", "email": "bmjpn@example.com", "country": "Uruguay", "status": "Churned", "currency": "BOB", "city": "Rees", "is_active": false, "unit": "mL"}, {"company": "Soylent Corp", "email": "ckpjff@example.com", "country": "Uruguay", "status": "Paused", "currency": "BOB", "city": "Guxi", "is_active": true, "unit": "mL"}, {"company": "Wonka Co", "email": "phkda@example.com", "country": "New Caledonia", "status": "Paused", "currency": "HTG", "city": "Chorley", "is_active": false, "unit": "in"}, {"company": "Stark Industries", "email": "fkbj@test.org", "country": "Zimbabwe", "status": "Active", "currency": "HTG", "city": "Guxi", "is_active": false, "unit": "km"}, {"company": "Cyberdyne", "email": "bcebdj@corp.io", "country": "New Caledonia", "status": "Churned", "currency": "BOB", "city": "Chorley", "is_active": true, "unit": "mL"}, {"company": "Soylent Corp", "email": "ehba@example.com", "country": "Uruguay", "status": "Churned", "currency": "HTG", "city": "Guxi", "is_active": false, "unit": "in"}, {"company": "Globex", "email": "aclggnj@mail.com", "country": "Uruguay", "status": "Paused", "currency": "HTG", "city": "Rees", "is_active": false, "unit": "mL"}, {"company": "Stark Industries", "email": "hjkhc@example.com", "country": "Uruguay", "status": "Active", "currency": "HTG", "city": "Kom Ombo", "is_active": true, "unit": "km"}, {"company": "Initech", "email": "omalddg@example.com", "country": "New Caledonia", "status": "Paused", "currency": "BOB", "city": "Chorley", "is_active": false, "unit": "in"}, {"company": "Globex", "email": "hekj@mail.com", "country": "Zimbabwe", "status": "Trial", "currency": "BOB", "city": "Chorley", "is_active": true, "unit": "mL"}, {"company": "Hooli", "email": "hchb@corp.io", "country": "Zimbabwe", "status": "Paused", "currency": "BOB", "city": "Kom Ombo", "is_active": false, "unit": "mL"}, {"company": "Acme Inc", "email": "lignk@mail.com", "country": "Uruguay", "status": "Trial", "currency": "HTG", "city": "Kom Ombo", "is_active": true, "unit": "in"}, {"company": "Stark Industries", "email": "kegegi@corp.io", "country": "Uruguay", "status": "Trial", "currency": "HTG", "city": "Guxi", "is_active": true, "unit": "km"}, {"company": "Globex", "email": "enjn@corp.io", "country": "New Caledonia", "status": "Trial", "currency": "HTG", "city": "Kom Ombo", "is_active": true, "unit": "mL"}, {"company": "Soylent Corp", "email": "nobmb@example.com", "country": "Zimbabwe", "status": "Churned", "currency": "BOB", "city": "Kom Ombo", "is_active": false, "unit": "km"}, {"company": "Wonka Co", "email": "nanoebg@corp.io", "country": "Uruguay", "status": "Active", "currency": "BOB", "city": "Guxi", "is_active": false, "unit": "in"}, {"company": "Hooli", "email": "jdcjo@example.com", "country": "New Caledonia", "status": "Churned", "currency": "HTG", "city": "Chorley", "is_active": true, "unit": "in"}, {"company": "Cyberdyne", "email": "mpdhhc@mail.com", "country": "Uruguay", "status": "Trial", "currency": "BOB", "city": "Rees", "is_active": true, "unit": "km"}, {"company": "Globex", "email": "diiep@example.com", "country": "New Caledonia", "status": "Trial", "currency": "BOB", "city": "Kom Ombo", "is_active": true, "unit": "mL"}, {"company": "Soylent Corp", "email": "canh@test.org", "country": "Zimbabwe", "status": "Churned", "currency": "BOB", "city": "Kom Ombo", "is_active": false, "unit": "in"}, {"company": "Umbrella", "email": "mbhfa@corp.io", "country": "Uruguay", "status": "Active", "currency": "BOB", "city": "Chorley", "is_active": false, "unit": "mL"}, {"company": "Initech", "email": "cgfhbn@test.org", "country": "New Caledonia", "status": "Active", "currency": "HTG", "city": "Kom Ombo", "is_active": false, "unit": "in"}, {"company": "Wonka Co", "email": "nnmk@test.org", "country": "Zimbabwe", "status": "Trial", "currency": "BOB", "city": "Guxi", "is_active": true, "unit": "mL"}, {"company": "Globex", "email": "dammb@test.org", "country": "Uruguay", "status": "Churned", "currency": "BOB", "city": "Rees", "is_active": false, "unit": "in"}, {"company": "Stark Industries", "email": "mehg@example.com", "country": "Uruguay", "status": "Trial", "currency": "HTG", "city": "Kom Ombo", "is_active": true, "unit": "km"}, {"company": "Stark Industries", "email": "eibhe@corp.io", "country": "Uruguay", "status": "Paused", "currency": "HTG", "city": "Kom Ombo", "is_active": true, "unit": "km"}], "dirty_cols": ["company", "email", "country", "status", "currency", "city", "is_active", "unit"], "clean_cols": ["company", "email", "country", "status", "currency", "city", "is_active", "unit"], "plan": {"dataset_summary": "61 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"EASTERN REPUBLIC OF URUGUYA": "Uruguay", "Zw": "Zimbabwe", "ury": "Uruguay", "ZW": "Zimbabwe", "nc": "New Caledonia", "URY": "Uruguay", "ZWE": "Zimbabwe", "zwe": "Zimbabwe", "Zwe": "Zimbabwe", "uy": "Uruguay", "NCL": "New Caledonia", "New Caledonnia": "New Caledonia", "Eastern Republic Of Urusuay": "Uruguay", "Eastern Republic of Uruguay": "Uruguay", "zimbabwe": "Zimbabwe", "new caledonia": "New Caledonia", "EASTERN REPUBLIC OF URUUGUAY": "Uruguay", "NEW CALEDONIA": "New Caledonia", "EASTERN REPUBLIC OF URUGUAY": "Uruguay", "NC": "New Caledonia", "ncl": "New Caledonia", "Eastern Republic Of Uruguay": "Uruguay", "eastern republic of uruguay": "Uruguay", "uruguay": "Uruguay", "zw": "Zimbabwe", "Eastern Republoc of Uruguay": "Uruguay", "UY": "Uruguay"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"canceled": "Churned", "active": "Active", "churned": "Churned", "on hold": "Paused", "On Hold": "Paused", "paused": "Paused", "trial": "Trial", "pauhed": "Paused", "PAUSED": "Paused", "FREE TRIAL": "Trial", "TRIIAL": "Trial", "Trihl": "Trial", "ACTVIE": "Active", "TRIAL": "Trial", "free trial": "Trial", "cancelled": "Churned", "ACTIVE": "Active", "pased": "Paused", "ATIVE": "Active", "TIAL": "Trial", "Churnud": "Churned"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"htg": "HTG", "Bob": "BOB", "GOURDE": "HTG", "Gourde": "HTG", "Boliviano": "BOB", "bob": "BOB", "boliviano": "BOB", "Htg": "HTG", "gourde": "HTG", "BOLIVIANO": "BOB"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kom ombo": "Kom Ombo", "rees": "Rees", "GUXI": "Guxi", "Gui": "Guxi", "chorley": "Chorley", "CHORLEY": "Chorley", "guxi": "Guxi", "Choorley": "Chorley", "KOM OMBO": "Kom Ombo", "Gxui": "Guxi", "guhi": "Guxi", "Komm Ombo": "Kom Ombo"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mls": "mL", "IN": "in", "inch": "in", "ml": "mL", "\"": "in", "MILLLIITER": "mL", "In": "in", "kilometrs": "km", "kilometers": "km", "inches": "in", "INCHES": "in", "milliliter": "mL", "Inch": "in", "kilometer": "km", "ML": "mL", "millilter": "mL", "icnh": "in", "Km": "km"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "5963548353", "state": "LA", "currency": "BTN", "rate": "19.3%", "company": "Globex", "extra": ""}, {"phone": "521-924-1825", "state": "LA", "currency": "gel", "rate": "61.1%", "company": "Acme Inc", "extra": ""}, {"phone": "5273003117", "state": "UT", "currency": "Ngultrum", "rate": "11.1%", "company": "   Acme Inc", "extra": ""}, {"phone": "542.201.8624", "state": "utlh", "currency": "ngultrum", "rate": "90.6%", "company": "   Globex", "extra": ""}, {"phone": "593-951-6526", "state": "la", "currency": "ngultrum", "rate": "46.9%", "company": "   Globex", "extra": ""}, {"phone": "586.701.9633", "state": "ut", "currency": "Dalasi", "rate": "13.3%", "company": "Umbrella ", "extra": ""}, {"phone": "548.022.7287", "state": "Utah", "currency": "GEL", "rate": "80.1%", "company": "Initech", "extra": ""}, {"phone": "(533)425-5895", "state": " Louisiana ", "currency": "Dalasi   ", "rate": "48.5%", "company": "Stark Industries   ", "extra": ""}, {"phone": "503.466.1858", "state": "UT", "currency": "BTN", "rate": "53.6%", "company": "Umbrella", "extra": ""}, {"phone": "559-087-5325", "state": " ut ", "currency": "BTN", "rate": "47.9%", "company": "  Stark Industries", "extra": ""}, {"phone": "5464268117", "state": "LOUISIANA", "currency": "Dalasi", "rate": "73.4%", "company": "Soylent Corp", "extra": ""}, {"phone": "520-038-2710", "state": "Utah", "currency": " BTN ", "rate": "2.0%", "company": "Umbrella  ", "extra": ""}, {"phone": "(544)910-2070", "state": "LOUISIANA", "currency": "BTN", "rate": "29.1%", "company": "Umbrella", "extra": ""}, {"phone": "(577)461-7309", "state": "  LOUISIANA", "currency": "lari", "rate": "12.9%", "company": "Wonka Co  ", "extra": ""}, {"phone": "529-584-2017", "state": "LA", "currency": "BTN", "rate": "41.1%", "company": "Cyberdyne", "extra": ""}, {"phone": "(578)130-0523", "state": "Uttah", "currency": "Ngultrum", "rate": "98.8%", "company": " Stark Industries", "extra": ""}, {"phone": "5840407939", "state": "LA", "currency": "GEL", "rate": "47.3%", "company": "Umbrella  ", "extra": ""}, {"phone": "5735514923", "state": " UTAH ", "currency": "  Ngultrum", "rate": "29.6%", "company": "Cyberdyne", "extra": ""}, {"phone": "(542)994-3421", "state": "ut", "currency": "BTN", "rate": "42.2%", "company": "Wonka Co  ", "extra": ""}, {"phone": "(554)977-1333", "state": "UT", "currency": "GEL", "rate": "14.8%", "company": "Initech", "extra": ""}, {"phone": "5025388979", "state": " utah ", "currency": "GEL", "rate": "24.8%", "company": "Wonka Co ", "extra": ""}, {"phone": "575-615-9109", "state": "LA ", "currency": "Dalasi", "rate": "43.7%", "company": "Vehement", "extra": ""}, {"phone": "5711125268", "state": " ut", "currency": "gmd", "rate": "30.8%", "company": "Initech  ", "extra": ""}, {"phone": "500.937.8587", "state": "Utah", "currency": "GMD", "rate": "97.2%", "company": "Wonka Co", "extra": ""}, {"phone": "520-070-4623", "state": "   Louisiana", "currency": " Dalasi", "rate": "68.1%", "company": "Cyberdyne", "extra": ""}, {"phone": "5963548353", "state": "LA", "currency": "BTN", "rate": "19.3%", "company": "Globex", "extra": ""}, {"phone": "586.701.9633", "state": "ut", "currency": "Dalasi", "rate": "13.3%", "company": "Umbrella ", "extra": ""}], "clean": [{"phone": "(596) 354-8353", "state": "Louisiana", "currency": "BTN", "rate": 0.193, "company": "Globex"}, {"phone": "(521) 924-1825", "state": "Louisiana", "currency": "GEL", "rate": 0.611, "company": "Acme Inc"}, {"phone": "(527) 300-3117", "state": "Utah", "currency": "BTN", "rate": 0.111, "company": "Acme Inc"}, {"phone": "(542) 201-8624", "state": "Utah", "currency": "BTN", "rate": 0.9059999999999999, "company": "Globex"}, {"phone": "(593) 951-6526", "state": "Louisiana", "currency": "BTN", "rate": 0.469, "company": "Globex"}, {"phone": "(586) 701-9633", "state": "Utah", "currency": "GMD", "rate": 0.133, "company": "Umbrella"}, {"phone": "(548) 022-7287", "state": "Utah", "currency": "GEL", "rate": 0.8009999999999999, "company": "Initech"}, {"phone": "(533) 425-5895", "state": "Louisiana", "currency": "GMD", "rate": 0.485, "company": "Stark Industries"}, {"phone": "(503) 466-1858", "state": "Utah", "currency": "BTN", "rate": 0.536, "company": "Umbrella"}, {"phone": "(559) 087-5325", "state": "Utah", "currency": "BTN", "rate": 0.479, "company": "Stark Industries"}, {"phone": "(546) 426-8117", "state": "Louisiana", "currency": "GMD", "rate": 0.7340000000000001, "company": "Soylent Corp"}, {"phone": "(520) 038-2710", "state": "Utah", "currency": "BTN", "rate": 0.02, "company": "Umbrella"}, {"phone": "(544) 910-2070", "state": "Louisiana", "currency": "BTN", "rate": 0.29100000000000004, "company": "Umbrella"}, {"phone": "(577) 461-7309", "state": "Louisiana", "currency": "GEL", "rate": 0.129, "company": "Wonka Co"}, {"phone": "(529) 584-2017", "state": "Louisiana", "currency": "BTN", "rate": 0.41100000000000003, "company": "Cyberdyne"}, {"phone": "(578) 130-0523", "state": "Utah", "currency": "BTN", "rate": 0.988, "company": "Stark Industries"}, {"phone": "(584) 040-7939", "state": "Louisiana", "currency": "GEL", "rate": 0.473, "company": "Umbrella"}, {"phone": "(573) 551-4923", "state": "Utah", "currency": "BTN", "rate": 0.29600000000000004, "company": "Cyberdyne"}, {"phone": "(542) 994-3421", "state": "Utah", "currency": "BTN", "rate": 0.42200000000000004, "company": "Wonka Co"}, {"phone": "(554) 977-1333", "state": "Utah", "currency": "GEL", "rate": 0.14800000000000002, "company": "Initech"}, {"phone": "(502) 538-8979", "state": "Utah", "currency": "GEL", "rate": 0.248, "company": "Wonka Co"}, {"phone": "(575) 615-9109", "state": "Louisiana", "currency": "GMD", "rate": 0.43700000000000006, "company": "Vehement"}, {"phone": "(571) 112-5268", "state": "Utah", "currency": "GMD", "rate": 0.308, "company": "Initech"}, {"phone": "(500) 937-8587", "state": "Utah", "currency": "GMD", "rate": 0.972, "company": "Wonka Co"}, {"phone": "(520) 070-4623", "state": "Louisiana", "currency": "GMD", "rate": 0.6809999999999999, "company": "Cyberdyne"}], "dirty_cols": ["phone", "state", "currency", "rate", "company", "extra"], "clean_cols": ["phone", "state", "currency", "rate", "company"], "plan": {"dataset_summary": "27 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LA": "Louisiana", "UT": "Utah", "utlh": "Utah", "la": "Louisiana", "ut": "Utah", "LOUISIANA": "Louisiana", "Uttah": "Utah", "UTAH": "Utah", "utah": "Utah"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"gel": "GEL", "Ngultrum": "BTN", "ngultrum": "BTN", "Dalasi": "GMD", "lari": "GEL", "gmd": "GMD"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"name": "David Park", "state": "Um", "is_active": "N", "amount": "$4,648.75", "rate": "48.4%", "currency": "Kwanza", "department": " HR ", "country": "Benin"}, {"name": "Lena Park", "state": "WY", "is_active": "No", "amount": "$6,742.27", "rate": "71.4%", "currency": "aoa", "department": "Biz Dev", "country": "Viet Nam"}, {"name": "Carol Petrov", "state": "Pennsylvania", "is_active": "F", "amount": "$5,623.73", "rate": "89.2%", "currency": "CVE", "department": " people ", "country": "benin"}, {"name": "Sara  Ali", "state": "UM  ", "is_active": "F", "amount": "3.798,18", "rate": "75.9%", "currency": "AOA", "department": "people", "country": "Socialist Republic of Viet Nam"}, {"name": "Bob Petrov", "state": "United States Minor Outlying Islands", "is_active": "TRUE", "amount": "8.023,26", "rate": "32.1%", "currency": "AOA", "department": " Hr ", "country": "Socialist Republic of Viet Nam"}, {"name": "Grace Park   ", "state": "Kentucky", "is_active": "0", "amount": "3.686,98", "rate": "73.3%", "currency": "Cve", "department": "Biz Dev", "country": " ESH"}, {"name": "Bob Park   ", "state": "UM", "is_active": "Y", "amount": "1.916,11", "rate": "51.7%", "currency": "cabo verde escudo", "department": "   Hr", "country": "Esh"}, {"name": "Frank Adams", "state": "WY", "is_active": "N", "amount": "$266.73", "rate": "96.3%", "currency": "  Kwanza", "department": "people", "country": "ESH"}, {"name": "Priya Reyes", "state": "TEXAS", "is_active": "No", "amount": "2.998,64", "rate": "64.7%", "currency": "CVE", "department": "people ops", "country": "BEN"}, {"name": "   Sara Novak", "state": "pennsylvania", "is_active": "true", "amount": "$2,303.15", "rate": "60.4%", "currency": "kwanza", "department": "PEOPLE OPS", "country": "western sahara"}, {"name": "Mona  Adams", "state": " tx ", "is_active": "TRUE", "amount": "$737.22", "rate": "68.3%", "currency": "kwanza", "department": " sales ", "country": "sx"}, {"name": "Ivan Johnson", "state": "united states minor outlying islands  ", "is_active": "F", "amount": "$2,691.33", "rate": "2.6%", "currency": "AOA", "department": "People", "country": "KGZ"}, {"name": "Grace  Reyes", "state": "Wyoming", "is_active": "T", "amount": "$7,831.71", "rate": "8.9%", "currency": "CVE", "department": "Sales", "country": "Republic of Benin  "}, {"name": "Karl Khan ", "state": "Wyoming", "is_active": "1", "amount": "None", "rate": "72.1%", "currency": "Cabo Verde Escudo ", "department": "people ops", "country": "BENIN  "}, {"name": "David Wong", "state": "TX   ", "is_active": "Y", "amount": "5.542,77", "rate": "51.8%", "currency": "AOA", "department": "Biz  Dev", "country": "VNM"}, {"name": "Bob Wong", "state": "texas", "is_active": "No", "amount": "$2,361.91", "rate": "72.3%", "currency": "AOA", "department": "   PEOPL EOPS", "country": "ESH"}, {"name": "Sara Brandt", "state": "TX", "is_active": "FALSE", "amount": "-", "rate": "50.5%", "currency": "AOA", "department": "   Human RResources", "country": "ben"}, {"name": "   Eve Fischer", "state": "wy", "is_active": "true", "amount": "8.792,46", "rate": "31.5%", "currency": "AOA", "department": "BIZ DEV", "country": "kg"}, {"name": " David Reyes", "state": " WY ", "is_active": "true", "amount": "$689.01", "rate": "27.3%", "currency": "KWNAZA", "department": "Sales", "country": "Western Sahara"}, {"name": "Priya Lee", "state": "WY", "is_active": "F", "amount": "1.837,61", "rate": "74.5%", "currency": "Cabo Verde Escudo", "department": "HUMAN RESOURCES", "country": "VN"}, {"name": "Ivan Novak", "state": "texas", "is_active": "Y", "amount": "TBD", "rate": "29.3%", "currency": "CVE", "department": "Biz Dev", "country": "Sx  "}, {"name": "Grace Diaz ", "state": "UM", "is_active": "true", "amount": "$5,189.51", "rate": "2.6%", "currency": "CVE", "department": " People ", "country": "kg"}, {"name": "Eve  Brandt", "state": "TX", "is_active": "TRUE", "amount": "1.496,88", "rate": "26.2%", "currency": "Cabo Verde Escudo", "department": "People Ops", "country": "vnm "}, {"name": "Mona Fischer ", "state": "tx", "is_active": "N", "amount": "$5,867.50", "rate": "95.3%", "currency": " Caabo Verde Escudo", "department": "   Biz Dev", "country": "KGZ"}, {"name": "Sara  Fischer", "state": "Kentucky", "is_active": "TRUE", "amount": "1.794,37", "rate": "31.1%", "currency": "CVE", "department": "Biz Dbv", "country": "SX  "}, {"name": "Lena Khan   ", "state": "KENTUCKY ", "is_active": "F", "amount": "$2,517.68", "rate": "90.5%", "currency": "cabo verde escudo", "department": "BIZ DEV", "country": "eh"}, {"name": "Alice Park", "state": "Wyomig", "is_active": "FALSE", "amount": "7.399,85", "rate": "16.3%", "currency": "aoa", "department": "PEOPLE OPS", "country": "EH  "}, {"name": "  Bob Adams", "state": "PA", "is_active": "false", "amount": "?", "rate": "46.6%", "currency": "Kwanza", "department": " People ", "country": "SXM"}, {"name": "Alice Wong", "state": "  Pennsylvania", "is_active": "FALSE", "amount": "286,33", "rate": "98.6%", "currency": "Cabbo  Verde Escudo", "department": "biz dev", "country": "Esh"}, {"name": "  Priya Novak", "state": "PA", "is_active": "true", "amount": "7.575,55", "rate": "51.9%", "currency": "  AOA", "department": "  BIZ  DEV", "country": "VN"}, {"name": "Alice  Lee", "state": "kentucy", "is_active": "0", "amount": "4.954,17", "rate": "35.7%", "currency": "cabo verde escudo", "department": " Biz Dev", "country": "REPUBLIC OF  BENIN"}, {"name": "Lena Brandt", "state": " um ", "is_active": "0", "amount": "$8,918.96", "rate": "49.2%", "currency": "kwanza", "department": "Human Resources", "country": "Vist Nam"}, {"name": "   Eve Adams", "state": "um", "is_active": "false", "amount": "7.183,94", "rate": "35.0%", "currency": "AOA", "department": "  Sales", "country": " SOCIALIST REPUBLIC OF VIET NAM"}, {"name": "Judy Cruz", "state": "WY", "is_active": "true", "amount": "$1,085.25", "rate": "67.1%", "currency": "Cabo Verde Escudo", "department": "Sales", "country": "EH"}, {"name": "Frank  Reyes", "state": "WY", "is_active": "0", "amount": "5.973,65", "rate": "42.9%", "currency": "AOA", "department": "People", "country": "Republic ofmBenin"}, {"name": "Lena Moore", "state": "  United States Minor Outlying Islands", "is_active": "N", "amount": "None", "rate": "57.8%", "currency": "kwanza", "department": "people ops", "country": "Esh"}, {"name": " Judy Lee", "state": "United States Minor Outlying Islands", "is_active": "0", "amount": "5.320,01", "rate": "65.1%", "currency": "Aoa", "department": "HR", "country": "western sahara"}, {"name": "Mona Moore  ", "state": " wyoming ", "is_active": "FALSE", "amount": "$7,602.16", "rate": "26.4%", "currency": "CABO VERDE ESCUDO", "department": "Sales", "country": "Socialist Republic of Viet Nam"}, {"name": "Priya Wong", "state": "PA", "is_active": "true", "amount": "4.470,48", "rate": "90.1%", "currency": "CVE", "department": "People", "country": "kyrgyzstan"}, {"name": "Karl  Petrov", "state": " Pa ", "is_active": "FALSE", "amount": "na", "rate": "86.5%", "currency": "AOA", "department": "People", "country": "kyrgyzstan"}, {"name": "Heidi Diaz", "state": " kentucky ", "is_active": "Yes", "amount": "4.926,04", "rate": "73.3%", "currency": "kwanza ", "department": "BIZ DEV", "country": "Socialist Republic of Viet Nam   "}, {"name": "Grace Adams", "state": "Ky", "is_active": "No", "amount": "$2,481.80", "rate": "80.9%", "currency": "AOA", "department": "biz dev", "country": "sxm"}, {"name": "David Diaz", "state": " PENNSYLVANIA ", "is_active": "TRUE", "amount": "6.597,44", "rate": "88.7%", "currency": "AOA", "department": "Sales", "country": "sx"}, {"name": "Mona  Novak", "state": "Pennsylvania", "is_active": "FALSE", "amount": "5.487,27", "rate": "9.2%", "currency": "KWANZA", "department": "People OOps", "country": "esh"}, {"name": "Heidi  Brandt", "state": "Kentucky", "is_active": "No", "amount": "4.018,82", "rate": "25.3%", "currency": "AOA", "department": " people ops", "country": "eh"}, {"name": "David Novak ", "state": "UM ", "is_active": "Yes", "amount": "3.250,97", "rate": "57.9%", "currency": "   CVE", "department": "salse", "country": "Kg"}, {"name": "Bob Lee", "state": "TEAS", "is_active": "Yes", "amount": "2.953,19", "rate": "77.6%", "currency": " CVE ", "department": "Bi zDev", "country": " SXM "}, {"name": "Lena Lee  ", "state": "tx", "is_active": "1", "amount": "$6,534.06", "rate": "31.3%", "currency": "Kwanza", "department": "Human Resources", "country": " SX"}, {"name": "Ivan Fischer ", "state": "WY", "is_active": "true", "amount": "$6,498.30", "rate": "33.4%", "currency": "Kwanza", "department": "Biz  Dev", "country": " sx"}, {"name": "David Johnson", "state": "Wyoming", "is_active": "T", "amount": "$8,460.10", "rate": "23.0%", "currency": " Cve ", "department": "  Sales", "country": "KGZ"}, {"name": "Ivan Diaz", "state": "texas", "is_active": "1", "amount": "$920.93", "rate": "71.4%", "currency": "aoa  ", "department": "Sales", "country": "VN"}, {"name": "Eve Park ", "state": " TX ", "is_active": "false", "amount": "$5,364.62", "rate": "40.1%", "currency": "kwanza", "department": "Peoplp Ops", "country": "SINT MAARTEN (DUTCH PART)"}, {"name": "Grace Ali", "state": "united sttates minor outlying islands", "is_active": "true", "amount": "0", "rate": "80.7%", "currency": " Aoa ", "department": "People", "country": "   ESH"}, {"name": "  Mona Wong", "state": "Wyoming", "is_active": "0", "amount": "6.317,04", "rate": "45.0%", "currency": "Kwana", "department": "Pemple", "country": "Western Sahara"}, {"name": " Lena Khan", "state": "Kentucky", "is_active": "false", "amount": "6.657,55", "rate": "8.2%", "currency": "Kwaza", "department": "BIZ DEV", "country": "  esh"}, {"name": "Bob  Cruz", "state": " Wy ", "is_active": "No", "amount": "5.613,77", "rate": "95.9%", "currency": "Cabo Verde Escudo", "department": "Bi zDev", "country": "Sint Maarten (Dutch part)"}, {"name": "Alice Cruz  ", "state": "WYOMING", "is_active": "1", "amount": "None", "rate": "74.4%", "currency": "Kwanza", "department": "   hr", "country": "   Vn"}, {"name": "   Heidi Moore", "state": "ky", "is_active": "T", "amount": "$7,688.85", "rate": "16.1%", "currency": "Kwanza", "department": "Biz Dev", "country": "Sxm"}, {"name": "", "state": "", "is_active": "", "amount": "", "rate": "", "currency": "", "department": "", "country": ""}], "clean": [{"name": "David Park", "state": "United States Minor Outlying Islands", "is_active": false, "amount": 4648.75, "rate": 0.484, "currency": "AOA", "department": "Human Resources", "country": "Benin"}, {"name": "Lena Park", "state": "Wyoming", "is_active": false, "amount": 6742.27, "rate": 0.7140000000000001, "currency": "AOA", "department": "Sales", "country": "Viet Nam"}, {"name": "Carol Petrov", "state": "Pennsylvania", "is_active": false, "amount": 5623.73, "rate": 0.892, "currency": "CVE", "department": "Human Resources", "country": "Benin"}, {"name": "Sara Ali", "state": "United States Minor Outlying Islands", "is_active": false, "amount": 3798.18, "rate": 0.759, "currency": "AOA", "department": "Human Resources", "country": "Viet Nam"}, {"name": "Bob Petrov", "state": "United States Minor Outlying Islands", "is_active": true, "amount": 8023.26, "rate": 0.321, "currency": "AOA", "department": "Human Resources", "country": "Viet Nam"}, {"name": "Grace Park", "state": "Kentucky", "is_active": false, "amount": 3686.98, "rate": 0.733, "currency": "CVE", "department": "Sales", "country": "Western Sahara"}, {"name": "Bob Park", "state": "United States Minor Outlying Islands", "is_active": true, "amount": 1916.11, "rate": 0.517, "currency": "CVE", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Frank Adams", "state": "Wyoming", "is_active": false, "amount": 266.73, "rate": 0.963, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Priya Reyes", "state": "Texas", "is_active": false, "amount": 2998.64, "rate": 0.647, "currency": "CVE", "department": "Human Resources", "country": "Benin"}, {"name": "Sara Novak", "state": "Pennsylvania", "is_active": true, "amount": 2303.15, "rate": 0.604, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Mona Adams", "state": "Texas", "is_active": true, "amount": 737.22, "rate": 0.6829999999999999, "currency": "AOA", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "Ivan Johnson", "state": "United States Minor Outlying Islands", "is_active": false, "amount": 2691.33, "rate": 0.026000000000000002, "currency": "AOA", "department": "Human Resources", "country": "Kyrgyzstan"}, {"name": "Grace Reyes", "state": "Wyoming", "is_active": true, "amount": 7831.71, "rate": 0.08900000000000001, "currency": "CVE", "department": "Sales", "country": "Benin"}, {"name": "Karl Khan", "state": "Wyoming", "is_active": true, "amount": NaN, "rate": 0.721, "currency": "CVE", "department": "Human Resources", "country": "Benin"}, {"name": "David Wong", "state": "Texas", "is_active": true, "amount": 5542.77, "rate": 0.518, "currency": "AOA", "department": "Sales", "country": "Viet Nam"}, {"name": "Bob Wong", "state": "Texas", "is_active": false, "amount": 2361.91, "rate": 0.723, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Sara Brandt", "state": "Texas", "is_active": false, "amount": NaN, "rate": 0.505, "currency": "AOA", "department": "Human Resources", "country": "Benin"}, {"name": "Eve Fischer", "state": "Wyoming", "is_active": true, "amount": 8792.46, "rate": 0.315, "currency": "AOA", "department": "Sales", "country": "Kyrgyzstan"}, {"name": "David Reyes", "state": "Wyoming", "is_active": true, "amount": 689.01, "rate": 0.273, "currency": "AOA", "department": "Sales", "country": "Western Sahara"}, {"name": "Priya Lee", "state": "Wyoming", "is_active": false, "amount": 1837.61, "rate": 0.745, "currency": "CVE", "department": "Human Resources", "country": "Viet Nam"}, {"name": "Ivan Novak", "state": "Texas", "is_active": true, "amount": NaN, "rate": 0.293, "currency": "CVE", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "Grace Diaz", "state": "United States Minor Outlying Islands", "is_active": true, "amount": 5189.51, "rate": 0.026000000000000002, "currency": "CVE", "department": "Human Resources", "country": "Kyrgyzstan"}, {"name": "Eve Brandt", "state": "Texas", "is_active": true, "amount": 1496.88, "rate": 0.262, "currency": "CVE", "department": "Human Resources", "country": "Viet Nam"}, {"name": "Mona Fischer", "state": "Texas", "is_active": false, "amount": 5867.5, "rate": 0.953, "currency": "CVE", "department": "Sales", "country": "Kyrgyzstan"}, {"name": "Sara Fischer", "state": "Kentucky", "is_active": true, "amount": 1794.37, "rate": 0.311, "currency": "CVE", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "Lena Khan", "state": "Kentucky", "is_active": false, "amount": 2517.68, "rate": 0.905, "currency": "CVE", "department": "Sales", "country": "Western Sahara"}, {"name": "Alice Park", "state": "Wyoming", "is_active": false, "amount": 7399.85, "rate": 0.163, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Bob Adams", "state": "Pennsylvania", "is_active": false, "amount": NaN, "rate": 0.466, "currency": "AOA", "department": "Human Resources", "country": "Sint Maarten (Dutch part)"}, {"name": "Alice Wong", "state": "Pennsylvania", "is_active": false, "amount": 286.33, "rate": 0.986, "currency": "CVE", "department": "Sales", "country": "Western Sahara"}, {"name": "Priya Novak", "state": "Pennsylvania", "is_active": true, "amount": 7575.55, "rate": 0.519, "currency": "AOA", "department": "Sales", "country": "Viet Nam"}, {"name": "Alice Lee", "state": "Kentucky", "is_active": false, "amount": 4954.17, "rate": 0.35700000000000004, "currency": "CVE", "department": "Sales", "country": "Benin"}, {"name": "Lena Brandt", "state": "United States Minor Outlying Islands", "is_active": false, "amount": 8918.96, "rate": 0.49200000000000005, "currency": "AOA", "department": "Human Resources", "country": "Viet Nam"}, {"name": "Eve Adams", "state": "United States Minor Outlying Islands", "is_active": false, "amount": 7183.94, "rate": 0.35, "currency": "AOA", "department": "Sales", "country": "Viet Nam"}, {"name": "Judy Cruz", "state": "Wyoming", "is_active": true, "amount": 1085.25, "rate": 0.6709999999999999, "currency": "CVE", "department": "Sales", "country": "Western Sahara"}, {"name": "Frank Reyes", "state": "Wyoming", "is_active": false, "amount": 5973.65, "rate": 0.429, "currency": "AOA", "department": "Human Resources", "country": "Benin"}, {"name": "Lena Moore", "state": "United States Minor Outlying Islands", "is_active": false, "amount": NaN, "rate": 0.578, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Judy Lee", "state": "United States Minor Outlying Islands", "is_active": false, "amount": 5320.01, "rate": 0.6509999999999999, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Mona Moore", "state": "Wyoming", "is_active": false, "amount": 7602.16, "rate": 0.264, "currency": "CVE", "department": "Sales", "country": "Viet Nam"}, {"name": "Priya Wong", "state": "Pennsylvania", "is_active": true, "amount": 4470.48, "rate": 0.9009999999999999, "currency": "CVE", "department": "Human Resources", "country": "Kyrgyzstan"}, {"name": "Karl Petrov", "state": "Pennsylvania", "is_active": false, "amount": NaN, "rate": 0.865, "currency": "AOA", "department": "Human Resources", "country": "Kyrgyzstan"}, {"name": "Heidi Diaz", "state": "Kentucky", "is_active": true, "amount": 4926.04, "rate": 0.733, "currency": "AOA", "department": "Sales", "country": "Viet Nam"}, {"name": "Grace Adams", "state": "Kentucky", "is_active": false, "amount": 2481.8, "rate": 0.809, "currency": "AOA", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "David Diaz", "state": "Pennsylvania", "is_active": true, "amount": 6597.44, "rate": 0.887, "currency": "AOA", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "Mona Novak", "state": "Pennsylvania", "is_active": false, "amount": 5487.27, "rate": 0.092, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Heidi Brandt", "state": "Kentucky", "is_active": false, "amount": 4018.82, "rate": 0.253, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "David Novak", "state": "United States Minor Outlying Islands", "is_active": true, "amount": 3250.97, "rate": 0.579, "currency": "CVE", "department": "Sales", "country": "Kyrgyzstan"}, {"name": "Bob Lee", "state": "Texas", "is_active": true, "amount": 2953.19, "rate": 0.7759999999999999, "currency": "CVE", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "Lena Lee", "state": "Texas", "is_active": true, "amount": 6534.06, "rate": 0.313, "currency": "AOA", "department": "Human Resources", "country": "Sint Maarten (Dutch part)"}, {"name": "Ivan Fischer", "state": "Wyoming", "is_active": true, "amount": 6498.3, "rate": 0.33399999999999996, "currency": "AOA", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "David Johnson", "state": "Wyoming", "is_active": true, "amount": 8460.1, "rate": 0.23, "currency": "CVE", "department": "Sales", "country": "Kyrgyzstan"}, {"name": "Ivan Diaz", "state": "Texas", "is_active": true, "amount": 920.93, "rate": 0.7140000000000001, "currency": "AOA", "department": "Sales", "country": "Viet Nam"}, {"name": "Eve Park", "state": "Texas", "is_active": false, "amount": 5364.62, "rate": 0.401, "currency": "AOA", "department": "Human Resources", "country": "Sint Maarten (Dutch part)"}, {"name": "Grace Ali", "state": "United States Minor Outlying Islands", "is_active": true, "amount": 0.0, "rate": 0.807, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Mona Wong", "state": "Wyoming", "is_active": false, "amount": 6317.04, "rate": 0.45, "currency": "AOA", "department": "Human Resources", "country": "Western Sahara"}, {"name": "Lena Khan", "state": "Kentucky", "is_active": false, "amount": 6657.55, "rate": 0.08199999999999999, "currency": "AOA", "department": "Sales", "country": "Western Sahara"}, {"name": "Bob Cruz", "state": "Wyoming", "is_active": false, "amount": 5613.77, "rate": 0.9590000000000001, "currency": "CVE", "department": "Sales", "country": "Sint Maarten (Dutch part)"}, {"name": "Alice Cruz", "state": "Wyoming", "is_active": true, "amount": NaN, "rate": 0.7440000000000001, "currency": "AOA", "department": "Human Resources", "country": "Viet Nam"}, {"name": "Heidi Moore", "state": "Kentucky", "is_active": true, "amount": 7688.85, "rate": 0.161, "currency": "AOA", "department": "Sales", "country": "Sint Maarten (Dutch part)"}], "dirty_cols": ["name", "state", "is_active", "amount", "rate", "currency", "department", "country"], "clean_cols": ["name", "state", "is_active", "amount", "rate", "currency", "department", "country"], "plan": {"dataset_summary": "59 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Um": "United States Minor Outlying Islands", "WY": "Wyoming", "UM": "United States Minor Outlying Islands", "TEXAS": "Texas", "pennsylvania": "Pennsylvania", "tx": "Texas", "united states minor outlying islands": "United States Minor Outlying Islands", "TX": "Texas", "texas": "Texas", "wy": "Wyoming", "KENTUCKY": "Kentucky", "Wyomig": "Wyoming", "PA": "Pennsylvania", "kentucy": "Kentucky", "um": "United States Minor Outlying Islands", "wyoming": "Wyoming", "Pa": "Pennsylvania", "kentucky": "Kentucky", "Ky": "Kentucky", "PENNSYLVANIA": "Pennsylvania", "TEAS": "Texas", "united sttates minor outlying islands": "United States Minor Outlying Islands", "Wy": "Wyoming", "WYOMING": "Wyoming", "ky": "Kentucky"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Kwanza": "AOA", "aoa": "AOA", "Cve": "CVE", "cabo verde escudo": "CVE", "kwanza": "AOA", "Cabo Verde Escudo": "CVE", "KWNAZA": "AOA", "Caabo Verde Escudo": "CVE", "Cabbo Verde Escudo": "CVE", "Aoa": "AOA", "CABO VERDE ESCUDO": "CVE", "KWANZA": "AOA", "Kwana": "AOA", "Kwaza": "AOA"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HR": "Human Resources", "Biz Dev": "Sales", "people": "Human Resources", "Hr": "Human Resources", "people ops": "Human Resources", "PEOPLE OPS": "Human Resources", "sales": "Sales", "People": "Human Resources", "PEOPL EOPS": "Human Resources", "Human RResources": "Human Resources", "BIZ DEV": "Sales", "HUMAN RESOURCES": "Human Resources", "People Ops": "Human Resources", "Biz Dbv": "Sales", "biz dev": "Sales", "BIZ  DEV": "Sales", "People OOps": "Human Resources", "salse": "Sales", "Bi zDev": "Sales", "Peoplp Ops": "Human Resources", "Pemple": "Human Resources", "hr": "Human Resources"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"benin": "Benin", "Socialist Republic of Viet Nam": "Viet Nam", "ESH": "Western Sahara", "Esh": "Western Sahara", "BEN": "Benin", "western sahara": "Western Sahara", "sx": "Sint Maarten (Dutch part)", "KGZ": "Kyrgyzstan", "Republic of Benin": "Benin", "BENIN": "Benin", "VNM": "Viet Nam", "ben": "Benin", "kg": "Kyrgyzstan", "VN": "Viet Nam", "Sx": "Sint Maarten (Dutch part)", "vnm": "Viet Nam", "SX": "Sint Maarten (Dutch part)", "eh": "Western Sahara", "EH": "Western Sahara", "SXM": "Sint Maarten (Dutch part)", "REPUBLIC OF BENIN": "Benin", "Vist Nam": "Viet Nam", "SOCIALIST REPUBLIC OF VIET NAM": "Viet Nam", "Republic ofmBenin": "Benin", "kyrgyzstan": "Kyrgyzstan", "sxm": "Sint Maarten (Dutch part)", "esh": "Western Sahara", "Kg": "Kyrgyzstan", "SINT MAARTEN (DUTCH PART)": "Sint Maarten (Dutch part)", "Vn": "Viet Nam", "Sxm": "Sint Maarten (Dutch part)"}, "rationale": "Unified 31 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"unit": "ML", "amount": "1.181,17", "phone": "(526)845-0927", "notes2": ""}, {"unit": "lb", "amount": "8.111,92", "phone": "567-077-6513", "notes2": ""}, {"unit": "Lb", "amount": "6.228,82", "phone": "589-300-8549", "notes2": ""}, {"unit": " KG ", "amount": "3.171,44", "phone": "529.789.9549", "notes2": ""}, {"unit": "milliliter", "amount": "6.166,80", "phone": "506-803-8114", "notes2": ""}, {"unit": "kg", "amount": "8.245,96", "phone": "5857401899", "notes2": ""}, {"unit": " ml ", "amount": "-", "phone": "512.287.9976", "notes2": ""}, {"unit": "lbs", "amount": "8.941,92", "phone": "(547)457-0405", "notes2": ""}, {"unit": "kilogram", "amount": "?", "phone": "533-150-2698", "notes2": ""}, {"unit": "kilograms", "amount": "$8,569.08", "phone": "(556)465-5129", "notes2": ""}, {"unit": "pounds", "amount": "$964.04", "phone": "5398261645", "notes2": ""}, {"unit": " mL ", "amount": "5.835,21", "phone": "514-738-3494", "notes2": ""}, {"unit": "milliliter", "amount": "$7,094.21", "phone": "596.746.9115", "notes2": ""}, {"unit": "POUND", "amount": "$410.03", "phone": "5801645249", "notes2": ""}, {"unit": "LBS", "amount": "7.061,05", "phone": "(545)859-7122", "notes2": ""}, {"unit": "Kg", "amount": "$7,950.04", "phone": "564.491.3708", "notes2": ""}, {"unit": " pound", "amount": "1.890,31", "phone": "562-616-6916", "notes2": ""}, {"unit": "   milliliter", "amount": "5.960,20", "phone": "5538980614", "notes2": ""}, {"unit": "lbs", "amount": "0", "phone": "508-592-6884", "notes2": ""}, {"unit": "KG", "amount": "7.638,85", "phone": "542-933-3045", "notes2": ""}, {"unit": "mls", "amount": "7.067,61", "phone": "(528)443-5107", "notes2": ""}, {"unit": "milliliter", "amount": "651,73", "phone": "5906728266", "notes2": ""}, {"unit": "  ML", "amount": "2.878,52", "phone": "(533)128-8465", "notes2": ""}, {"unit": "   milliliter", "amount": "5.960,20", "phone": "5538980614", "notes2": ""}], "clean": [{"unit": "mL", "amount": 1181.17, "phone": "(526) 845-0927"}, {"unit": "lb", "amount": 8111.92, "phone": "(567) 077-6513"}, {"unit": "lb", "amount": 6228.82, "phone": "(589) 300-8549"}, {"unit": "kg", "amount": 3171.44, "phone": "(529) 789-9549"}, {"unit": "mL", "amount": 6166.8, "phone": "(506) 803-8114"}, {"unit": "kg", "amount": 8245.96, "phone": "(585) 740-1899"}, {"unit": "mL", "amount": NaN, "phone": "(512) 287-9976"}, {"unit": "lb", "amount": 8941.92, "phone": "(547) 457-0405"}, {"unit": "kg", "amount": NaN, "phone": "(533) 150-2698"}, {"unit": "kg", "amount": 8569.08, "phone": "(556) 465-5129"}, {"unit": "lb", "amount": 964.04, "phone": "(539) 826-1645"}, {"unit": "mL", "amount": 5835.21, "phone": "(514) 738-3494"}, {"unit": "mL", "amount": 7094.21, "phone": "(596) 746-9115"}, {"unit": "lb", "amount": 410.03, "phone": "(580) 164-5249"}, {"unit": "lb", "amount": 7061.05, "phone": "(545) 859-7122"}, {"unit": "kg", "amount": 7950.04, "phone": "(564) 491-3708"}, {"unit": "lb", "amount": 1890.31, "phone": "(562) 616-6916"}, {"unit": "mL", "amount": 5960.2, "phone": "(553) 898-0614"}, {"unit": "lb", "amount": 0.0, "phone": "(508) 592-6884"}, {"unit": "kg", "amount": 7638.85, "phone": "(542) 933-3045"}, {"unit": "mL", "amount": 7067.61, "phone": "(528) 443-5107"}, {"unit": "mL", "amount": 651.73, "phone": "(590) 672-8266"}, {"unit": "mL", "amount": 2878.52, "phone": "(533) 128-8465"}], "dirty_cols": ["unit", "amount", "phone", "notes2"], "clean_cols": ["unit", "amount", "phone"], "plan": {"dataset_summary": "24 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ML": "mL", "Lb": "lb", "KG": "kg", "milliliter": "mL", "ml": "mL", "lbs": "lb", "kilogram": "kg", "kilograms": "kg", "pounds": "lb", "POUND": "lb", "LBS": "lb", "Kg": "kg", "pound": "lb", "mls": "mL"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"company": "Soylent Corp", "city": "smyrna", "phone": "(540)073-4928", "state": "me", "is_active": "T", "country": "slovenia", "rate": "24.8%", "notes2": ""}, {"company": "Cyberdyne", "city": " delft ", "phone": "501-685-8535", "state": "Maine", "is_active": "false", "country": "  Pakistan", "rate": "14.9%", "notes2": ""}, {"company": "Hooli ", "city": "Delft", "phone": "533-272-8570", "state": "Maine", "is_active": "T", "country": "MKD", "rate": "31.7%", "notes2": ""}, {"company": "   Stark Industries", "city": "Nowshera Kalan", "phone": "(594)253-9055", "state": "maine", "is_active": "TRUE", "country": "   Slovenia", "rate": "92.0%", "notes2": ""}, {"company": "Umbrella", "city": "Zawiercie   ", "phone": "530.355.2876", "state": "iowa", "is_active": "0", "country": "SI", "rate": "10.9%", "notes2": ""}, {"company": "  Globex", "city": "ZAWIERCIE", "phone": "598.107.2200", "state": "Louisiana", "is_active": "Yes", "country": "SI ", "rate": "89.4%", "notes2": ""}, {"company": "Initech", "city": "nowshera kalan", "phone": "5366328294", "state": "me", "is_active": "TRUE", "country": "   North Macedonia", "rate": "87.6%", "notes2": ""}, {"company": "Umbrella", "city": "zawieercie", "phone": "519.201.4571", "state": "Maine", "is_active": "FALSE", "country": "MK", "rate": "66.1%", "notes2": ""}, {"company": "Wonka  Co", "city": "Nowshera Kalan", "phone": "5373456184", "state": "LA", "is_active": "Y", "country": "   pakistan", "rate": "7.8%", "notes2": ""}, {"company": "Stark Industries ", "city": "smyran", "phone": "560-187-4462", "state": "MAINE", "is_active": "F", "country": "pk", "rate": "61.2%", "notes2": ""}, {"company": "Stark Industries", "city": "delft", "phone": "5337452073", "state": " maine ", "is_active": "TRUE", "country": "Asm", "rate": "48.9%", "notes2": ""}, {"company": " Hooli", "city": "Nowshera Kalan", "phone": "505.384.3974", "state": "LA", "is_active": "Y", "country": "SI", "rate": "49.7%", "notes2": ""}, {"company": "Globex", "city": "   Smyrna", "phone": "5478775024", "state": "Maine", "is_active": "N", "country": "Republic of North Macedonia", "rate": "12.7%", "notes2": ""}, {"company": "Vehement", "city": " ZAWIERCIE", "phone": "5386297384", "state": " IOWA ", "is_active": "FALSE", "country": "republic of north macedonia", "rate": "95.1%", "notes2": ""}, {"company": "Soylent Corp   ", "city": "Nowshera Kalan", "phone": "5265519569", "state": "iowa", "is_active": "F", "country": "ASM", "rate": "7.7%", "notes2": ""}, {"company": "Umbrella", "city": "Delft", "phone": "506-293-2412", "state": "LA", "is_active": "N", "country": "North Macedonia", "rate": "13.1%", "notes2": ""}, {"company": "Wonka  Co", "city": "Delft", "phone": "503-747-4785", "state": "   Maine", "is_active": "false", "country": "  ASM", "rate": "25.8%", "notes2": ""}, {"company": "Stark Industries ", "city": "Smyrna", "phone": "565-958-6287", "state": "IA", "is_active": "true", "country": "republic of north macedonia ", "rate": "63.5%", "notes2": ""}, {"company": "Initech  ", "city": "NOWSHERA KALAN", "phone": "542-274-6413", "state": "ME", "is_active": "1", "country": "PK", "rate": "78.3%", "notes2": ""}, {"company": "Soylent Corp", "city": "Delft", "phone": "5621553710", "state": "iowa ", "is_active": "FALSE", "country": "asm", "rate": "51.7%", "notes2": ""}, {"company": "Umbrella", "city": "delft", "phone": "(539)216-4342", "state": "Louisiana", "is_active": "Yes", "country": "SVN", "rate": "31.0%", "notes2": ""}, {"company": "Stark Industries", "city": "smyrna", "phone": "547-000-1409", "state": "Iowa", "is_active": "T", "country": " MK ", "rate": "93.8%", "notes2": ""}, {"company": "Cyberdyne  ", "city": "Delft ", "phone": "503-096-0056", "state": "LA", "is_active": "N", "country": "republic of slovenia", "rate": "48.4%", "notes2": ""}, {"company": "Soylent Corp", "city": "Smyrna", "phone": "563.812.3381", "state": "IA", "is_active": "N", "country": "slovenia", "rate": "76.7%", "notes2": ""}, {"company": "Hooli ", "city": "Smyrna", "phone": "5054019681", "state": "Iowa", "is_active": "0", "country": "Aemrican Samoa", "rate": "82.5%", "notes2": ""}, {"company": "Wonka Co", "city": "nowshera kalan", "phone": "(569)787-0850", "state": " Mine ", "is_active": "Y", "country": "PK", "rate": "33.7%", "notes2": ""}, {"company": "Stark  Industries", "city": "ZAWIERCIE", "phone": "5639129632", "state": "Me", "is_active": "N", "country": "Pak", "rate": "74.9%", "notes2": ""}, {"company": "Stark Industries ", "city": "zawiercie", "phone": "578.467.9895", "state": "LA", "is_active": "F", "country": "AMERIICAN  SAMOA", "rate": "33.0%", "notes2": ""}, {"company": "Stark Industries   ", "city": " smyrna ", "phone": "589-453-8344", "state": "Louisiana", "is_active": "false", "country": "Islamic Republic of Pakistan", "rate": "40.6%", "notes2": ""}, {"company": "  Soylent Corp", "city": "Nowshera Kalan", "phone": "5639954298", "state": "ME", "is_active": "0", "country": "pakistan", "rate": "49.1%", "notes2": ""}, {"company": "Stark Industries ", "city": "nowshera kalan", "phone": "5652047075", "state": "ME ", "is_active": "No", "country": "MKD", "rate": "53.8%", "notes2": ""}, {"company": "Hooli", "city": "Smyrna", "phone": "(581)182-7817", "state": "Maine", "is_active": "F", "country": "islamic republic of pakistan   ", "rate": "22.8%", "notes2": ""}, {"company": "Soylent Corp  ", "city": "SMYRNA  ", "phone": "591-183-4974", "state": "la", "is_active": "N", "country": "MK", "rate": "85.2%", "notes2": ""}, {"company": "Acme Inc", "city": " Nowshera Kalan", "phone": "5300301138", "state": "Ia  ", "is_active": "Y", "country": "PK", "rate": "5.4%", "notes2": ""}, {"company": "Initech", "city": "nowshera kalan", "phone": "547-247-7165", "state": " Iowa ", "is_active": "F", "country": "Republic of North Macedonia", "rate": "44.8%", "notes2": ""}, {"company": "Vehement", "city": "Smyrna", "phone": "(580)344-0437", "state": "LA", "is_active": "N", "country": "SI", "rate": "51.2%", "notes2": ""}, {"company": "Globex", "city": "nowshera kalan ", "phone": "598.554.9176", "state": "ME   ", "is_active": "No", "country": "slovenia", "rate": "74.1%", "notes2": ""}, {"company": "Acme Inc", "city": "Nowshera Kalan", "phone": "5844029393", "state": "ME", "is_active": "TRUE", "country": "SI", "rate": "59.5%", "notes2": ""}, {"company": "Hooli ", "city": "Nowshera Kalan", "phone": "509.027.7520", "state": "Maine", "is_active": "N", "country": "ASM", "rate": "13.3%", "notes2": ""}, {"company": "Acme Inc  ", "city": "Zawiercie", "phone": "5802624391", "state": "Maine", "is_active": "Y", "country": "ASM", "rate": "52.5%", "notes2": ""}, {"company": "Cyberdyne", "city": "zawiercie", "phone": "(540)169-2466", "state": "maine ", "is_active": "Yes", "country": "PK", "rate": "50.0%", "notes2": ""}, {"company": " Cyberdyne ", "city": " Delft ", "phone": "535.265.8023", "state": "  Maine", "is_active": "FALSE", "country": "American Samoa   ", "rate": "24.0%", "notes2": ""}, {"company": "Cyberdyne", "city": "Zawiercie  ", "phone": "5769837303", "state": "iowa", "is_active": "F", "country": "as", "rate": "97.4%", "notes2": ""}, {"company": "  Cyberdyne", "city": " DELFT", "phone": "544-803-7381", "state": " Iowa", "is_active": "F", "country": "MKD", "rate": "32.7%", "notes2": ""}, {"company": "Stark Industries", "city": "nowshera kalan", "phone": "5901618039", "state": " Me", "is_active": "0", "country": "REPUBLIC OF SLOVENIA", "rate": "84.3%", "notes2": ""}, {"company": "  Wonka Co", "city": "Delft", "phone": "525-244-1208", "state": "IA", "is_active": "FALSE", "country": "MKD", "rate": "18.1%", "notes2": ""}, {"company": "  Acme Inc", "city": "DELFT", "phone": "570.579.2471", "state": " ia ", "is_active": "false", "country": "PAK", "rate": "62.2%", "notes2": ""}, {"company": "Umbrella", "city": "Delft", "phone": "586.492.2056", "state": "iowa", "is_active": "false", "country": "MK", "rate": "63.5%", "notes2": ""}, {"company": "Globex", "city": "delft", "phone": "584.676.5661", "state": "iowa   ", "is_active": "0", "country": "Slovenia", "rate": "57.4%", "notes2": ""}, {"company": "Globex", "city": "delft", "phone": "590-760-4731", "state": "LA", "is_active": "false", "country": "Slovenia", "rate": "71.5%", "notes2": ""}, {"company": "  Cyberdyne", "city": "Zawiercie", "phone": "541.890.9158", "state": "maine", "is_active": "Yes", "country": "American Samoa", "rate": "46.0%", "notes2": ""}, {"company": "Initech", "city": "Nowshera Kalan", "phone": "549-689-5800", "state": "la", "is_active": "TRUE", "country": "mk", "rate": "76.1%", "notes2": ""}, {"company": "   Wonka Co", "city": "smyrqa", "phone": "584.243.9281", "state": "IA", "is_active": "No", "country": "Republic of Slovenia", "rate": "82.3%", "notes2": ""}, {"company": "Cyberdyne", "city": "Nowshera Kalan", "phone": "525.039.7552", "state": "  Maine", "is_active": "Y", "country": "North Macedonia", "rate": "54.2%", "notes2": ""}, {"company": "Hooli", "city": "nowshera kalan", "phone": "585.442.4208", "state": " Iowa", "is_active": "F", "country": "North Macedonia", "rate": "70.2%", "notes2": ""}, {"company": "Soylent Corp  ", "city": "zawiercie", "phone": "(584)252-3277", "state": "ME", "is_active": "false", "country": "ISLAMIC REPUBLIC OF PAKISTAN", "rate": "31.2%", "notes2": ""}], "clean": [{"company": "Soylent Corp", "city": "Smyrna", "phone": "(540) 073-4928", "state": "Maine", "is_active": true, "country": "Slovenia", "rate": 0.248}, {"company": "Cyberdyne", "city": "Delft", "phone": "(501) 685-8535", "state": "Maine", "is_active": false, "country": "Pakistan", "rate": 0.149}, {"company": "Hooli", "city": "Delft", "phone": "(533) 272-8570", "state": "Maine", "is_active": true, "country": "North Macedonia", "rate": 0.317}, {"company": "Stark Industries", "city": "Nowshera Kalan", "phone": "(594) 253-9055", "state": "Maine", "is_active": true, "country": "Slovenia", "rate": 0.92}, {"company": "Umbrella", "city": "Zawiercie", "phone": "(530) 355-2876", "state": "Iowa", "is_active": false, "country": "Slovenia", "rate": 0.109}, {"company": "Globex", "city": "Zawiercie", "phone": "(598) 107-2200", "state": "Louisiana", "is_active": true, "country": "Slovenia", "rate": 0.894}, {"company": "Initech", "city": "Nowshera Kalan", "phone": "(536) 632-8294", "state": "Maine", "is_active": true, "country": "North Macedonia", "rate": 0.8759999999999999}, {"company": "Umbrella", "city": "Zawiercie", "phone": "(519) 201-4571", "state": "Maine", "is_active": false, "country": "North Macedonia", "rate": 0.6609999999999999}, {"company": "Wonka Co", "city": "Nowshera Kalan", "phone": "(537) 345-6184", "state": "Louisiana", "is_active": true, "country": "Pakistan", "rate": 0.078}, {"company": "Stark Industries", "city": "Smyrna", "phone": "(560) 187-4462", "state": "Maine", "is_active": false, "country": "Pakistan", "rate": 0.612}, {"company": "Stark Industries", "city": "Delft", "phone": "(533) 745-2073", "state": "Maine", "is_active": true, "country": "American Samoa", "rate": 0.489}, {"company": "Hooli", "city": "Nowshera Kalan", "phone": "(505) 384-3974", "state": "Louisiana", "is_active": true, "country": "Slovenia", "rate": 0.49700000000000005}, {"company": "Globex", "city": "Smyrna", "phone": "(547) 877-5024", "state": "Maine", "is_active": false, "country": "North Macedonia", "rate": 0.127}, {"company": "Vehement", "city": "Zawiercie", "phone": "(538) 629-7384", "state": "Iowa", "is_active": false, "country": "North Macedonia", "rate": 0.951}, {"company": "Soylent Corp", "city": "Nowshera Kalan", "phone": "(526) 551-9569", "state": "Iowa", "is_active": false, "country": "American Samoa", "rate": 0.077}, {"company": "Umbrella", "city": "Delft", "phone": "(506) 293-2412", "state": "Louisiana", "is_active": false, "country": "North Macedonia", "rate": 0.131}, {"company": "Wonka Co", "city": "Delft", "phone": "(503) 747-4785", "state": "Maine", "is_active": false, "country": "American Samoa", "rate": 0.258}, {"company": "Stark Industries", "city": "Smyrna", "phone": "(565) 958-6287", "state": "Iowa", "is_active": true, "country": "North Macedonia", "rate": 0.635}, {"company": "Initech", "city": "Nowshera Kalan", "phone": "(542) 274-6413", "state": "Maine", "is_active": true, "country": "Pakistan", "rate": 0.7829999999999999}, {"company": "Soylent Corp", "city": "Delft", "phone": "(562) 155-3710", "state": "Iowa", "is_active": false, "country": "American Samoa", "rate": 0.517}, {"company": "Umbrella", "city": "Delft", "phone": "(539) 216-4342", "state": "Louisiana", "is_active": true, "country": "Slovenia", "rate": 0.31}, {"company": "Stark Industries", "city": "Smyrna", "phone": "(547) 000-1409", "state": "Iowa", "is_active": true, "country": "North Macedonia", "rate": 0.938}, {"company": "Cyberdyne", "city": "Delft", "phone": "(503) 096-0056", "state": "Louisiana", "is_active": false, "country": "Slovenia", "rate": 0.484}, {"company": "Soylent Corp", "city": "Smyrna", "phone": "(563) 812-3381", "state": "Iowa", "is_active": false, "country": "Slovenia", "rate": 0.767}, {"company": "Hooli", "city": "Smyrna", "phone": "(505) 401-9681", "state": "Iowa", "is_active": false, "country": "American Samoa", "rate": 0.825}, {"company": "Wonka Co", "city": "Nowshera Kalan", "phone": "(569) 787-0850", "state": "Maine", "is_active": true, "country": "Pakistan", "rate": 0.337}, {"company": "Stark Industries", "city": "Zawiercie", "phone": "(563) 912-9632", "state": "Maine", "is_active": false, "country": "Pakistan", "rate": 0.7490000000000001}, {"company": "Stark Industries", "city": "Zawiercie", "phone": "(578) 467-9895", "state": "Louisiana", "is_active": false, "country": "American Samoa", "rate": 0.33}, {"company": "Stark Industries", "city": "Smyrna", "phone": "(589) 453-8344", "state": "Louisiana", "is_active": false, "country": "Pakistan", "rate": 0.406}, {"company": "Soylent Corp", "city": "Nowshera Kalan", "phone": "(563) 995-4298", "state": "Maine", "is_active": false, "country": "Pakistan", "rate": 0.491}, {"company": "Stark Industries", "city": "Nowshera Kalan", "phone": "(565) 204-7075", "state": "Maine", "is_active": false, "country": "North Macedonia", "rate": 0.5379999999999999}, {"company": "Hooli", "city": "Smyrna", "phone": "(581) 182-7817", "state": "Maine", "is_active": false, "country": "Pakistan", "rate": 0.228}, {"company": "Soylent Corp", "city": "Smyrna", "phone": "(591) 183-4974", "state": "Louisiana", "is_active": false, "country": "North Macedonia", "rate": 0.852}, {"company": "Acme Inc", "city": "Nowshera Kalan", "phone": "(530) 030-1138", "state": "Iowa", "is_active": true, "country": "Pakistan", "rate": 0.054000000000000006}, {"company": "Initech", "city": "Nowshera Kalan", "phone": "(547) 247-7165", "state": "Iowa", "is_active": false, "country": "North Macedonia", "rate": 0.44799999999999995}, {"company": "Vehement", "city": "Smyrna", "phone": "(580) 344-0437", "state": "Louisiana", "is_active": false, "country": "Slovenia", "rate": 0.512}, {"company": "Globex", "city": "Nowshera Kalan", "phone": "(598) 554-9176", "state": "Maine", "is_active": false, "country": "Slovenia", "rate": 0.741}, {"company": "Acme Inc", "city": "Nowshera Kalan", "phone": "(584) 402-9393", "state": "Maine", "is_active": true, "country": "Slovenia", "rate": 0.595}, {"company": "Hooli", "city": "Nowshera Kalan", "phone": "(509) 027-7520", "state": "Maine", "is_active": false, "country": "American Samoa", "rate": 0.133}, {"company": "Acme Inc", "city": "Zawiercie", "phone": "(580) 262-4391", "state": "Maine", "is_active": true, "country": "American Samoa", "rate": 0.525}, {"company": "Cyberdyne", "city": "Zawiercie", "phone": "(540) 169-2466", "state": "Maine", "is_active": true, "country": "Pakistan", "rate": 0.5}, {"company": "Cyberdyne", "city": "Delft", "phone": "(535) 265-8023", "state": "Maine", "is_active": false, "country": "American Samoa", "rate": 0.24}, {"company": "Cyberdyne", "city": "Zawiercie", "phone": "(576) 983-7303", "state": "Iowa", "is_active": false, "country": "American Samoa", "rate": 0.9740000000000001}, {"company": "Cyberdyne", "city": "Delft", "phone": "(544) 803-7381", "state": "Iowa", "is_active": false, "country": "North Macedonia", "rate": 0.327}, {"company": "Stark Industries", "city": "Nowshera Kalan", "phone": "(590) 161-8039", "state": "Maine", "is_active": false, "country": "Slovenia", "rate": 0.843}, {"company": "Wonka Co", "city": "Delft", "phone": "(525) 244-1208", "state": "Iowa", "is_active": false, "country": "North Macedonia", "rate": 0.18100000000000002}, {"company": "Acme Inc", "city": "Delft", "phone": "(570) 579-2471", "state": "Iowa", "is_active": false, "country": "Pakistan", "rate": 0.622}, {"company": "Umbrella", "city": "Delft", "phone": "(586) 492-2056", "state": "Iowa", "is_active": false, "country": "North Macedonia", "rate": 0.635}, {"company": "Globex", "city": "Delft", "phone": "(584) 676-5661", "state": "Iowa", "is_active": false, "country": "Slovenia", "rate": 0.574}, {"company": "Globex", "city": "Delft", "phone": "(590) 760-4731", "state": "Louisiana", "is_active": false, "country": "Slovenia", "rate": 0.715}, {"company": "Cyberdyne", "city": "Zawiercie", "phone": "(541) 890-9158", "state": "Maine", "is_active": true, "country": "American Samoa", "rate": 0.46}, {"company": "Initech", "city": "Nowshera Kalan", "phone": "(549) 689-5800", "state": "Louisiana", "is_active": true, "country": "North Macedonia", "rate": 0.7609999999999999}, {"company": "Wonka Co", "city": "Smyrna", "phone": "(584) 243-9281", "state": "Iowa", "is_active": false, "country": "Slovenia", "rate": 0.823}, {"company": "Cyberdyne", "city": "Nowshera Kalan", "phone": "(525) 039-7552", "state": "Maine", "is_active": true, "country": "North Macedonia", "rate": 0.542}, {"company": "Hooli", "city": "Nowshera Kalan", "phone": "(585) 442-4208", "state": "Iowa", "is_active": false, "country": "North Macedonia", "rate": 0.7020000000000001}, {"company": "Soylent Corp", "city": "Zawiercie", "phone": "(584) 252-3277", "state": "Maine", "is_active": false, "country": "Pakistan", "rate": 0.312}], "dirty_cols": ["company", "city", "phone", "state", "is_active", "country", "rate", "notes2"], "clean_cols": ["company", "city", "phone", "state", "is_active", "country", "rate"], "plan": {"dataset_summary": "56 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"smyrna": "Smyrna", "delft": "Delft", "ZAWIERCIE": "Zawiercie", "nowshera kalan": "Nowshera Kalan", "zawieercie": "Zawiercie", "smyran": "Smyrna", "NOWSHERA KALAN": "Nowshera Kalan", "zawiercie": "Zawiercie", "SMYRNA": "Smyrna", "DELFT": "Delft", "smyrqa": "Smyrna"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"me": "Maine", "maine": "Maine", "iowa": "Iowa", "LA": "Louisiana", "MAINE": "Maine", "IOWA": "Iowa", "IA": "Iowa", "ME": "Maine", "Mine": "Maine", "Me": "Maine", "la": "Louisiana", "Ia": "Iowa", "ia": "Iowa"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"slovenia": "Slovenia", "MKD": "North Macedonia", "SI": "Slovenia", "MK": "North Macedonia", "pakistan": "Pakistan", "pk": "Pakistan", "Asm": "American Samoa", "Republic of North Macedonia": "North Macedonia", "republic of north macedonia": "North Macedonia", "ASM": "American Samoa", "PK": "Pakistan", "asm": "American Samoa", "SVN": "Slovenia", "republic of slovenia": "Slovenia", "Aemrican Samoa": "American Samoa", "Pak": "Pakistan", "AMERIICAN SAMOA": "American Samoa", "Islamic Republic of Pakistan": "Pakistan", "islamic republic of pakistan": "Pakistan", "as": "American Samoa", "REPUBLIC OF SLOVENIA": "Slovenia", "PAK": "Pakistan", "mk": "North Macedonia", "Republic of Slovenia": "Slovenia", "ISLAMIC REPUBLIC OF PAKISTAN": "Pakistan"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"currency": "algerian dinar", "rate": "15.6%", "city": "Mugina", "department": "Legal"}, {"currency": "Algerian Dinar", "rate": "52.6%", "city": "Mugija", "department": "Legal & Compliance"}, {"currency": "gnf", "rate": "28.4%", "city": "   Kampong Dungun", "department": "Accounting"}, {"currency": "dzd", "rate": "21.3%", "city": "MUGINA", "department": "Cust Support"}, {"currency": "GUINEAN FRANC", "rate": "17.0%", "city": "Belize City", "department": "cust support"}, {"currency": " ALGERIAN DINAR", "rate": "11.3%", "city": "Mugina", "department": "Legal"}, {"currency": "Algerian Dinar", "rate": "0.4%", "city": "BELIZE CITY", "department": "people ops"}, {"currency": "  guinean franc", "rate": "21.5%", "city": "BELIZE CITY", "department": " finance "}, {"currency": "algerian dinar", "rate": "54.4%", "city": "mugina", "department": "HR"}, {"currency": "DZD", "rate": "12.2%", "city": "Mugina", "department": "Accounting"}, {"currency": "dzd", "rate": "96.3%", "city": "Mugina ", "department": " Support"}, {"currency": " gnf ", "rate": "62.1%", "city": "Waldkirch", "department": "Cust Support"}, {"currency": " Dzd ", "rate": "53.5%", "city": "Waldkrch", "department": "Customer Support"}, {"currency": "   guinean franc", "rate": "14.4%", "city": "Mugina", "department": " Legal"}, {"currency": "dzd", "rate": "97.0%", "city": "Waldkirch", "department": "Legal & Compliance"}, {"currency": "Guinean Franc", "rate": "63.7%", "city": "belize city", "department": "   fin"}, {"currency": "gnf", "rate": "46.3%", "city": "Belize City", "department": "Cust Support"}, {"currency": "Algerian Dinar", "rate": "34.2%", "city": "Kampong Dungun", "department": "People Ops"}, {"currency": "GUINEAN FRANC", "rate": "1.4%", "city": "Mugina ", "department": " FIN "}, {"currency": "Guinean Franc", "rate": "13.7%", "city": "Beelize City", "department": "legal & compliance"}, {"currency": " DZD ", "rate": "5.2%", "city": "WALDKIIRCH ", "department": "Accounting"}, {"currency": "DZD", "rate": "40.0%", "city": "Kampong Dungun", "department": "Accounting"}, {"currency": " DZD ", "rate": "81.3%", "city": "mugina", "department": "People Ops"}, {"currency": "dzd   ", "rate": "38.0%", "city": "Belize City", "department": "PEOOPLE"}, {"currency": "ALGERIAN DINAR", "rate": "12.5%", "city": "Mugina", "department": " LEGAL"}, {"currency": "gnf", "rate": "28.4%", "city": "   Kampong Dungun", "department": "Accounting"}, {"currency": "gnf", "rate": "46.3%", "city": "Belize City", "department": "Cust Support"}], "clean": [{"currency": "DZD", "rate": 0.156, "city": "Mugina", "department": "Legal"}, {"currency": "DZD", "rate": 0.526, "city": "Mugina", "department": "Legal"}, {"currency": "GNF", "rate": 0.284, "city": "Kampong Dungun", "department": "Finance"}, {"currency": "DZD", "rate": 0.213, "city": "Mugina", "department": "Customer Support"}, {"currency": "GNF", "rate": 0.17, "city": "Belize City", "department": "Customer Support"}, {"currency": "DZD", "rate": 0.113, "city": "Mugina", "department": "Legal"}, {"currency": "DZD", "rate": 0.004, "city": "Belize City", "department": "Human Resources"}, {"currency": "GNF", "rate": 0.215, "city": "Belize City", "department": "Finance"}, {"currency": "DZD", "rate": 0.544, "city": "Mugina", "department": "Human Resources"}, {"currency": "DZD", "rate": 0.122, "city": "Mugina", "department": "Finance"}, {"currency": "DZD", "rate": 0.963, "city": "Mugina", "department": "Customer Support"}, {"currency": "GNF", "rate": 0.621, "city": "Waldkirch", "department": "Customer Support"}, {"currency": "DZD", "rate": 0.535, "city": "Waldkirch", "department": "Customer Support"}, {"currency": "GNF", "rate": 0.14400000000000002, "city": "Mugina", "department": "Legal"}, {"currency": "DZD", "rate": 0.97, "city": "Waldkirch", "department": "Legal"}, {"currency": "GNF", "rate": 0.637, "city": "Belize City", "department": "Finance"}, {"currency": "GNF", "rate": 0.46299999999999997, "city": "Belize City", "department": "Customer Support"}, {"currency": "DZD", "rate": 0.342, "city": "Kampong Dungun", "department": "Human Resources"}, {"currency": "GNF", "rate": 0.013999999999999999, "city": "Mugina", "department": "Finance"}, {"currency": "GNF", "rate": 0.13699999999999998, "city": "Belize City", "department": "Legal"}, {"currency": "DZD", "rate": 0.052000000000000005, "city": "Waldkirch", "department": "Finance"}, {"currency": "DZD", "rate": 0.4, "city": "Kampong Dungun", "department": "Finance"}, {"currency": "DZD", "rate": 0.813, "city": "Mugina", "department": "Human Resources"}, {"currency": "DZD", "rate": 0.38, "city": "Belize City", "department": "Human Resources"}, {"currency": "DZD", "rate": 0.125, "city": "Mugina", "department": "Legal"}], "dirty_cols": ["currency", "rate", "city", "department"], "clean_cols": ["currency", "rate", "city", "department"], "plan": {"dataset_summary": "27 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"algerian dinar": "DZD", "Algerian Dinar": "DZD", "gnf": "GNF", "dzd": "DZD", "GUINEAN FRANC": "GNF", "ALGERIAN DINAR": "DZD", "guinean franc": "GNF", "Dzd": "DZD", "Guinean Franc": "GNF"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mugija": "Mugina", "MUGINA": "Mugina", "BELIZE CITY": "Belize City", "mugina": "Mugina", "Waldkrch": "Waldkirch", "belize city": "Belize City", "Beelize City": "Belize City", "WALDKIIRCH": "Waldkirch"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Legal & Compliance": "Legal", "Accounting": "Finance", "Cust Support": "Customer Support", "cust support": "Customer Support", "people ops": "Human Resources", "finance": "Finance", "HR": "Human Resources", "Support": "Customer Support", "fin": "Finance", "People Ops": "Human Resources", "FIN": "Finance", "legal & compliance": "Legal", "PEOOPLE": "Human Resources", "LEGAL": "Legal"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "No", "signup_date": "45112", "name": "Carol Fischer", "status": " LOST ", "city": " birigui ", "currency": "BIF  ", "email": "DIEN@EXAMPLE.COM", "job_title": "Chief Executive Officer"}, {"is_active": "Yes", "signup_date": "2023-07-05", "name": "Omar Wong", "status": "In Progress", "city": "  Bukit Panjang New Town", "currency": "Burundi Franc", "email": "mfbpoi@test.org", "job_title": "Sr. Engineer"}, {"is_active": "T", "signup_date": "2023-05-08", "name": "Priya Novak", "status": "closed lost", "city": "BUKIT PANJANG NEW TOWN", "currency": "nakfa", "email": " AMJCM@MAIL.COM ", "job_title": "Sr Engineer"}, {"is_active": "F", "signup_date": "2023-06-15", "name": "David  Smith", "status": "won", "city": "Birigui", "currency": " BIF", "email": "kmchgb@example.com", "job_title": "ae"}, {"is_active": "true", "signup_date": "2023-05-16", "name": "Frank Lee", "status": "won", "city": " BIRIGUI ", "currency": "   BUURNDI FRANC", "email": "  HIFBPKP@MAIL.COM", "job_title": "Senior Eng"}, {"is_active": "true", "signup_date": "2023-05-01", "name": " Sara Adams", "status": " won ", "city": "Birigui", "currency": "ERN", "email": "jofbfp@example.com", "job_title": "PM"}, {"is_active": "T", "signup_date": "28 Sep 2023", "name": "Lena Brandt", "status": "LOST", "city": "Qujing", "currency": "Danish Krone", "email": "FNEDBE@TEST.ORG", "job_title": "Account Executive"}, {"is_active": "true", "signup_date": "44999", "name": "Eve Park", "status": "COSED-WON", "city": "qujing", "currency": "Nakfa", "email": "bopnbj@corp.io", "job_title": "prod mgr"}, {"is_active": "true", "signup_date": "2023-01-05", "name": "Eve Reyes  ", "status": "loost", "city": "Qujnig", "currency": "Burundi Franc", "email": "EJDBFH@EXAMPLE.COM", "job_title": "PM"}, {"is_active": "No", "signup_date": "6 Jul 2023", "name": "Lena Smith", "status": "closed lost", "city": "Bukit Panjang New Town", "currency": "danish krone", "email": "hchdnab@mail.com", "job_title": "ACCOUNT EXECUTIVE"}, {"is_active": "TRUE", "signup_date": "5 Apr 2023", "name": "Omar Brandt", "status": "In Progress", "city": "birigui", "currency": "DKK", "email": "OGLBB@MAIL.COM", "job_title": "AE"}, {"is_active": "false", "signup_date": "2023-11-19", "name": "Judy Cruz", "status": "WON", "city": "buikt panjang new town", "currency": "Burundi Franc", "email": "OELMEDE@MAIL.COM", "job_title": "Acct Exec"}, {"is_active": "FALSE", "signup_date": "45043", "name": "Grace Ali   ", "status": "CLOSED  WON", "city": "Birigui", "currency": "danish krone", "email": "KPBOMG@MAIL.COM", "job_title": "ceo"}, {"is_active": "FALSE", "signup_date": "45166", "name": "Karl Adams", "status": "In Progress", "city": "Bukit Panjang New Town", "currency": "bif", "email": "gbbmdoi@corp.io  ", "job_title": "AE"}, {"is_active": "1", "signup_date": "2023-08-28", "name": "Sara Khan   ", "status": "in progress", "city": "Biirgui", "currency": "ERN", "email": "NIKHGJC@EXAMPLE.COM", "job_title": "   Senior Eng"}, {"is_active": "Y", "signup_date": "6 Nov 2023", "name": "Karl Moore   ", "status": "won", "city": "BUKIT PANJANG NEW TOWN", "currency": "BIF", "email": "NFDKGG@MAIL.COM", "job_title": "Prod Mgr"}, {"is_active": "N", "signup_date": "45013", "name": "  Karl Park", "status": "Lost", "city": "Qujiog", "currency": "nakfa", "email": "kdfhmb@test.org", "job_title": "Account  Executive"}, {"is_active": "TRUE", "signup_date": "25 Aug 2023", "name": "Judy Cruz", "status": " lost ", "city": "Birigui  ", "currency": "burundi franc", "email": "ohnle@mail.com", "job_title": "Snr  Engineer"}, {"is_active": "TRUE", "signup_date": "18 Aug 2023", "name": "Mona Cruz", "status": "Lost", "city": "BUKIT PANJANG NEW TOWN", "currency": "Danish Krone", "email": "LLEJA@EXAMPLE.COM", "job_title": "SENIOR ENGINEER"}, {"is_active": "0", "signup_date": "4/20/2023", "name": "Omar Brandt", "status": "WON", "city": "Birigui", "currency": "Ern", "email": " HOAIGMJ@MAIL.COM", "job_title": "c.e.o."}, {"is_active": "0", "signup_date": "45065", "name": "Sara Ali", "status": "WON", "city": "Bukit Panjang New Town", "currency": "DANISH KRONE   ", "email": "DIECBKC@EXAMPLE.COM", "job_title": "CEO"}, {"is_active": "FALSE", "signup_date": "9/13/2023", "name": "Heidi Wong", "status": "In Pogress", "city": "Qujing", "currency": "NAKFA   ", "email": "kkbae@corp.io", "job_title": "PM"}, {"is_active": "0", "signup_date": "3 Mar 2023", "name": "  Carol Reyes", "status": "  lost", "city": "qujing", "currency": "Danish Krone", "email": "iebbo@corp.io", "job_title": "ACCW EXEC"}, {"is_active": "No", "signup_date": "44944", "name": "Sara Smith", "status": "  CLOSED WON", "city": "qujiing", "currency": "  ERN", "email": "KABAGJ@MAIL.COM", "job_title": "CEO"}, {"is_active": "FALSE", "signup_date": "11/24/2023", "name": "Bob Park", "status": "closed-won", "city": "birigui", "currency": "Burundi Franc", "email": "oedbod@example.com", "job_title": "acct exec"}, {"is_active": "No", "signup_date": "11/20/2023", "name": "Priya Fischer   ", "status": "WIP", "city": "Bukit Panjang New Town", "currency": "Burundi Franc", "email": "ceed@test.org", "job_title": " ceo "}, {"is_active": "No", "signup_date": "45162", "name": "Mona  Khan", "status": "Lost", "city": "Biirigui", "currency": "  nakfa", "email": "bcbefb@mail.com", "job_title": "Prod Mgr"}, {"is_active": "FALSE", "signup_date": "2023-06-23", "name": "Judy Smith ", "status": "lost", "city": "BIRIGUI", "currency": "DKK", "email": "bklhoa@corp.io", "job_title": "Product Manager"}, {"is_active": "1", "signup_date": "12/19/2023", "name": "Sara Wong", "status": "Won", "city": " BIRIGUI ", "currency": "burundi franc ", "email": "DMKHHA@TEST.ORG", "job_title": "PM"}, {"is_active": "true", "signup_date": "2023-05-10", "name": "Grace  Adams", "status": "closed lost", "city": "  birigui", "currency": "bif", "email": "fbcfkog@example.com", "job_title": "product  manager"}, {"is_active": "F", "signup_date": "27 Jul 2023", "name": "Frank  Brandt", "status": "Won", "city": "Birigui", "currency": " BIF ", "email": "NEJDKJ@CORP.IO ", "job_title": " Senior Engineer"}, {"is_active": "0", "signup_date": "11/14/2023", "name": "   Lena Ali", "status": "Won", "city": "Bukit Panjang New Town", "currency": "bif", "email": "  pbcb@corp.io", "job_title": "account executive   "}, {"is_active": "false", "signup_date": "2023-04-23", "name": "Alice Khan", "status": "   Lost", "city": "qujing", "currency": "danish krone", "email": "BABKKC@MAIL.COM", "job_title": "SENIOR ENG"}, {"is_active": "0", "signup_date": "26 Jul 2023", "name": "Judy Smith ", "status": " Lost ", "city": "Birigui", "currency": "Danish Krone", "email": "FLOAO@TEST.ORG", "job_title": "Chief Executive Officvr"}, {"is_active": "No", "signup_date": "8/7/2023", "name": "David  Fischer", "status": " WON ", "city": "Qujing", "currency": "dkk", "email": "dbgkobe@test.org", "job_title": "AE"}, {"is_active": "F", "signup_date": "4/28/2023", "name": "Carol  Cruz", "status": " won ", "city": " Bukit Panjang New Town", "currency": "burundi franc ", "email": "KJKCJJI@CORP.IO", "job_title": "SENIOR ENG"}, {"is_active": "F", "signup_date": "8 Jun 2023", "name": "Sara Smith", "status": "Ongoing", "city": "Qujing", "currency": "DKK", "email": "cgma@corp.io", "job_title": "C.E.O."}, {"is_active": "TRUE", "signup_date": "2023-01-27", "name": "Ivan Adams", "status": "  Won", "city": "Qujing ", "currency": "BIF", "email": "KBAOHB@EXAMPLE.COM ", "job_title": "Ae"}, {"is_active": "No", "signup_date": "10 Aug 2023", "name": "Frank Adams", "status": " ClosedLost ", "city": "BUKIT PANJANG NEW TOWN", "currency": "DKK", "email": "LJBOMM@MAIL.COM", "job_title": "SR. ENGINEER"}, {"is_active": "No", "signup_date": "45047", "name": "Karl Lee", "status": "won", "city": "qujing  ", "currency": "dkk", "email": "AAMOKAI@MAIL.COM", "job_title": "Chief  Executive Officer"}, {"is_active": "FALSE", "signup_date": "2023-01-24", "name": "Omar  Ali", "status": "Lost", "city": "Birigui", "currency": "DANISH KRONE", "email": "AOOHJ@EXAMPLE.COM ", "job_title": "  Prod Manager"}, {"is_active": "FALSE", "signup_date": "45043", "name": "Grace Ali   ", "status": "CLOSED  WON", "city": "Birigui", "currency": "danish krone", "email": "KPBOMG@MAIL.COM", "job_title": "ceo"}, {"is_active": "false", "signup_date": "2023-04-23", "name": "Alice Khan", "status": "   Lost", "city": "qujing", "currency": "danish krone", "email": "BABKKC@MAIL.COM", "job_title": "SENIOR ENG"}, {"is_active": "", "signup_date": "", "name": "", "status": "", "city": "", "currency": "", "email": "", "job_title": ""}], "clean": [{"is_active": false, "signup_date": "2023-07-05", "name": "Carol Fischer", "status": "Lost", "city": "Birigui", "currency": "BIF", "email": "dien@example.com", "job_title": "Chief Executive Officer"}, {"is_active": true, "signup_date": "2023-07-05", "name": "Omar Wong", "status": "In Progress", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "mfbpoi@test.org", "job_title": "Senior Engineer"}, {"is_active": true, "signup_date": "2023-05-08", "name": "Priya Novak", "status": "Lost", "city": "Bukit Panjang New Town", "currency": "ERN", "email": "amjcm@mail.com", "job_title": "Senior Engineer"}, {"is_active": false, "signup_date": "2023-06-15", "name": "David Smith", "status": "Won", "city": "Birigui", "currency": "BIF", "email": "kmchgb@example.com", "job_title": "Account Executive"}, {"is_active": true, "signup_date": "2023-05-16", "name": "Frank Lee", "status": "Won", "city": "Birigui", "currency": "BIF", "email": "hifbpkp@mail.com", "job_title": "Senior Engineer"}, {"is_active": true, "signup_date": "2023-05-01", "name": "Sara Adams", "status": "Won", "city": "Birigui", "currency": "ERN", "email": "jofbfp@example.com", "job_title": "Product Manager"}, {"is_active": true, "signup_date": "2023-09-28", "name": "Lena Brandt", "status": "Lost", "city": "Qujing", "currency": "DKK", "email": "fnedbe@test.org", "job_title": "Account Executive"}, {"is_active": true, "signup_date": "2023-03-14", "name": "Eve Park", "status": "Won", "city": "Qujing", "currency": "ERN", "email": "bopnbj@corp.io", "job_title": "Product Manager"}, {"is_active": true, "signup_date": "2023-01-05", "name": "Eve Reyes", "status": "Lost", "city": "Qujing", "currency": "BIF", "email": "ejdbfh@example.com", "job_title": "Product Manager"}, {"is_active": false, "signup_date": "2023-07-06", "name": "Lena Smith", "status": "Lost", "city": "Bukit Panjang New Town", "currency": "DKK", "email": "hchdnab@mail.com", "job_title": "Account Executive"}, {"is_active": true, "signup_date": "2023-04-05", "name": "Omar Brandt", "status": "In Progress", "city": "Birigui", "currency": "DKK", "email": "oglbb@mail.com", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-11-19", "name": "Judy Cruz", "status": "Won", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "oelmede@mail.com", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-04-27", "name": "Grace Ali", "status": "Won", "city": "Birigui", "currency": "DKK", "email": "kpbomg@mail.com", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-08-28", "name": "Karl Adams", "status": "In Progress", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "gbbmdoi@corp.io", "job_title": "Account Executive"}, {"is_active": true, "signup_date": "2023-08-28", "name": "Sara Khan", "status": "In Progress", "city": "Birigui", "currency": "ERN", "email": "nikhgjc@example.com", "job_title": "Senior Engineer"}, {"is_active": true, "signup_date": "2023-11-06", "name": "Karl Moore", "status": "Won", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "nfdkgg@mail.com", "job_title": "Product Manager"}, {"is_active": false, "signup_date": "2023-03-28", "name": "Karl Park", "status": "Lost", "city": "Qujing", "currency": "ERN", "email": "kdfhmb@test.org", "job_title": "Account Executive"}, {"is_active": true, "signup_date": "2023-08-25", "name": "Judy Cruz", "status": "Lost", "city": "Birigui", "currency": "BIF", "email": "ohnle@mail.com", "job_title": "Senior Engineer"}, {"is_active": true, "signup_date": "2023-08-18", "name": "Mona Cruz", "status": "Lost", "city": "Bukit Panjang New Town", "currency": "DKK", "email": "lleja@example.com", "job_title": "Senior Engineer"}, {"is_active": false, "signup_date": "2023-04-20", "name": "Omar Brandt", "status": "Won", "city": "Birigui", "currency": "ERN", "email": "hoaigmj@mail.com", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-05-19", "name": "Sara Ali", "status": "Won", "city": "Bukit Panjang New Town", "currency": "DKK", "email": "diecbkc@example.com", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-09-13", "name": "Heidi Wong", "status": "In Progress", "city": "Qujing", "currency": "ERN", "email": "kkbae@corp.io", "job_title": "Product Manager"}, {"is_active": false, "signup_date": "2023-03-03", "name": "Carol Reyes", "status": "Lost", "city": "Qujing", "currency": "DKK", "email": "iebbo@corp.io", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-01-18", "name": "Sara Smith", "status": "Won", "city": "Qujing", "currency": "ERN", "email": "kabagj@mail.com", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-11-24", "name": "Bob Park", "status": "Won", "city": "Birigui", "currency": "BIF", "email": "oedbod@example.com", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-11-20", "name": "Priya Fischer", "status": "In Progress", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "ceed@test.org", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-08-24", "name": "Mona Khan", "status": "Lost", "city": "Birigui", "currency": "ERN", "email": "bcbefb@mail.com", "job_title": "Product Manager"}, {"is_active": false, "signup_date": "2023-06-23", "name": "Judy Smith", "status": "Lost", "city": "Birigui", "currency": "DKK", "email": "bklhoa@corp.io", "job_title": "Product Manager"}, {"is_active": true, "signup_date": "2023-12-19", "name": "Sara Wong", "status": "Won", "city": "Birigui", "currency": "BIF", "email": "dmkhha@test.org", "job_title": "Product Manager"}, {"is_active": true, "signup_date": "2023-05-10", "name": "Grace Adams", "status": "Lost", "city": "Birigui", "currency": "BIF", "email": "fbcfkog@example.com", "job_title": "Product Manager"}, {"is_active": false, "signup_date": "2023-07-27", "name": "Frank Brandt", "status": "Won", "city": "Birigui", "currency": "BIF", "email": "nejdkj@corp.io", "job_title": "Senior Engineer"}, {"is_active": false, "signup_date": "2023-11-14", "name": "Lena Ali", "status": "Won", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "pbcb@corp.io", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-04-23", "name": "Alice Khan", "status": "Lost", "city": "Qujing", "currency": "DKK", "email": "babkkc@mail.com", "job_title": "Senior Engineer"}, {"is_active": false, "signup_date": "2023-07-26", "name": "Judy Smith", "status": "Lost", "city": "Birigui", "currency": "DKK", "email": "floao@test.org", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-08-07", "name": "David Fischer", "status": "Won", "city": "Qujing", "currency": "DKK", "email": "dbgkobe@test.org", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-04-28", "name": "Carol Cruz", "status": "Won", "city": "Bukit Panjang New Town", "currency": "BIF", "email": "kjkcjji@corp.io", "job_title": "Senior Engineer"}, {"is_active": false, "signup_date": "2023-06-08", "name": "Sara Smith", "status": "In Progress", "city": "Qujing", "currency": "DKK", "email": "cgma@corp.io", "job_title": "Chief Executive Officer"}, {"is_active": true, "signup_date": "2023-01-27", "name": "Ivan Adams", "status": "Won", "city": "Qujing", "currency": "BIF", "email": "kbaohb@example.com", "job_title": "Account Executive"}, {"is_active": false, "signup_date": "2023-08-10", "name": "Frank Adams", "status": "Lost", "city": "Bukit Panjang New Town", "currency": "DKK", "email": "ljbomm@mail.com", "job_title": "Senior Engineer"}, {"is_active": false, "signup_date": "2023-05-01", "name": "Karl Lee", "status": "Won", "city": "Qujing", "currency": "DKK", "email": "aamokai@mail.com", "job_title": "Chief Executive Officer"}, {"is_active": false, "signup_date": "2023-01-24", "name": "Omar Ali", "status": "Lost", "city": "Birigui", "currency": "DKK", "email": "aoohj@example.com", "job_title": "Product Manager"}], "dirty_cols": ["is_active", "signup_date", "name", "status", "city", "currency", "email", "job_title"], "clean_cols": ["is_active", "signup_date", "name", "status", "city", "currency", "email", "job_title"], "plan": {"dataset_summary": "44 rows × 8 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LOST": "Lost", "closed lost": "Lost", "won": "Won", "COSED-WON": "Won", "loost": "Lost", "WON": "Won", "CLOSED WON": "Won", "in progress": "In Progress", "lost": "Lost", "In Pogress": "In Progress", "closed-won": "Won", "WIP": "In Progress", "Ongoing": "In Progress", "ClosedLost": "Lost"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"birigui": "Birigui", "BUKIT PANJANG NEW TOWN": "Bukit Panjang New Town", "BIRIGUI": "Birigui", "qujing": "Qujing", "Qujnig": "Qujing", "buikt panjang new town": "Bukit Panjang New Town", "Biirgui": "Birigui", "Qujiog": "Qujing", "qujiing": "Qujing", "Biirigui": "Birigui"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Burundi Franc": "BIF", "nakfa": "ERN", "BUURNDI FRANC": "BIF", "Danish Krone": "DKK", "Nakfa": "ERN", "danish krone": "DKK", "bif": "BIF", "burundi franc": "BIF", "Ern": "ERN", "DANISH KRONE": "DKK", "NAKFA": "ERN", "dkk": "DKK"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Sr. Engineer": "Senior Engineer", "Sr Engineer": "Senior Engineer", "ae": "Account Executive", "Senior Eng": "Senior Engineer", "PM": "Product Manager", "prod mgr": "Product Manager", "ACCOUNT EXECUTIVE": "Account Executive", "AE": "Account Executive", "Acct Exec": "Account Executive", "ceo": "Chief Executive Officer", "Prod Mgr": "Product Manager", "Snr Engineer": "Senior Engineer", "SENIOR ENGINEER": "Senior Engineer", "c.e.o.": "Chief Executive Officer", "CEO": "Chief Executive Officer", "ACCW EXEC": "Account Executive", "acct exec": "Account Executive", "product manager": "Product Manager", "account executive": "Account Executive", "SENIOR ENG": "Senior Engineer", "Chief Executive Officvr": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer", "Ae": "Account Executive", "SR. ENGINEER": "Senior Engineer", "Prod Manager": "Product Manager"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "Tanggu", "signup_date": "11/13/2023", "status": "low", "email": "calcdej@test.org", "country": "ESTONIA", "notes2": ""}, {"city": "Sokal", "signup_date": "2023-08-12", "status": "medium", "email": "CKDIH@EXAMPLE.COM", "country": "ee", "notes2": ""}, {"city": "Irosin", "signup_date": "2023-02-02", "status": "Low", "email": "AAHPP@EXAMPLE.COM", "country": "CG", "notes2": ""}, {"city": "Tanggu", "signup_date": "7 Oct 2023", "status": "low", "email": "   BCOLGGE@EXAMPLE.COM", "country": "republic of estonia", "notes2": ""}, {"city": "Sokal", "signup_date": "5/25/2023", "status": "low   ", "email": "beido@mail.com", "country": "CG", "notes2": ""}, {"city": "sokal", "signup_date": "2 Oct 2023", "status": "med", "email": "bcjem@example.com", "country": "luxembourg", "notes2": ""}, {"city": "Tanggu", "signup_date": "19 Aug 2023", "status": "medium", "email": "IPGBG@TEST.ORG   ", "country": "Luxembourg", "notes2": ""}, {"city": "Sokal", "signup_date": "45223", "status": "  MEDIUM", "email": " NOLJCBD@MAIL.COM", "country": "Luxembourg   ", "notes2": ""}, {"city": " Danjiangkou ", "signup_date": "2023-06-05", "status": "h", "email": "oggdb@test.org ", "country": "Repulbic of Estonia", "notes2": ""}, {"city": " Irosin ", "signup_date": "24 May 2023", "status": "P2", "email": "pfcabj@mail.com", "country": "lu", "notes2": ""}, {"city": "sokal", "signup_date": "17 Mar 2023", "status": "p3", "email": "JGHPFLO@TEST.ORG", "country": "Republic of the Congo", "notes2": ""}, {"city": "Sokal", "signup_date": "2023-12-14", "status": "L", "email": "oienlpb@example.com", "country": "Ee", "notes2": ""}, {"city": "Tanggu", "signup_date": "2023-12-01", "status": "  low", "email": "MECJ@MAIL.COM", "country": "est", "notes2": ""}, {"city": " Irosin ", "signup_date": "24 May 2023", "status": "P2", "email": "pfcabj@mail.com", "country": "lu", "notes2": ""}, {"city": "", "signup_date": "", "status": "", "email": "", "country": "", "notes2": ""}], "clean": [{"city": "Tanggu", "signup_date": "2023-11-13", "status": "Low", "email": "calcdej@test.org", "country": "Estonia"}, {"city": "Sokal", "signup_date": "2023-08-12", "status": "Medium", "email": "ckdih@example.com", "country": "Estonia"}, {"city": "Irosin", "signup_date": "2023-02-02", "status": "Low", "email": "aahpp@example.com", "country": "Congo"}, {"city": "Tanggu", "signup_date": "2023-10-07", "status": "Low", "email": "bcolgge@example.com", "country": "Estonia"}, {"city": "Sokal", "signup_date": "2023-05-25", "status": "Low", "email": "beido@mail.com", "country": "Congo"}, {"city": "Sokal", "signup_date": "2023-10-02", "status": "Medium", "email": "bcjem@example.com", "country": "Luxembourg"}, {"city": "Tanggu", "signup_date": "2023-08-19", "status": "Medium", "email": "ipgbg@test.org", "country": "Luxembourg"}, {"city": "Sokal", "signup_date": "2023-10-24", "status": "Medium", "email": "noljcbd@mail.com", "country": "Luxembourg"}, {"city": "Danjiangkou", "signup_date": "2023-06-05", "status": "High", "email": "oggdb@test.org", "country": "Estonia"}, {"city": "Irosin", "signup_date": "2023-05-24", "status": "Medium", "email": "pfcabj@mail.com", "country": "Luxembourg"}, {"city": "Sokal", "signup_date": "2023-03-17", "status": "Low", "email": "jghpflo@test.org", "country": "Congo"}, {"city": "Sokal", "signup_date": "2023-12-14", "status": "Low", "email": "oienlpb@example.com", "country": "Estonia"}, {"city": "Tanggu", "signup_date": "2023-12-01", "status": "Low", "email": "mecj@mail.com", "country": "Estonia"}], "dirty_cols": ["city", "signup_date", "status", "email", "country", "notes2"], "clean_cols": ["city", "signup_date", "status", "email", "country"], "plan": {"dataset_summary": "15 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sokal": "Sokal"}, "rationale": "Unified 1 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"low": "Low", "medium": "Medium", "med": "Medium", "MEDIUM": "Medium", "h": "High", "P2": "Medium", "p3": "Low", "L": "Low"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ESTONIA": "Estonia", "ee": "Estonia", "CG": "Congo", "republic of estonia": "Estonia", "luxembourg": "Luxembourg", "Repulbic of Estonia": "Estonia", "lu": "Luxembourg", "Republic of the Congo": "Congo", "Ee": "Estonia", "est": "Estonia"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "Danish Krone", "department": "Information Technology", "email": "   HNKCL@TEST.ORG", "unit": "   mi"}, {"currency": "DKK", "department": "It", "email": "  hbpn@example.com", "unit": "killgrams"}, {"currency": "Byn", "department": "ops", "email": "imbjpf@corp.io", "unit": "miles"}, {"currency": "Quetzal", "department": " i.t. ", "email": "khkpeda@mail.com", "unit": "ml"}, {"currency": "Danish  Krone", "department": " marketing ", "email": "DLMPJ@TEST.ORG", "unit": "kiloggram   "}, {"currency": "Danish Krone", "department": "OPS", "email": "IBJDPI@TEST.ORG  ", "unit": "mls"}, {"currency": "gtq", "department": "growth   ", "email": "imffccp@example.com", "unit": " KG "}, {"currency": "danish krone", "department": "marketing  ", "email": "jbmfkkh@corp.io", "unit": "kilogram"}, {"currency": "Gtq", "department": "OPERATIONS", "email": "fbafg@example.com", "unit": "MI"}, {"currency": "belarusian ruble", "department": "  Information Technology", "email": "fffgeh@corp.io", "unit": "kilogram"}, {"currency": "CDF", "department": "Growth", "email": "MIGOA@CORP.IO", "unit": "Kg "}, {"currency": "Danish Krone", "department": "accounting", "email": "   LBBJNJ@TEST.ORG", "unit": "Mi"}, {"currency": "DKK", "department": "Accounitng", "email": "odeikc@mail.com", "unit": "  Mi"}, {"currency": " DKK ", "department": "   IT", "email": "ppebokj@corp.io", "unit": "KG"}, {"currency": "danish krone", "department": "it", "email": " bojco@test.org ", "unit": "mi"}, {"currency": "Congolese Franc", "department": "Finance", "email": "   NONEJ@CORP.IO", "unit": "ML"}, {"currency": "QUETZAL", "department": "GROWTH", "email": "PBIDP@CORP.IO", "unit": "mls"}, {"currency": "   CONGOLESE FRANC", "department": "marketiing", "email": "CHGON@TEST.ORG", "unit": "MILES"}, {"currency": "GTQ ", "department": "ops", "email": "PFJOMP@CORP.IO", "unit": "Mi"}, {"currency": "DKK", "department": "ops", "email": " PKIHO@CORP.IO ", "unit": "mls"}, {"currency": "danish krone", "department": "   Fin", "email": " COBGC@MAIL.COM ", "unit": "mi"}, {"currency": "GTQ", "department": "GROWTH", "email": "nikagh@example.com", "unit": "mile"}, {"currency": "cdf", "department": "I.T.", "email": "GIPGF@TEST.ORG", "unit": "Kg"}, {"currency": "CDF", "department": "Mkt", "email": "  nfbo@example.com", "unit": "miles"}, {"currency": "   Gtq", "department": "Growth", "email": "okafd@mail.com", "unit": "MI"}, {"currency": "Gtq", "department": "OPERATIONS", "email": "fbafg@example.com", "unit": "MI"}, {"currency": "QUETZAL", "department": "GROWTH", "email": "PBIDP@CORP.IO", "unit": "mls"}, {"currency": "", "department": "", "email": "", "unit": ""}], "clean": [{"currency": "DKK", "department": "Information Technology", "email": "hnkcl@test.org", "unit": "mi"}, {"currency": "DKK", "department": "Information Technology", "email": "hbpn@example.com", "unit": "kg"}, {"currency": "BYN", "department": "Operations", "email": "imbjpf@corp.io", "unit": "mi"}, {"currency": "GTQ", "department": "Information Technology", "email": "khkpeda@mail.com", "unit": "mL"}, {"currency": "DKK", "department": "Marketing", "email": "dlmpj@test.org", "unit": "kg"}, {"currency": "DKK", "department": "Operations", "email": "ibjdpi@test.org", "unit": "mL"}, {"currency": "GTQ", "department": "Marketing", "email": "imffccp@example.com", "unit": "kg"}, {"currency": "DKK", "department": "Marketing", "email": "jbmfkkh@corp.io", "unit": "kg"}, {"currency": "GTQ", "department": "Operations", "email": "fbafg@example.com", "unit": "mi"}, {"currency": "BYN", "department": "Information Technology", "email": "fffgeh@corp.io", "unit": "kg"}, {"currency": "CDF", "department": "Marketing", "email": "migoa@corp.io", "unit": "kg"}, {"currency": "DKK", "department": "Finance", "email": "lbbjnj@test.org", "unit": "mi"}, {"currency": "DKK", "department": "Finance", "email": "odeikc@mail.com", "unit": "mi"}, {"currency": "DKK", "department": "Information Technology", "email": "ppebokj@corp.io", "unit": "kg"}, {"currency": "DKK", "department": "Information Technology", "email": "bojco@test.org", "unit": "mi"}, {"currency": "CDF", "department": "Finance", "email": "nonej@corp.io", "unit": "mL"}, {"currency": "GTQ", "department": "Marketing", "email": "pbidp@corp.io", "unit": "mL"}, {"currency": "CDF", "department": "Marketing", "email": "chgon@test.org", "unit": "mi"}, {"currency": "GTQ", "department": "Operations", "email": "pfjomp@corp.io", "unit": "mi"}, {"currency": "DKK", "department": "Operations", "email": "pkiho@corp.io", "unit": "mL"}, {"currency": "DKK", "department": "Finance", "email": "cobgc@mail.com", "unit": "mi"}, {"currency": "GTQ", "department": "Marketing", "email": "nikagh@example.com", "unit": "mi"}, {"currency": "CDF", "department": "Information Technology", "email": "gipgf@test.org", "unit": "kg"}, {"currency": "CDF", "department": "Marketing", "email": "nfbo@example.com", "unit": "mi"}, {"currency": "GTQ", "department": "Marketing", "email": "okafd@mail.com", "unit": "mi"}], "dirty_cols": ["currency", "department", "email", "unit"], "clean_cols": ["currency", "department", "email", "unit"], "plan": {"dataset_summary": "28 rows × 4 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Danish Krone": "DKK", "Byn": "BYN", "Quetzal": "GTQ", "gtq": "GTQ", "danish krone": "DKK", "Gtq": "GTQ", "belarusian ruble": "BYN", "Congolese Franc": "CDF", "QUETZAL": "GTQ", "CONGOLESE FRANC": "CDF", "cdf": "CDF"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"It": "Information Technology", "ops": "Operations", "i.t.": "Information Technology", "marketing": "Marketing", "OPS": "Operations", "growth": "Marketing", "OPERATIONS": "Operations", "Growth": "Marketing", "accounting": "Finance", "Accounitng": "Finance", "IT": "Information Technology", "it": "Information Technology", "GROWTH": "Marketing", "marketiing": "Marketing", "Fin": "Finance", "I.T.": "Information Technology", "Mkt": "Marketing"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"killgrams": "kg", "miles": "mi", "ml": "mL", "kiloggram": "kg", "mls": "mL", "KG": "kg", "kilogram": "kg", "MI": "mi", "Kg": "kg", "Mi": "mi", "ML": "mL", "MILES": "mi", "mile": "mi"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "Yes", "job_title": "VICE PRESIDENT", "phone": "5749732077", "unnamed": ""}, {"is_active": "T", "job_title": "vp", "phone": "552.186.2677", "unnamed": ""}, {"is_active": "true", "job_title": "Engineering Mgr  ", "phone": "522.578.6987", "unnamed": ""}, {"is_active": "F", "job_title": "vice pres", "phone": "(527)924-1728", "unnamed": ""}, {"is_active": "No", "job_title": "Vice Pres", "phone": "5292623203", "unnamed": ""}, {"is_active": "Y", "job_title": "eng  manager", "phone": "5298980103", "unnamed": ""}, {"is_active": "F", "job_title": "engineering mgr", "phone": "(525)429-6414", "unnamed": ""}, {"is_active": "Y", "job_title": "sales  representative", "phone": "5073485233", "unnamed": ""}, {"is_active": "0", "job_title": "  Engineering Mgr", "phone": "(509)742-5448", "unnamed": ""}, {"is_active": "FALSE", "job_title": "engineering manager", "phone": "504-231-5010", "unnamed": ""}, {"is_active": "Yes", "job_title": "VICE PRES", "phone": "590-203-5637", "unnamed": ""}, {"is_active": "Y", "job_title": "eng mgr", "phone": "(526)600-9977", "unnamed": ""}, {"is_active": "Y", "job_title": "Sales Representative", "phone": "(530)300-4356", "unnamed": ""}, {"is_active": "true", "job_title": "Rep", "phone": "532-591-2810", "unnamed": ""}, {"is_active": "true", "job_title": "Engineering Manager", "phone": "520-374-0455", "unnamed": ""}], "clean": [{"is_active": true, "job_title": "Vice President", "phone": "(574) 973-2077"}, {"is_active": true, "job_title": "Vice President", "phone": "(552) 186-2677"}, {"is_active": true, "job_title": "Engineering Manager", "phone": "(522) 578-6987"}, {"is_active": false, "job_title": "Vice President", "phone": "(527) 924-1728"}, {"is_active": false, "job_title": "Vice President", "phone": "(529) 262-3203"}, {"is_active": true, "job_title": "Engineering Manager", "phone": "(529) 898-0103"}, {"is_active": false, "job_title": "Engineering Manager", "phone": "(525) 429-6414"}, {"is_active": true, "job_title": "Sales Representative", "phone": "(507) 348-5233"}, {"is_active": false, "job_title": "Engineering Manager", "phone": "(509) 742-5448"}, {"is_active": false, "job_title": "Engineering Manager", "phone": "(504) 231-5010"}, {"is_active": true, "job_title": "Vice President", "phone": "(590) 203-5637"}, {"is_active": true, "job_title": "Engineering Manager", "phone": "(526) 600-9977"}, {"is_active": true, "job_title": "Sales Representative", "phone": "(530) 300-4356"}, {"is_active": true, "job_title": "Sales Representative", "phone": "(532) 591-2810"}, {"is_active": true, "job_title": "Engineering Manager", "phone": "(520) 374-0455"}], "dirty_cols": ["is_active", "job_title", "phone", "unnamed"], "clean_cols": ["is_active", "job_title", "phone"], "plan": {"dataset_summary": "15 rows × 4 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"VICE PRESIDENT": "Vice President", "vp": "Vice President", "Engineering Mgr": "Engineering Manager", "vice pres": "Vice President", "Vice Pres": "Vice President", "eng manager": "Engineering Manager", "engineering mgr": "Engineering Manager", "sales representative": "Sales Representative", "engineering manager": "Engineering Manager", "VICE PRES": "Vice President", "eng mgr": "Engineering Manager", "Rep": "Sales Representative"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"unit": "L", "city": "BUENG KUM", "rate": "17.7%", "job_title": "ceo", "email": "delbg@example.com", "department": "Fin", "col_x": ""}, {"unit": "MI", "city": "Bueng Kum", "rate": "72.2%", "job_title": "Acct Exec", "email": "GEBAJJ@MAIL.COM", "department": "PRODUCT", "col_x": ""}, {"unit": "pounds", "city": "   nyala", "rate": "37.7%", "job_title": "Accuont Executive", "email": "mbih@mail.com", "department": "Pm", "col_x": ""}, {"unit": "lb", "city": "Nyala   ", "rate": "70.8%", "job_title": "chief executive officer", "email": "ALPIGKB@CORP.IO", "department": "accounting", "col_x": ""}, {"unit": "\"  ", "city": "Nyala", "rate": "86.3%", "job_title": " c.e.o.", "email": " BNOLKLO@CORP.IO ", "department": "Fin", "col_x": ""}, {"unit": "MILLES", "city": "nyala", "rate": "96.3%", "job_title": "C.E.O.", "email": "HHIKEF@CORP.IO ", "department": "Prod", "col_x": ""}, {"unit": "Inches", "city": "Nyala", "rate": "95.6%", "job_title": "ae", "email": "CDGAO@EXAMPLE.COM", "department": "  Fin", "col_x": ""}, {"unit": "L", "city": "Nyala ", "rate": "0.2%", "job_title": "c.e.o.", "email": " objhmkg@corp.io ", "department": "Product", "col_x": ""}, {"unit": "inch ", "city": "Nyala", "rate": "28.7%", "job_title": "AE", "email": "   bdad@test.org", "department": "prod", "col_x": ""}, {"unit": " Mile ", "city": "  nyla", "rate": "41.6%", "job_title": "ae", "email": "ipmgm@mail.com", "department": "product", "col_x": ""}, {"unit": "mi ", "city": "BUENG KUM   ", "rate": "33.9%", "job_title": "  Chief Executive Officr", "email": "CLBHALP@TEST.ORG ", "department": "Fin", "col_x": ""}, {"unit": "MI", "city": " Nyala ", "rate": "22.6%", "job_title": "Chief Executive Officer", "email": "egaln@mail.com", "department": " Fin ", "col_x": ""}, {"unit": "IN", "city": "nyrla", "rate": "91.1%", "job_title": "CHIEF EXECUTIVE OFFICER", "email": "oemocde@example.com", "department": "fin", "col_x": ""}, {"unit": " Lb ", "city": "   Bueng Kum", "rate": "77.5%", "job_title": " CEO ", "email": "bholpg@example.com  ", "department": "Fin", "col_x": ""}, {"unit": "liters", "city": " BUENG KUM", "rate": "9.3%", "job_title": "C.E.O.", "email": " PANN@EXAMPLE.COM ", "department": "FIN", "col_x": ""}, {"unit": "in", "city": "BUENG KUM", "rate": "38.5%", "job_title": " Ae ", "email": "hlhm@corp.io", "department": " PM", "col_x": ""}, {"unit": "l", "city": "Nyala", "rate": "8.5%", "job_title": "   Ae", "email": "   AJMI@TEST.ORG", "department": "Accounting", "col_x": ""}, {"unit": "mile", "city": "bueng kum", "rate": "41.5%", "job_title": "chief executive officer", "email": " pddco@test.org ", "department": "Product", "col_x": ""}, {"unit": "inches  ", "city": "Bueng Kum", "rate": "88.8%", "job_title": "AE  ", "email": "elngfe@corp.io", "department": "accounting", "col_x": ""}, {"unit": "litre", "city": "Bueng Kum", "rate": "54.6%", "job_title": "chief executive officer", "email": "IIFIB@CORP.IO", "department": "prod", "col_x": ""}, {"unit": "l", "city": "nyaal", "rate": "10.3%", "job_title": "Account Executive", "email": "enphp@mail.com", "department": "Fin", "col_x": ""}, {"unit": "MILES", "city": " Nyala ", "rate": "79.1%", "job_title": "  AE", "email": "   nhmck@test.org", "department": "PM", "col_x": ""}, {"unit": "l", "city": "  Bueng Kum", "rate": "27.7%", "job_title": "AE", "email": "NJICNDP@EXAMPLE.COM", "department": "fin", "col_x": ""}, {"unit": " \"", "city": "Bueng Kum", "rate": "2.5%", "job_title": "CEO", "email": "   EBABDH@EXAMPLE.COM", "department": "Product   ", "col_x": ""}], "clean": [{"unit": "L", "city": "Bueng Kum", "rate": 0.177, "job_title": "Chief Executive Officer", "email": "delbg@example.com", "department": "Finance"}, {"unit": "mi", "city": "Bueng Kum", "rate": 0.722, "job_title": "Account Executive", "email": "gebajj@mail.com", "department": "Product"}, {"unit": "lb", "city": "Nyala", "rate": 0.377, "job_title": "Account Executive", "email": "mbih@mail.com", "department": "Product"}, {"unit": "lb", "city": "Nyala", "rate": 0.708, "job_title": "Chief Executive Officer", "email": "alpigkb@corp.io", "department": "Finance"}, {"unit": "in", "city": "Nyala", "rate": 0.863, "job_title": "Chief Executive Officer", "email": "bnolklo@corp.io", "department": "Finance"}, {"unit": "mi", "city": "Nyala", "rate": 0.963, "job_title": "Chief Executive Officer", "email": "hhikef@corp.io", "department": "Product"}, {"unit": "in", "city": "Nyala", "rate": 0.956, "job_title": "Account Executive", "email": "cdgao@example.com", "department": "Finance"}, {"unit": "L", "city": "Nyala", "rate": 0.002, "job_title": "Chief Executive Officer", "email": "objhmkg@corp.io", "department": "Product"}, {"unit": "in", "city": "Nyala", "rate": 0.287, "job_title": "Account Executive", "email": "bdad@test.org", "department": "Product"}, {"unit": "mi", "city": "Nyala", "rate": 0.41600000000000004, "job_title": "Account Executive", "email": "ipmgm@mail.com", "department": "Product"}, {"unit": "mi", "city": "Bueng Kum", "rate": 0.33899999999999997, "job_title": "Chief Executive Officer", "email": "clbhalp@test.org", "department": "Finance"}, {"unit": "mi", "city": "Nyala", "rate": 0.226, "job_title": "Chief Executive Officer", "email": "egaln@mail.com", "department": "Finance"}, {"unit": "in", "city": "Nyala", "rate": 0.9109999999999999, "job_title": "Chief Executive Officer", "email": "oemocde@example.com", "department": "Finance"}, {"unit": "lb", "city": "Bueng Kum", "rate": 0.775, "job_title": "Chief Executive Officer", "email": "bholpg@example.com", "department": "Finance"}, {"unit": "L", "city": "Bueng Kum", "rate": 0.09300000000000001, "job_title": "Chief Executive Officer", "email": "pann@example.com", "department": "Finance"}, {"unit": "in", "city": "Bueng Kum", "rate": 0.385, "job_title": "Account Executive", "email": "hlhm@corp.io", "department": "Product"}, {"unit": "L", "city": "Nyala", "rate": 0.085, "job_title": "Account Executive", "email": "ajmi@test.org", "department": "Finance"}, {"unit": "mi", "city": "Bueng Kum", "rate": 0.415, "job_title": "Chief Executive Officer", "email": "pddco@test.org", "department": "Product"}, {"unit": "in", "city": "Bueng Kum", "rate": 0.888, "job_title": "Account Executive", "email": "elngfe@corp.io", "department": "Finance"}, {"unit": "L", "city": "Bueng Kum", "rate": 0.546, "job_title": "Chief Executive Officer", "email": "iifib@corp.io", "department": "Product"}, {"unit": "L", "city": "Nyala", "rate": 0.10300000000000001, "job_title": "Account Executive", "email": "enphp@mail.com", "department": "Finance"}, {"unit": "mi", "city": "Nyala", "rate": 0.7909999999999999, "job_title": "Account Executive", "email": "nhmck@test.org", "department": "Product"}, {"unit": "L", "city": "Bueng Kum", "rate": 0.27699999999999997, "job_title": "Account Executive", "email": "njicndp@example.com", "department": "Finance"}, {"unit": "in", "city": "Bueng Kum", "rate": 0.025, "job_title": "Chief Executive Officer", "email": "ebabdh@example.com", "department": "Product"}], "dirty_cols": ["unit", "city", "rate", "job_title", "email", "department", "col_x"], "clean_cols": ["unit", "city", "rate", "job_title", "email", "department"], "plan": {"dataset_summary": "24 rows × 7 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MI": "mi", "pounds": "lb", "\"": "in", "MILLES": "mi", "Inches": "in", "inch": "in", "Mile": "mi", "IN": "in", "Lb": "lb", "liters": "L", "l": "L", "mile": "mi", "inches": "in", "litre": "L", "MILES": "mi"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BUENG KUM": "Bueng Kum", "nyala": "Nyala", "nyla": "Nyala", "nyrla": "Nyala", "bueng kum": "Bueng Kum", "nyaal": "Nyala"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ceo": "Chief Executive Officer", "Acct Exec": "Account Executive", "Accuont Executive": "Account Executive", "chief executive officer": "Chief Executive Officer", "c.e.o.": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer", "ae": "Account Executive", "AE": "Account Executive", "Chief Executive Officr": "Chief Executive Officer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "CEO": "Chief Executive Officer", "Ae": "Account Executive"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fin": "Finance", "PRODUCT": "Product", "Pm": "Product", "accounting": "Finance", "Prod": "Product", "prod": "Product", "product": "Product", "fin": "Finance", "FIN": "Finance", "PM": "Product", "Accounting": "Finance"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "2023-09-19", "is_active": "0", "job_title": "Sr Engineer", "name": "Eve Johnson", "extra": ""}, {"signup_date": "2/2/2023", "is_active": "false", "job_title": "Chief Executive Officer", "name": "Grace Wong   ", "extra": ""}, {"signup_date": "2023-05-02", "is_active": "FALSE", "job_title": "sr  engineer", "name": "Eve Ali ", "extra": ""}, {"signup_date": "12 Jun 2023", "is_active": "TRUE", "job_title": "Administrative Assistant", "name": "Heidi  Lee", "extra": ""}, {"signup_date": "45264", "is_active": "No", "job_title": "  ceo", "name": "Grace Brandt", "extra": ""}, {"signup_date": "8/18/2023", "is_active": "N", "job_title": "Engineering Manager", "name": "  Mona Johnson", "extra": ""}, {"signup_date": "45119", "is_active": "0", "job_title": "Admin", "name": "David Diaz   ", "extra": ""}, {"signup_date": "2/21/2023", "is_active": "Y", "job_title": "Senior Eng", "name": "Alice Lee", "extra": ""}, {"signup_date": "4/3/2023", "is_active": "F", "job_title": "Snr Engineer", "name": "   Mona Ali", "extra": ""}, {"signup_date": "25 Aug 2023", "is_active": "FALSE", "job_title": "Sr Engineer", "name": "Omar Khan  ", "extra": ""}, {"signup_date": "11/7/2023", "is_active": "FALSE", "job_title": "   Sr Engineer", "name": "Karl Fischer  ", "extra": ""}, {"signup_date": "8/7/2023", "is_active": "1", "job_title": "administrative assistant", "name": "Carol Novak ", "extra": ""}], "clean": [{"signup_date": "2023-09-19", "is_active": false, "job_title": "Senior Engineer", "name": "Eve Johnson"}, {"signup_date": "2023-02-02", "is_active": false, "job_title": "Chief Executive Officer", "name": "Grace Wong"}, {"signup_date": "2023-05-02", "is_active": false, "job_title": "Senior Engineer", "name": "Eve Ali"}, {"signup_date": "2023-06-12", "is_active": true, "job_title": "Administrative Assistant", "name": "Heidi Lee"}, {"signup_date": "2023-12-04", "is_active": false, "job_title": "Chief Executive Officer", "name": "Grace Brandt"}, {"signup_date": "2023-08-18", "is_active": false, "job_title": "Engineering Manager", "name": "Mona Johnson"}, {"signup_date": "2023-07-12", "is_active": false, "job_title": "Administrative Assistant", "name": "David Diaz"}, {"signup_date": "2023-02-21", "is_active": true, "job_title": "Senior Engineer", "name": "Alice Lee"}, {"signup_date": "2023-04-03", "is_active": false, "job_title": "Senior Engineer", "name": "Mona Ali"}, {"signup_date": "2023-08-25", "is_active": false, "job_title": "Senior Engineer", "name": "Omar Khan"}, {"signup_date": "2023-11-07", "is_active": false, "job_title": "Senior Engineer", "name": "Karl Fischer"}, {"signup_date": "2023-08-07", "is_active": true, "job_title": "Administrative Assistant", "name": "Carol Novak"}], "dirty_cols": ["signup_date", "is_active", "job_title", "name", "extra"], "clean_cols": ["signup_date", "is_active", "job_title", "name"], "plan": {"dataset_summary": "12 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Sr Engineer": "Senior Engineer", "sr engineer": "Senior Engineer", "ceo": "Chief Executive Officer", "Admin": "Administrative Assistant", "Senior Eng": "Senior Engineer", "Snr Engineer": "Senior Engineer", "administrative assistant": "Administrative Assistant"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"city": "   Valparai", "name": "Lena Adams", "job_title": "Eng Manfger", "signup_date": "2023-11-16", "status": "l", "department": "pm"}, {"city": "Rees   ", "name": "Mona Brandt", "job_title": "Engg Mgr", "signup_date": "2023-02-14", "status": " P1 ", "department": "CS"}, {"city": "Rees", "name": "Eve  Ali", "job_title": "Eng Manager", "signup_date": "23 Sep 2023", "status": "High", "department": "PM"}, {"city": "Valparai", "name": "Sara Cruz", "job_title": "v.p.", "signup_date": "5 Sep 2023", "status": "high", "department": "Prod"}, {"city": "valparai", "name": "Eve Lee   ", "job_title": " Engineering Mgr", "signup_date": "2023-07-09", "status": "H", "department": "prod"}, {"city": " Valparai ", "name": "Frank Fischer ", "job_title": "ENGINEERING MANAGER", "signup_date": "2023-09-02", "status": "  p3", "department": "  PM"}, {"city": "REES", "name": "Carol Moore", "job_title": "VP", "signup_date": "25 Dec 2023", "status": "P3", "department": "Product"}, {"city": "Valparai", "name": "Alice Lee", "job_title": "eng manager", "signup_date": "45125", "status": "L", "department": "it"}, {"city": "REES", "name": "Ivan Brandt", "job_title": "Engineering Manager   ", "signup_date": "2023-05-07", "status": "L", "department": " information technology"}, {"city": "Valparai", "name": "Grace Khan", "job_title": "vice pres", "signup_date": "2023-08-09", "status": " P3", "department": "prod"}, {"city": "Tam  Giang", "name": "Priya Ali  ", "job_title": "vp", "signup_date": "2023-11-27", "status": "high", "department": "Product"}, {"city": "Valparai", "name": "Frank Reyes  ", "job_title": "eng mnaager", "signup_date": "3 Jun 2023", "status": "high", "department": "Information Technology"}, {"city": "Valparrai", "name": "Judy  Lee", "job_title": "vp", "signup_date": "45031", "status": "   p3", "department": "CUST SUPPORT"}, {"city": "valparai", "name": "   Ivan Ali", "job_title": "Engineering Manager", "signup_date": "22 Jun 2023", "status": "high", "department": "Customer Support"}, {"city": "Sheoganj", "name": "Alice Diaz", "job_title": "Engineerng Mgr", "signup_date": "45051", "status": "   HIGH", "department": "CS"}, {"city": "VALPARAI", "name": "Ivan Cruz   ", "job_title": "VICE  PRES", "signup_date": "44941", "status": "HIGH", "department": "Product"}, {"city": "Sheoganj", "name": "   Lena Khan", "job_title": " VP ", "signup_date": "2023-03-18", "status": "HIGH", "department": "INFORMATION  TECHNOLOGY"}, {"city": "TAM GIANG   ", "name": "Ivan Adams", "job_title": "engineering manager", "signup_date": "2023-04-20", "status": "  HIGH", "department": "Information Technology"}], "clean": [{"city": "Valparai", "name": "Lena Adams", "job_title": "Engineering Manager", "signup_date": "2023-11-16", "status": "Low", "department": "Product"}, {"city": "Rees", "name": "Mona Brandt", "job_title": "Engineering Manager", "signup_date": "2023-02-14", "status": "High", "department": "Customer Support"}, {"city": "Rees", "name": "Eve Ali", "job_title": "Engineering Manager", "signup_date": "2023-09-23", "status": "High", "department": "Product"}, {"city": "Valparai", "name": "Sara Cruz", "job_title": "Vice President", "signup_date": "2023-09-05", "status": "High", "department": "Product"}, {"city": "Valparai", "name": "Eve Lee", "job_title": "Engineering Manager", "signup_date": "2023-07-09", "status": "High", "department": "Product"}, {"city": "Valparai", "name": "Frank Fischer", "job_title": "Engineering Manager", "signup_date": "2023-09-02", "status": "Low", "department": "Product"}, {"city": "Rees", "name": "Carol Moore", "job_title": "Vice President", "signup_date": "2023-12-25", "status": "Low", "department": "Product"}, {"city": "Valparai", "name": "Alice Lee", "job_title": "Engineering Manager", "signup_date": "2023-07-18", "status": "Low", "department": "Information Technology"}, {"city": "Rees", "name": "Ivan Brandt", "job_title": "Engineering Manager", "signup_date": "2023-05-07", "status": "Low", "department": "Information Technology"}, {"city": "Valparai", "name": "Grace Khan", "job_title": "Vice President", "signup_date": "2023-08-09", "status": "Low", "department": "Product"}, {"city": "Tam Giang", "name": "Priya Ali", "job_title": "Vice President", "signup_date": "2023-11-27", "status": "High", "department": "Product"}, {"city": "Valparai", "name": "Frank Reyes", "job_title": "Engineering Manager", "signup_date": "2023-06-03", "status": "High", "department": "Information Technology"}, {"city": "Valparai", "name": "Judy Lee", "job_title": "Vice President", "signup_date": "2023-04-15", "status": "Low", "department": "Customer Support"}, {"city": "Valparai", "name": "Ivan Ali", "job_title": "Engineering Manager", "signup_date": "2023-06-22", "status": "High", "department": "Customer Support"}, {"city": "Sheoganj", "name": "Alice Diaz", "job_title": "Engineering Manager", "signup_date": "2023-05-05", "status": "High", "department": "Customer Support"}, {"city": "Valparai", "name": "Ivan Cruz", "job_title": "Vice President", "signup_date": "2023-01-15", "status": "High", "department": "Product"}, {"city": "Sheoganj", "name": "Lena Khan", "job_title": "Vice President", "signup_date": "2023-03-18", "status": "High", "department": "Information Technology"}, {"city": "Tam Giang", "name": "Ivan Adams", "job_title": "Engineering Manager", "signup_date": "2023-04-20", "status": "High", "department": "Information Technology"}], "dirty_cols": ["city", "name", "job_title", "signup_date", "status", "department"], "clean_cols": ["city", "name", "job_title", "signup_date", "status", "department"], "plan": {"dataset_summary": "18 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"valparai": "Valparai", "REES": "Rees", "Valparrai": "Valparai", "VALPARAI": "Valparai", "TAM GIANG": "Tam Giang"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Eng Manfger": "Engineering Manager", "Engg Mgr": "Engineering Manager", "Eng Manager": "Engineering Manager", "v.p.": "Vice President", "Engineering Mgr": "Engineering Manager", "ENGINEERING MANAGER": "Engineering Manager", "VP": "Vice President", "eng manager": "Engineering Manager", "vice pres": "Vice President", "vp": "Vice President", "eng mnaager": "Engineering Manager", "Engineerng Mgr": "Engineering Manager", "VICE PRES": "Vice President", "engineering manager": "Engineering Manager"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"l": "Low", "P1": "High", "high": "High", "H": "High", "p3": "Low", "P3": "Low", "L": "Low", "HIGH": "High"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pm": "Product", "CS": "Customer Support", "PM": "Product", "Prod": "Product", "prod": "Product", "it": "Information Technology", "information technology": "Information Technology", "CUST SUPPORT": "Customer Support", "INFORMATION TECHNOLOGY": "Information Technology"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "false", "company": "Globex", "amount": "$6,629.66", "name": "David Petrov"}, {"is_active": "FALSE", "company": "Stark Industries", "amount": "3.245,34", "name": "Priya Moore"}, {"is_active": "F", "company": "Acme Inc", "amount": "3.291,87", "name": "   Mona Johnson"}, {"is_active": "FALSE", "company": "Acme Inc", "amount": "1.074,09", "name": " Lena Reyes"}, {"is_active": "F", "company": "Cyberdyne", "amount": "8.857,36", "name": "Eve Diaz  "}, {"is_active": "false", "company": "Acme Inc", "amount": "N/A", "name": "Lena Ali"}, {"is_active": "0", "company": "Soylent Corp   ", "amount": "$7,590.52", "name": "Lena Adams"}, {"is_active": "0", "company": "Cyberdyne", "amount": "1.973,52", "name": "Ivan Adams"}, {"is_active": "TRUE", "company": "Umbrella", "amount": "6.678,94", "name": "  Sara Fischer"}, {"is_active": "TRUE", "company": " Globex", "amount": "$7,606.79", "name": "Carol Lee"}, {"is_active": "FALSE", "company": "Vehement", "amount": "6.973,60", "name": "Bob Johnson"}, {"is_active": "F", "company": "Hooli", "amount": "6.002,58", "name": "Judy  Fischer"}, {"is_active": "true", "company": "   Wonka Co", "amount": "$6,385.94", "name": "David  Reyes"}, {"is_active": "FALSE", "company": " Globex", "amount": "$6,936.44", "name": "Lena Brandt"}, {"is_active": "FALSE", "company": "Vehement  ", "amount": "2.167,15", "name": "Carol Ali"}, {"is_active": "Y", "company": "Umbrella", "amount": "$1,896.54", "name": "Sara Diaz  "}, {"is_active": "false", "company": "  Vehement", "amount": "$5,379.59", "name": "Eve Cruz   "}, {"is_active": "0", "company": "Globex", "amount": "2.758,42", "name": "Judy Park"}, {"is_active": "N", "company": "Hooli", "amount": "5.210,42", "name": "Mona  Smith"}, {"is_active": "Y", "company": "Stark Industries", "amount": "-", "name": "Grace Ali"}, {"is_active": "Y", "company": "Soylent  Corp", "amount": "5.752,09", "name": "Alice Brandt"}, {"is_active": "Y", "company": "Initech", "amount": "$8,156.10", "name": "  David Novak"}], "clean": [{"is_active": false, "company": "Globex", "amount": 6629.66, "name": "David Petrov"}, {"is_active": false, "company": "Stark Industries", "amount": 3245.34, "name": "Priya Moore"}, {"is_active": false, "company": "Acme Inc", "amount": 3291.87, "name": "Mona Johnson"}, {"is_active": false, "company": "Acme Inc", "amount": 1074.09, "name": "Lena Reyes"}, {"is_active": false, "company": "Cyberdyne", "amount": 8857.36, "name": "Eve Diaz"}, {"is_active": false, "company": "Acme Inc", "amount": NaN, "name": "Lena Ali"}, {"is_active": false, "company": "Soylent Corp", "amount": 7590.52, "name": "Lena Adams"}, {"is_active": false, "company": "Cyberdyne", "amount": 1973.52, "name": "Ivan Adams"}, {"is_active": true, "company": "Umbrella", "amount": 6678.94, "name": "Sara Fischer"}, {"is_active": true, "company": "Globex", "amount": 7606.79, "name": "Carol Lee"}, {"is_active": false, "company": "Vehement", "amount": 6973.6, "name": "Bob Johnson"}, {"is_active": false, "company": "Hooli", "amount": 6002.58, "name": "Judy Fischer"}, {"is_active": true, "company": "Wonka Co", "amount": 6385.94, "name": "David Reyes"}, {"is_active": false, "company": "Globex", "amount": 6936.44, "name": "Lena Brandt"}, {"is_active": false, "company": "Vehement", "amount": 2167.15, "name": "Carol Ali"}, {"is_active": true, "company": "Umbrella", "amount": 1896.54, "name": "Sara Diaz"}, {"is_active": false, "company": "Vehement", "amount": 5379.59, "name": "Eve Cruz"}, {"is_active": false, "company": "Globex", "amount": 2758.42, "name": "Judy Park"}, {"is_active": false, "company": "Hooli", "amount": 5210.42, "name": "Mona Smith"}, {"is_active": true, "company": "Stark Industries", "amount": NaN, "name": "Grace Ali"}, {"is_active": true, "company": "Soylent Corp", "amount": 5752.09, "name": "Alice Brandt"}, {"is_active": true, "company": "Initech", "amount": 8156.1, "name": "David Novak"}], "dirty_cols": ["is_active", "company", "amount", "name"], "clean_cols": ["is_active", "company", "amount", "name"], "plan": {"dataset_summary": "22 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"is_active": "TRUE", "phone": "556-910-9569", "state": "mp", "amount": "$409.08", "unit": " kilograms ", "status": "pedning", "country": "LY", "col_x": ""}, {"is_active": "0", "phone": "(565)982-6679", "state": "south dakota", "amount": "$8,674.94", "unit": "kilogram", "status": "OVERDUE", "country": "Republic kf Bosnia and Herzegovina", "col_x": ""}, {"is_active": "T", "phone": "(555)672-6016", "state": "NORTHERN MARIANA ISLANDS", "amount": "730,27", "unit": "POUNDS", "status": " late", "country": "Bosnia And Herzegovina", "col_x": ""}, {"is_active": "true", "phone": "(526)204-2371", "state": "   South Dakota", "amount": "3.309,84", "unit": " lb ", "status": "overdue", "country": " LY ", "col_x": ""}, {"is_active": "No", "phone": "(566)979-9062", "state": "Sd", "amount": "$535.98", "unit": "KM", "status": "refuunded", "country": "  Libya", "col_x": ""}, {"is_active": "false", "phone": "5224431955", "state": "mp", "amount": "$2,763.73", "unit": "Kilogram", "status": "refunded", "country": "bosnia and herzegovina", "col_x": ""}, {"is_active": "0", "phone": "5840545900", "state": "sd", "amount": "2.922,14", "unit": "kilometer", "status": "Ovedrue", "country": "bh", "col_x": ""}, {"is_active": "0", "phone": "(595)996-7085", "state": "SD", "amount": "7.734,36", "unit": "kilometer", "status": "Overdue", "country": "Bahrain", "col_x": ""}, {"is_active": "1", "phone": "5365167297", "state": "Northern Mariana Islands", "amount": "$6,076.24", "unit": "Pounds", "status": "refund", "country": "Bahrain", "col_x": ""}, {"is_active": "Yes", "phone": "(573)945-2132", "state": "northern mariana islands", "amount": "6.909,91", "unit": "km", "status": "pending", "country": "Kingdom  of Bahrain", "col_x": ""}, {"is_active": "TRUE", "phone": "5836735242", "state": "mp", "amount": "$5,033.90", "unit": "lb", "status": "reffunded", "country": "Kingdom of Bahrain", "col_x": ""}, {"is_active": "true", "phone": "530.073.0133", "state": "Northern Mariana Islands ", "amount": "1.080,98", "unit": "km", "status": "Paid", "country": "lby", "col_x": ""}, {"is_active": "N", "phone": "(548)163-5489", "state": "northern mariana ijlands", "amount": "$851.53", "unit": "KM", "status": "PENDING   ", "country": "LBY", "col_x": ""}, {"is_active": "false", "phone": "5514269124", "state": "South Dakota", "amount": "$89.82", "unit": "Km", "status": "refunded", "country": "Kingdom of Bahrain", "col_x": ""}, {"is_active": "false", "phone": "(596)530-8184", "state": "Northern Mariana Islands", "amount": "2.450,48", "unit": "kilograms", "status": "REFUNDED", "country": "   Libya", "col_x": ""}, {"is_active": "1", "phone": "534-199-0765", "state": "Sd", "amount": "?", "unit": " Kilogrm", "status": "Pending", "country": "Bosnia and Herzegovina  ", "col_x": ""}, {"is_active": "false", "phone": "553-695-9263", "state": "Northern Mariana Islands", "amount": "$1,379.21", "unit": "Lb", "status": " refunded", "country": "Lby", "col_x": ""}, {"is_active": "true", "phone": "531.887.5349", "state": "SD", "amount": "8.964,01", "unit": "  Km", "status": "Paid", "country": "LBY", "col_x": ""}, {"is_active": "T", "phone": "502-554-1644", "state": "MP", "amount": "6.341,35", "unit": " Lbs ", "status": "refunded", "country": "  kingdhm of bahrain", "col_x": ""}, {"is_active": "false", "phone": "5109943444", "state": "south dakota", "amount": "2.782,32", "unit": "kilograms", "status": "late ", "country": "bosnia and herzegovina", "col_x": ""}, {"is_active": "true", "phone": "535-290-2444", "state": "SOTH DAKOTA", "amount": "$6,609.80", "unit": "kg", "status": "OVEDUE", "country": "BIH", "col_x": ""}, {"is_active": "T", "phone": "(583)279-3773", "state": "South Dakota", "amount": "N/A", "unit": "kipometers", "status": " refunded ", "country": "BHR", "col_x": ""}, {"is_active": "Yes", "phone": "(551)223-9448", "state": "South Dakoa", "amount": "2.204,95", "unit": "kg", "status": "PAD ", "country": "Lby", "col_x": ""}, {"is_active": "false", "phone": "5189008687", "state": "South Daktoa", "amount": "$4,014.86", "unit": "kg", "status": "Overdue", "country": "ba", "col_x": ""}, {"is_active": "Yes", "phone": "599.839.4546", "state": "SD", "amount": "7.192,89", "unit": "Kg", "status": "LATE", "country": "Bosnia And Herzegovina", "col_x": ""}, {"is_active": "0", "phone": "533-144-2238", "state": "South Dakota", "amount": "TBD", "unit": "Kg", "status": "due   ", "country": "republic of bosnia and herzegovina", "col_x": ""}, {"is_active": "0", "phone": "543.117.1131", "state": "MP  ", "amount": "2.210,41", "unit": " km ", "status": "PENDING", "country": "Libya", "col_x": ""}, {"is_active": "true", "phone": "(562)872-3145", "state": "northern mariana  islands", "amount": "7.813,92", "unit": "kg", "status": "Due", "country": " bh ", "col_x": ""}, {"is_active": "", "phone": "", "state": "", "amount": "", "unit": "", "status": "", "country": "", "col_x": ""}], "clean": [{"is_active": true, "phone": "(556) 910-9569", "state": "Northern Mariana Islands", "amount": 409.08, "unit": "kg", "status": "Pending", "country": "Libya"}, {"is_active": false, "phone": "(565) 982-6679", "state": "South Dakota", "amount": 8674.94, "unit": "kg", "status": "Overdue", "country": "Bosnia and Herzegovina"}, {"is_active": true, "phone": "(555) 672-6016", "state": "Northern Mariana Islands", "amount": 730.27, "unit": "lb", "status": "Overdue", "country": "Bosnia and Herzegovina"}, {"is_active": true, "phone": "(526) 204-2371", "state": "South Dakota", "amount": 3309.84, "unit": "lb", "status": "Overdue", "country": "Libya"}, {"is_active": false, "phone": "(566) 979-9062", "state": "South Dakota", "amount": 535.98, "unit": "km", "status": "Refunded", "country": "Libya"}, {"is_active": false, "phone": "(522) 443-1955", "state": "Northern Mariana Islands", "amount": 2763.73, "unit": "kg", "status": "Refunded", "country": "Bosnia and Herzegovina"}, {"is_active": false, "phone": "(584) 054-5900", "state": "South Dakota", "amount": 2922.14, "unit": "km", "status": "Overdue", "country": "Bahrain"}, {"is_active": false, "phone": "(595) 996-7085", "state": "South Dakota", "amount": 7734.36, "unit": "km", "status": "Overdue", "country": "Bahrain"}, {"is_active": true, "phone": "(536) 516-7297", "state": "Northern Mariana Islands", "amount": 6076.24, "unit": "lb", "status": "Refunded", "country": "Bahrain"}, {"is_active": true, "phone": "(573) 945-2132", "state": "Northern Mariana Islands", "amount": 6909.91, "unit": "km", "status": "Pending", "country": "Bahrain"}, {"is_active": true, "phone": "(583) 673-5242", "state": "Northern Mariana Islands", "amount": 5033.9, "unit": "lb", "status": "Refunded", "country": "Bahrain"}, {"is_active": true, "phone": "(530) 073-0133", "state": "Northern Mariana Islands", "amount": 1080.98, "unit": "km", "status": "Paid", "country": "Libya"}, {"is_active": false, "phone": "(548) 163-5489", "state": "Northern Mariana Islands", "amount": 851.53, "unit": "km", "status": "Pending", "country": "Libya"}, {"is_active": false, "phone": "(551) 426-9124", "state": "South Dakota", "amount": 89.82, "unit": "km", "status": "Refunded", "country": "Bahrain"}, {"is_active": false, "phone": "(596) 530-8184", "state": "Northern Mariana Islands", "amount": 2450.48, "unit": "kg", "status": "Refunded", "country": "Libya"}, {"is_active": true, "phone": "(534) 199-0765", "state": "South Dakota", "amount": NaN, "unit": "kg", "status": "Pending", "country": "Bosnia and Herzegovina"}, {"is_active": false, "phone": "(553) 695-9263", "state": "Northern Mariana Islands", "amount": 1379.21, "unit": "lb", "status": "Refunded", "country": "Libya"}, {"is_active": true, "phone": "(531) 887-5349", "state": "South Dakota", "amount": 8964.01, "unit": "km", "status": "Paid", "country": "Libya"}, {"is_active": true, "phone": "(502) 554-1644", "state": "Northern Mariana Islands", "amount": 6341.35, "unit": "lb", "status": "Refunded", "country": "Bahrain"}, {"is_active": false, "phone": "(510) 994-3444", "state": "South Dakota", "amount": 2782.32, "unit": "kg", "status": "Overdue", "country": "Bosnia and Herzegovina"}, {"is_active": true, "phone": "(535) 290-2444", "state": "South Dakota", "amount": 6609.8, "unit": "kg", "status": "Overdue", "country": "Bosnia and Herzegovina"}, {"is_active": true, "phone": "(583) 279-3773", "state": "South Dakota", "amount": NaN, "unit": "km", "status": "Refunded", "country": "Bahrain"}, {"is_active": true, "phone": "(551) 223-9448", "state": "South Dakota", "amount": 2204.95, "unit": "kg", "status": "Paid", "country": "Libya"}, {"is_active": false, "phone": "(518) 900-8687", "state": "South Dakota", "amount": 4014.86, "unit": "kg", "status": "Overdue", "country": "Bosnia and Herzegovina"}, {"is_active": true, "phone": "(599) 839-4546", "state": "South Dakota", "amount": 7192.89, "unit": "kg", "status": "Overdue", "country": "Bosnia and Herzegovina"}, {"is_active": false, "phone": "(533) 144-2238", "state": "South Dakota", "amount": NaN, "unit": "kg", "status": "Pending", "country": "Bosnia and Herzegovina"}, {"is_active": false, "phone": "(543) 117-1131", "state": "Northern Mariana Islands", "amount": 2210.41, "unit": "km", "status": "Pending", "country": "Libya"}, {"is_active": true, "phone": "(562) 872-3145", "state": "Northern Mariana Islands", "amount": 7813.92, "unit": "kg", "status": "Pending", "country": "Bahrain"}], "dirty_cols": ["is_active", "phone", "state", "amount", "unit", "status", "country", "col_x"], "clean_cols": ["is_active", "phone", "state", "amount", "unit", "status", "country"], "plan": {"dataset_summary": "29 rows × 8 columns. 7 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mp": "Northern Mariana Islands", "south dakota": "South Dakota", "NORTHERN MARIANA ISLANDS": "Northern Mariana Islands", "Sd": "South Dakota", "sd": "South Dakota", "SD": "South Dakota", "northern mariana islands": "Northern Mariana Islands", "northern mariana ijlands": "Northern Mariana Islands", "MP": "Northern Mariana Islands", "SOTH DAKOTA": "South Dakota", "South Dakoa": "South Dakota", "South Daktoa": "South Dakota"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilograms": "kg", "kilogram": "kg", "POUNDS": "lb", "KM": "km", "Kilogram": "kg", "kilometer": "km", "Pounds": "lb", "Km": "km", "Kilogrm": "kg", "Lb": "lb", "Lbs": "lb", "kipometers": "km", "Kg": "kg"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pedning": "Pending", "OVERDUE": "Overdue", "late": "Overdue", "overdue": "Overdue", "refuunded": "Refunded", "refunded": "Refunded", "Ovedrue": "Overdue", "refund": "Refunded", "pending": "Pending", "reffunded": "Refunded", "PENDING": "Pending", "REFUNDED": "Refunded", "OVEDUE": "Overdue", "PAD": "Paid", "LATE": "Overdue", "due": "Pending", "Due": "Pending"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LY": "Libya", "Republic kf Bosnia and Herzegovina": "Bosnia and Herzegovina", "Bosnia And Herzegovina": "Bosnia and Herzegovina", "bosnia and herzegovina": "Bosnia and Herzegovina", "bh": "Bahrain", "Kingdom of Bahrain": "Bahrain", "lby": "Libya", "LBY": "Libya", "Lby": "Libya", "kingdhm of bahrain": "Bahrain", "BIH": "Bosnia and Herzegovina", "BHR": "Bahrain", "ba": "Bosnia and Herzegovina", "republic of bosnia and herzegovina": "Bosnia and Herzegovina"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "5606393549", "currency": " btn ", "industry": "Education", "state": "south dakota", "city": " itajobi ", "company": "   Acme Inc"}, {"phone": "517.790.0907", "currency": "armenian dram", "industry": "  MEDICL", "state": "SD", "city": "Es Senia", "company": "   Acme Inc"}, {"phone": "565-540-2824", "currency": "Brunei Dollar", "industry": "EdTech", "state": "Utah", "city": "  Danjiangkou", "company": " Cyberdyne"}, {"phone": "5900074766", "currency": "AMD", "industry": "Medical", "state": "california", "city": "Itajobi", "company": "Stark Industries"}, {"phone": "5273770646", "currency": "  Brunei Dollar", "industry": "education", "state": "SD", "city": "itajobi", "company": "Vehement"}, {"phone": "508.440.8589", "currency": "Armenian Dram", "industry": "medical", "state": "SD", "city": " malema ", "company": "Umbrella"}, {"phone": "(569)799-0520", "currency": "djibouti franc   ", "industry": "education", "state": "california", "city": "Es Senia", "company": "Globex"}, {"phone": "(561)302-0405", "currency": "AMD", "industry": "Medical", "state": "  utah", "city": " Malema", "company": "Vehement"}, {"phone": "560.325.9400", "currency": "BTN  ", "industry": "oil & gas", "state": "sd", "city": "Danjiangkou", "company": "Vehement"}, {"phone": "(534)167-6923", "currency": "  amd", "industry": "Utilities", "state": " ca", "city": " Danjiangkou ", "company": "Initech"}, {"phone": "550-871-2397", "currency": "armenhan dram  ", "industry": " Ed", "state": " CALIFORNIA ", "city": "   Itajobi", "company": "Wonka Co"}, {"phone": "(525)278-3216", "currency": "djf", "industry": "Utilities", "state": "  South Dakota", "city": "Itajobi", "company": "Stark Industries"}, {"phone": "(585)219-2180", "currency": "BND   ", "industry": "Health Care", "state": "UTAH  ", "city": "MALEMA", "company": " Vehement "}, {"phone": "5606393549", "currency": " btn ", "industry": "Education", "state": "south dakota", "city": " itajobi ", "company": "   Acme Inc"}, {"phone": "5273770646", "currency": "  Brunei Dollar", "industry": "education", "state": "SD", "city": "itajobi", "company": "Vehement"}], "clean": [{"phone": "(560) 639-3549", "currency": "BTN", "industry": "Education", "state": "South Dakota", "city": "Itajobi", "company": "Acme Inc"}, {"phone": "(517) 790-0907", "currency": "AMD", "industry": "Healthcare", "state": "South Dakota", "city": "Es Senia", "company": "Acme Inc"}, {"phone": "(565) 540-2824", "currency": "BND", "industry": "Education", "state": "Utah", "city": "Danjiangkou", "company": "Cyberdyne"}, {"phone": "(590) 007-4766", "currency": "AMD", "industry": "Healthcare", "state": "California", "city": "Itajobi", "company": "Stark Industries"}, {"phone": "(527) 377-0646", "currency": "BND", "industry": "Education", "state": "South Dakota", "city": "Itajobi", "company": "Vehement"}, {"phone": "(508) 440-8589", "currency": "AMD", "industry": "Healthcare", "state": "South Dakota", "city": "Malema", "company": "Umbrella"}, {"phone": "(569) 799-0520", "currency": "DJF", "industry": "Education", "state": "California", "city": "Es Senia", "company": "Globex"}, {"phone": "(561) 302-0405", "currency": "AMD", "industry": "Healthcare", "state": "Utah", "city": "Malema", "company": "Vehement"}, {"phone": "(560) 325-9400", "currency": "BTN", "industry": "Energy", "state": "South Dakota", "city": "Danjiangkou", "company": "Vehement"}, {"phone": "(534) 167-6923", "currency": "AMD", "industry": "Energy", "state": "California", "city": "Danjiangkou", "company": "Initech"}, {"phone": "(550) 871-2397", "currency": "AMD", "industry": "Education", "state": "California", "city": "Itajobi", "company": "Wonka Co"}, {"phone": "(525) 278-3216", "currency": "DJF", "industry": "Energy", "state": "South Dakota", "city": "Itajobi", "company": "Stark Industries"}, {"phone": "(585) 219-2180", "currency": "BND", "industry": "Healthcare", "state": "Utah", "city": "Malema", "company": "Vehement"}], "dirty_cols": ["phone", "currency", "industry", "state", "city", "company"], "clean_cols": ["phone", "currency", "industry", "state", "city", "company"], "plan": {"dataset_summary": "15 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"btn": "BTN", "armenian dram": "AMD", "Brunei Dollar": "BND", "Armenian Dram": "AMD", "djibouti franc": "DJF", "amd": "AMD", "armenhan dram": "AMD", "djf": "DJF"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MEDICL": "Healthcare", "EdTech": "Education", "Medical": "Healthcare", "education": "Education", "medical": "Healthcare", "oil & gas": "Energy", "Utilities": "Energy", "Ed": "Education", "Health Care": "Healthcare"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"south dakota": "South Dakota", "SD": "South Dakota", "california": "California", "utah": "Utah", "sd": "South Dakota", "ca": "California", "CALIFORNIA": "California", "UTAH": "Utah"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"itajobi": "Itajobi", "malema": "Malema", "MALEMA": "Malema"}, "rationale": "Unified 3 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"country": "Wallis and Futuna", "rate": "0.3%", "name": "Karl Wong", "company": " Globex ", "job_title": "sales representative", "city": "pijijiapan"}, {"country": "TKM", "rate": "18.2%", "name": "Eve Diaz", "company": " Stark Industries", "job_title": "ENG MANAGEER", "city": "Pijijiapan"}, {"country": "Tkm", "rate": "96.2%", "name": "Bob Johnson", "company": "  Acme Inc", "job_title": "ceo", "city": "Taquari"}, {"country": "WLF", "rate": "87.4%", "name": "Bob  Lee", "company": "Hooli  ", "job_title": "chief executivve officer", "city": "Taqquari"}, {"country": "mm", "rate": "3.2%", "name": "Sara Cruz", "company": "  Soylent Corp", "job_title": "C.E.O.", "city": "Pijijiapan  "}, {"country": "TKM   ", "rate": "66.3%", "name": "Karl Johnson", "company": " Initech ", "job_title": "Engineering Mgr", "city": "Taquari"}, {"country": "REPUBLIC OF MYANMAR", "rate": "74.5%", "name": "Karl Novak", "company": "Vehement", "job_title": "engineering mgr", "city": "taquari"}, {"country": "Turkmenistan   ", "rate": "22.8%", "name": "Karl Reyes", "company": "   Hooli", "job_title": "ENG MGR", "city": " TAQUARI "}, {"country": "TKM", "rate": "56.6%", "name": "Bob  Ali", "company": "Vehement", "job_title": "Dit.", "city": "Pijijiapan"}, {"country": "  Turkmenistan", "rate": "35.5%", "name": "Heidi  Adams", "company": "Cyberdyne", "job_title": "Sales Jepresentative", "city": "khanty-mansiysk"}, {"country": "WF", "rate": "15.4%", "name": "Mona Brandt", "company": "Umbrella", "job_title": " Rep ", "city": " taquari "}, {"country": "MM", "rate": "84.1%", "name": "   David Ali", "company": "Acme Inc", "job_title": "Salem Rep", "city": "Taquari"}, {"country": "MM", "rate": "7.5%", "name": "Eve Wong", "company": "  Umbrella", "job_title": "Engineering Mgr ", "city": "Pijijiapan"}, {"country": " Myanmar", "rate": "85.3%", "name": " Lena Diaz", "company": "   Stark Industries", "job_title": "Director", "city": "khanty-mansiysk"}, {"country": "TURKMENISTAN", "rate": "93.6%", "name": "David Cruz", "company": "Vehement ", "job_title": " Dir", "city": "Buenos Aires"}, {"country": "wlf", "rate": "42.4%", "name": "   Sara Wong", "company": "Soylent Corp  ", "job_title": " diir.", "city": "ba"}, {"country": "MMR  ", "rate": "71.5%", "name": "Grace Novak", "company": "Globex  ", "job_title": "CEO", "city": "Taquari "}, {"country": "WLF", "rate": "11.9%", "name": "Heidi Cruz", "company": "Acme Inc", "job_title": "Eng Mgr", "city": "PIJIJIAPAN"}, {"country": "TKM", "rate": "63.3%", "name": "Frank Cruz   ", "company": "Vehement ", "job_title": "   rep", "city": "Pijijiapan"}, {"country": " myanmar ", "rate": "94.9%", "name": "   Omar Brandt", "company": "Acme Inc", "job_title": "  Rep", "city": "ba"}, {"country": "Myanmar", "rate": "67.5%", "name": "Grace  Petrov", "company": "Cyberdyne", "job_title": "Chief Executive Odficer", "city": "Khanty-Mansiysk"}, {"country": "TKM", "rate": "71.9%", "name": "Priya Fischer", "company": "Acme Inc", "job_title": "Director", "city": "Khanty-Mansiysk"}, {"country": "Wallis and Futuna", "rate": "39.1%", "name": "Judy Lee   ", "company": "Acme Inc", "job_title": " rep", "city": " PIJIJIAPAN"}, {"country": "myanmar", "rate": "84.7%", "name": "Priya Brandt  ", "company": "Cyberdyne", "job_title": "Dir", "city": "Buenos  Aires"}, {"country": "tkm", "rate": "82.9%", "name": "Priya Smith", "company": "Acme Inc", "job_title": "rep", "city": " Khanty-Mansiysk"}, {"country": "TURKMENISTAN", "rate": "93.6%", "name": "David Cruz", "company": "Vehement ", "job_title": " Dir", "city": "Buenos Aires"}, {"country": "Tkm", "rate": "96.2%", "name": "Bob Johnson", "company": "  Acme Inc", "job_title": "ceo", "city": "Taquari"}, {"country": "", "rate": "", "name": "", "company": "", "job_title": "", "city": ""}], "clean": [{"country": "Wallis and Futuna", "rate": 0.003, "name": "Karl Wong", "company": "Globex", "job_title": "Sales Representative", "city": "Pijijiapan"}, {"country": "Turkmenistan", "rate": 0.182, "name": "Eve Diaz", "company": "Stark Industries", "job_title": "Engineering Manager", "city": "Pijijiapan"}, {"country": "Turkmenistan", "rate": 0.9620000000000001, "name": "Bob Johnson", "company": "Acme Inc", "job_title": "Chief Executive Officer", "city": "Taquari"}, {"country": "Wallis and Futuna", "rate": 0.8740000000000001, "name": "Bob Lee", "company": "Hooli", "job_title": "Chief Executive Officer", "city": "Taquari"}, {"country": "Myanmar", "rate": 0.032, "name": "Sara Cruz", "company": "Soylent Corp", "job_title": "Chief Executive Officer", "city": "Pijijiapan"}, {"country": "Turkmenistan", "rate": 0.6629999999999999, "name": "Karl Johnson", "company": "Initech", "job_title": "Engineering Manager", "city": "Taquari"}, {"country": "Myanmar", "rate": 0.745, "name": "Karl Novak", "company": "Vehement", "job_title": "Engineering Manager", "city": "Taquari"}, {"country": "Turkmenistan", "rate": 0.228, "name": "Karl Reyes", "company": "Hooli", "job_title": "Engineering Manager", "city": "Taquari"}, {"country": "Turkmenistan", "rate": 0.5660000000000001, "name": "Bob Ali", "company": "Vehement", "job_title": "Director", "city": "Pijijiapan"}, {"country": "Turkmenistan", "rate": 0.355, "name": "Heidi Adams", "company": "Cyberdyne", "job_title": "Sales Representative", "city": "Khanty-Mansiysk"}, {"country": "Wallis and Futuna", "rate": 0.154, "name": "Mona Brandt", "company": "Umbrella", "job_title": "Sales Representative", "city": "Taquari"}, {"country": "Myanmar", "rate": 0.841, "name": "David Ali", "company": "Acme Inc", "job_title": "Sales Representative", "city": "Taquari"}, {"country": "Myanmar", "rate": 0.075, "name": "Eve Wong", "company": "Umbrella", "job_title": "Engineering Manager", "city": "Pijijiapan"}, {"country": "Myanmar", "rate": 0.853, "name": "Lena Diaz", "company": "Stark Industries", "job_title": "Director", "city": "Khanty-Mansiysk"}, {"country": "Turkmenistan", "rate": 0.9359999999999999, "name": "David Cruz", "company": "Vehement", "job_title": "Director", "city": "Buenos Aires"}, {"country": "Wallis and Futuna", "rate": 0.424, "name": "Sara Wong", "company": "Soylent Corp", "job_title": "Director", "city": "Buenos Aires"}, {"country": "Myanmar", "rate": 0.715, "name": "Grace Novak", "company": "Globex", "job_title": "Chief Executive Officer", "city": "Taquari"}, {"country": "Wallis and Futuna", "rate": 0.11900000000000001, "name": "Heidi Cruz", "company": "Acme Inc", "job_title": "Engineering Manager", "city": "Pijijiapan"}, {"country": "Turkmenistan", "rate": 0.633, "name": "Frank Cruz", "company": "Vehement", "job_title": "Sales Representative", "city": "Pijijiapan"}, {"country": "Myanmar", "rate": 0.9490000000000001, "name": "Omar Brandt", "company": "Acme Inc", "job_title": "Sales Representative", "city": "Buenos Aires"}, {"country": "Myanmar", "rate": 0.675, "name": "Grace Petrov", "company": "Cyberdyne", "job_title": "Chief Executive Officer", "city": "Khanty-Mansiysk"}, {"country": "Turkmenistan", "rate": 0.7190000000000001, "name": "Priya Fischer", "company": "Acme Inc", "job_title": "Director", "city": "Khanty-Mansiysk"}, {"country": "Wallis and Futuna", "rate": 0.391, "name": "Judy Lee", "company": "Acme Inc", "job_title": "Sales Representative", "city": "Pijijiapan"}, {"country": "Myanmar", "rate": 0.847, "name": "Priya Brandt", "company": "Cyberdyne", "job_title": "Director", "city": "Buenos Aires"}, {"country": "Turkmenistan", "rate": 0.8290000000000001, "name": "Priya Smith", "company": "Acme Inc", "job_title": "Sales Representative", "city": "Khanty-Mansiysk"}], "dirty_cols": ["country", "rate", "name", "company", "job_title", "city"], "clean_cols": ["country", "rate", "name", "company", "job_title", "city"], "plan": {"dataset_summary": "28 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TKM": "Turkmenistan", "Tkm": "Turkmenistan", "WLF": "Wallis and Futuna", "mm": "Myanmar", "REPUBLIC OF MYANMAR": "Myanmar", "WF": "Wallis and Futuna", "MM": "Myanmar", "TURKMENISTAN": "Turkmenistan", "wlf": "Wallis and Futuna", "MMR": "Myanmar", "myanmar": "Myanmar", "tkm": "Turkmenistan"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sales representative": "Sales Representative", "ENG MANAGEER": "Engineering Manager", "ceo": "Chief Executive Officer", "chief executivve officer": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer", "Engineering Mgr": "Engineering Manager", "engineering mgr": "Engineering Manager", "ENG MGR": "Engineering Manager", "Dit.": "Director", "Sales Jepresentative": "Sales Representative", "Rep": "Sales Representative", "Salem Rep": "Sales Representative", "Dir": "Director", "diir.": "Director", "CEO": "Chief Executive Officer", "Eng Mgr": "Engineering Manager", "rep": "Sales Representative", "Chief Executive Odficer": "Chief Executive Officer"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pijijiapan": "Pijijiapan", "Taqquari": "Taquari", "taquari": "Taquari", "TAQUARI": "Taquari", "khanty-mansiysk": "Khanty-Mansiysk", "ba": "Buenos Aires", "PIJIJIAPAN": "Pijijiapan"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "ENG  MANAGER", "signup_date": "7/27/2023", "currency": "Hkd", "rate": "59.3%"}, {"job_title": "Engineering  Mgr", "signup_date": "2023-05-03", "currency": " bermudian dollar", "rate": "32.2%"}, {"job_title": "VICE PRES", "signup_date": "2/21/2023", "currency": "bbd", "rate": "44.3%"}, {"job_title": "   Vice Pres", "signup_date": "2023-06-17", "currency": "bbd", "rate": "86.9%"}, {"job_title": "ENGINEERING MANAGER", "signup_date": "7 Sep 2023", "currency": "   BARBADOS DOLLAR", "rate": "30.1%"}, {"job_title": "vp", "signup_date": "4/7/2023", "currency": "BMD", "rate": "37.2%"}, {"job_title": "   Vice Pres", "signup_date": "2023-05-16", "currency": "BMD", "rate": "9.0%"}, {"job_title": "engineering manager", "signup_date": "3/7/2023", "currency": " BBD ", "rate": "9.0%"}, {"job_title": "Eng Manager", "signup_date": "2023-04-24", "currency": "Barbados Dollar", "rate": "9.8%"}, {"job_title": "V.P.", "signup_date": "1/19/2023", "currency": "WIR Euro", "rate": "42.3%"}, {"job_title": "VICE  PRESIDENT", "signup_date": "4 Jan 2023", "currency": "hong kong dollar", "rate": "6.4%"}, {"job_title": "Eng Mgr", "signup_date": "2023-06-02", "currency": "Hong Kong Dollar", "rate": "1.3%"}, {"job_title": "Eng Mgr", "signup_date": "6 Nov 2023", "currency": "bbd", "rate": "74.3%"}, {"job_title": "Engineering Mgr", "signup_date": "2023-12-15", "currency": "hkd", "rate": "49.6%"}, {"job_title": "Engineering  Mgr", "signup_date": "45217", "currency": "BMD", "rate": "59.6%"}, {"job_title": "Eng Manager", "signup_date": "2023-04-24", "currency": "Barbados Dollar", "rate": "9.8%"}, {"job_title": "Engineering  Mgr", "signup_date": "2023-05-03", "currency": " bermudian dollar", "rate": "32.2%"}], "clean": [{"job_title": "Engineering Manager", "signup_date": "2023-07-27", "currency": "HKD", "rate": 0.593}, {"job_title": "Engineering Manager", "signup_date": "2023-05-03", "currency": "BMD", "rate": 0.322}, {"job_title": "Vice President", "signup_date": "2023-02-21", "currency": "BBD", "rate": 0.44299999999999995}, {"job_title": "Vice President", "signup_date": "2023-06-17", "currency": "BBD", "rate": 0.8690000000000001}, {"job_title": "Engineering Manager", "signup_date": "2023-09-07", "currency": "BBD", "rate": 0.301}, {"job_title": "Vice President", "signup_date": "2023-04-07", "currency": "BMD", "rate": 0.37200000000000005}, {"job_title": "Vice President", "signup_date": "2023-05-16", "currency": "BMD", "rate": 0.09}, {"job_title": "Engineering Manager", "signup_date": "2023-03-07", "currency": "BBD", "rate": 0.09}, {"job_title": "Engineering Manager", "signup_date": "2023-04-24", "currency": "BBD", "rate": 0.098}, {"job_title": "Vice President", "signup_date": "2023-01-19", "currency": "CHE", "rate": 0.423}, {"job_title": "Vice President", "signup_date": "2023-01-04", "currency": "HKD", "rate": 0.064}, {"job_title": "Engineering Manager", "signup_date": "2023-06-02", "currency": "HKD", "rate": 0.013000000000000001}, {"job_title": "Engineering Manager", "signup_date": "2023-11-06", "currency": "BBD", "rate": 0.743}, {"job_title": "Engineering Manager", "signup_date": "2023-12-15", "currency": "HKD", "rate": 0.496}, {"job_title": "Engineering Manager", "signup_date": "2023-10-18", "currency": "BMD", "rate": 0.596}], "dirty_cols": ["job_title", "signup_date", "currency", "rate"], "clean_cols": ["job_title", "signup_date", "currency", "rate"], "plan": {"dataset_summary": "17 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ENG MANAGER": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "VICE PRES": "Vice President", "Vice Pres": "Vice President", "ENGINEERING MANAGER": "Engineering Manager", "vp": "Vice President", "engineering manager": "Engineering Manager", "Eng Manager": "Engineering Manager", "V.P.": "Vice President", "VICE PRESIDENT": "Vice President", "Eng Mgr": "Engineering Manager"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Hkd": "HKD", "bermudian dollar": "BMD", "bbd": "BBD", "BARBADOS DOLLAR": "BBD", "Barbados Dollar": "BBD", "WIR Euro": "CHE", "hong kong dollar": "HKD", "Hong Kong Dollar": "HKD", "hkd": "HKD"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"rate": "9.7%", "status": "High", "currency": " Boliiano", "job_title": "Account Executive  ", "city": "Deepew", "name": "Heidi  Smith", "notes2": ""}, {"rate": "91.0%", "status": "high", "currency": "boliviano", "job_title": "Senior Engineer", "city": "DEPEW", "name": "Mona Ali ", "notes2": ""}, {"rate": "95.3%", "status": " P1 ", "currency": "   BOB", "job_title": " Acct Exec", "city": "general mosconi", "name": "David Petrov", "notes2": ""}, {"rate": "99.5%", "status": "L ", "currency": "Bhd   ", "job_title": "director", "city": "Washington Dc", "name": "Mona Lee", "notes2": ""}, {"rate": "24.4%", "status": "Low", "currency": "BOB", "job_title": "Dir", "city": "general mosconi", "name": " Grace Park", "notes2": ""}, {"rate": "48.5%", "status": "P3 ", "currency": "Bahraini Dinar", "job_title": "Snr Engineer", "city": "DC", "name": "David  Brandt", "notes2": ""}, {"rate": "93.3%", "status": "High", "currency": "BOB", "job_title": " SNR ENGINEER", "city": "General Mosconi", "name": "Carol Adams  ", "notes2": ""}, {"rate": "56.3%", "status": "L", "currency": "bob", "job_title": "Dir", "city": "General Mosconi", "name": "Mona Petrov", "notes2": ""}, {"rate": "92.3%", "status": "L", "currency": "bob  ", "job_title": "   Account Execktive", "city": "Dessau", "name": "Karl Cruz", "notes2": ""}, {"rate": "88.1%", "status": " P1 ", "currency": " bob", "job_title": "Acct Exec", "city": "Dessau", "name": "Ivan Reyes  ", "notes2": ""}, {"rate": "38.6%", "status": "high", "currency": "Bahraini Dinar", "job_title": "Senior  Eng", "city": "General Mosconi", "name": "Bob Wong", "notes2": ""}, {"rate": "92.2%", "status": "High", "currency": "boliviano", "job_title": "Acct Exec", "city": "dessau", "name": "Alice Park", "notes2": ""}, {"rate": "26.6%", "status": " L ", "currency": "bob", "job_title": "DIR.", "city": "depew", "name": "Heidi Johnson", "notes2": ""}, {"rate": "86.4%", "status": "high", "currency": "BHD", "job_title": "AE", "city": "DC", "name": "Judy Reyes   ", "notes2": ""}, {"rate": "26.5%", "status": "p3", "currency": "bahraini  dinar", "job_title": "Snr Engineer", "city": "Wjshington  ", "name": "Bob  Smith", "notes2": ""}, {"rate": "55.8%", "status": "Low", "currency": "BHD  ", "job_title": "Acct Exec", "city": "Dessua", "name": "Eve Brandt", "notes2": ""}, {"rate": "10.5%", "status": "HIGH", "currency": "BOB", "job_title": "AE", "city": "general mosconi", "name": "Sara Wong", "notes2": ""}, {"rate": "2.8%", "status": " H", "currency": "Boliviano", "job_title": "Snr Engineer", "city": "Depew", "name": "Lena Diaz", "notes2": ""}, {"rate": "49.4%", "status": "l", "currency": "bhd", "job_title": "Senior Eng", "city": "General Mosconi", "name": "David  Khan", "notes2": ""}, {"rate": "49.4%", "status": "L  ", "currency": "Bahraini Dinar", "job_title": "SENIOR ENGINEER", "city": " Washington DC", "name": "Frank Fischer   ", "notes2": ""}, {"rate": "34.0%", "status": "Low", "currency": "  Bahraini Dinar", "job_title": "  Senior Eng", "city": "washington dc", "name": "Mona  Adams", "notes2": ""}, {"rate": "96.5%", "status": "Low ", "currency": "Boliviano", "job_title": "DIR.", "city": " Dc ", "name": "   Ivan Diaz", "notes2": ""}, {"rate": "1.9%", "status": "P3  ", "currency": "BOLIVIANO", "job_title": "Dir ", "city": "Washington DC", "name": "Lena Novak   ", "notes2": ""}, {"rate": "95.3%", "status": " P1 ", "currency": "   BOB", "job_title": " Acct Exec", "city": "general mosconi", "name": "David Petrov", "notes2": ""}, {"rate": "26.6%", "status": " L ", "currency": "bob", "job_title": "DIR.", "city": "depew", "name": "Heidi Johnson", "notes2": ""}, {"rate": "", "status": "", "currency": "", "job_title": "", "city": "", "name": "", "notes2": ""}], "clean": [{"rate": 0.09699999999999999, "status": "High", "currency": "BOB", "job_title": "Account Executive", "city": "Depew", "name": "Heidi Smith"}, {"rate": 0.91, "status": "High", "currency": "BOB", "job_title": "Senior Engineer", "city": "Depew", "name": "Mona Ali"}, {"rate": 0.953, "status": "High", "currency": "BOB", "job_title": "Account Executive", "city": "General Mosconi", "name": "David Petrov"}, {"rate": 0.995, "status": "Low", "currency": "BHD", "job_title": "Director", "city": "Washington", "name": "Mona Lee"}, {"rate": 0.244, "status": "Low", "currency": "BOB", "job_title": "Director", "city": "General Mosconi", "name": "Grace Park"}, {"rate": 0.485, "status": "Low", "currency": "BHD", "job_title": "Senior Engineer", "city": "Washington", "name": "David Brandt"}, {"rate": 0.9329999999999999, "status": "High", "currency": "BOB", "job_title": "Senior Engineer", "city": "General Mosconi", "name": "Carol Adams"}, {"rate": 0.563, "status": "Low", "currency": "BOB", "job_title": "Director", "city": "General Mosconi", "name": "Mona Petrov"}, {"rate": 0.9229999999999999, "status": "Low", "currency": "BOB", "job_title": "Account Executive", "city": "Dessau", "name": "Karl Cruz"}, {"rate": 0.8809999999999999, "status": "High", "currency": "BOB", "job_title": "Account Executive", "city": "Dessau", "name": "Ivan Reyes"}, {"rate": 0.386, "status": "High", "currency": "BHD", "job_title": "Senior Engineer", "city": "General Mosconi", "name": "Bob Wong"}, {"rate": 0.922, "status": "High", "currency": "BOB", "job_title": "Account Executive", "city": "Dessau", "name": "Alice Park"}, {"rate": 0.266, "status": "Low", "currency": "BOB", "job_title": "Director", "city": "Depew", "name": "Heidi Johnson"}, {"rate": 0.8640000000000001, "status": "High", "currency": "BHD", "job_title": "Account Executive", "city": "Washington", "name": "Judy Reyes"}, {"rate": 0.265, "status": "Low", "currency": "BHD", "job_title": "Senior Engineer", "city": "Washington", "name": "Bob Smith"}, {"rate": 0.5579999999999999, "status": "Low", "currency": "BHD", "job_title": "Account Executive", "city": "Dessau", "name": "Eve Brandt"}, {"rate": 0.105, "status": "High", "currency": "BOB", "job_title": "Account Executive", "city": "General Mosconi", "name": "Sara Wong"}, {"rate": 0.027999999999999997, "status": "High", "currency": "BOB", "job_title": "Senior Engineer", "city": "Depew", "name": "Lena Diaz"}, {"rate": 0.494, "status": "Low", "currency": "BHD", "job_title": "Senior Engineer", "city": "General Mosconi", "name": "David Khan"}, {"rate": 0.494, "status": "Low", "currency": "BHD", "job_title": "Senior Engineer", "city": "Washington", "name": "Frank Fischer"}, {"rate": 0.34, "status": "Low", "currency": "BHD", "job_title": "Senior Engineer", "city": "Washington", "name": "Mona Adams"}, {"rate": 0.965, "status": "Low", "currency": "BOB", "job_title": "Director", "city": "Washington", "name": "Ivan Diaz"}, {"rate": 0.019, "status": "Low", "currency": "BOB", "job_title": "Director", "city": "Washington", "name": "Lena Novak"}], "dirty_cols": ["rate", "status", "currency", "job_title", "city", "name", "notes2"], "clean_cols": ["rate", "status", "currency", "job_title", "city", "name"], "plan": {"dataset_summary": "26 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"high": "High", "P1": "High", "L": "Low", "P3": "Low", "p3": "Low", "HIGH": "High", "H": "High", "l": "Low"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Boliiano": "BOB", "boliviano": "BOB", "Bhd": "BHD", "Bahraini Dinar": "BHD", "bob": "BOB", "bahraini dinar": "BHD", "Boliviano": "BOB", "bhd": "BHD", "BOLIVIANO": "BOB"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Acct Exec": "Account Executive", "director": "Director", "Dir": "Director", "Snr Engineer": "Senior Engineer", "SNR ENGINEER": "Senior Engineer", "Account Execktive": "Account Executive", "Senior Eng": "Senior Engineer", "DIR.": "Director", "AE": "Account Executive", "SENIOR ENGINEER": "Senior Engineer"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Deepew": "Depew", "DEPEW": "Depew", "general mosconi": "General Mosconi", "Washington Dc": "Washington", "DC": "Washington", "dessau": "Dessau", "depew": "Depew", "Wjshington": "Washington", "Dessua": "Dessau", "Washington DC": "Washington", "washington dc": "Washington", "Dc": "Washington"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"is_active": "FALSE", "rate": "1.1%", "city": "Summerville", "unnamed": ""}, {"is_active": "N", "rate": "93.5%", "city": "tatarka", "unnamed": ""}, {"is_active": "TRUE", "rate": "82.7%", "city": "Tatarka", "unnamed": ""}, {"is_active": "Y", "rate": "17.1%", "city": "Stockholm", "unnamed": ""}, {"is_active": "0", "rate": "28.6%", "city": "tatarka", "unnamed": ""}, {"is_active": "Y", "rate": "85.6%", "city": "Stockholm", "unnamed": ""}, {"is_active": "N", "rate": "76.8%", "city": "nairobi", "unnamed": ""}, {"is_active": "true", "rate": "37.1%", "city": "Nairobi", "unnamed": ""}, {"is_active": "FALSE", "rate": "36.3%", "city": "stockholm", "unnamed": ""}, {"is_active": "N", "rate": "86.2%", "city": "Summerville", "unnamed": ""}, {"is_active": "TRUE", "rate": "84.6%", "city": "Stockholm", "unnamed": ""}, {"is_active": "No", "rate": "38.1%", "city": "stockholm", "unnamed": ""}, {"is_active": "false", "rate": "51.4%", "city": "   Tatarka", "unnamed": ""}, {"is_active": "", "rate": "", "city": "", "unnamed": ""}], "clean": [{"is_active": false, "rate": 0.011000000000000001, "city": "Summerville"}, {"is_active": false, "rate": 0.935, "city": "Tatarka"}, {"is_active": true, "rate": 0.8270000000000001, "city": "Tatarka"}, {"is_active": true, "rate": 0.171, "city": "Stockholm"}, {"is_active": false, "rate": 0.28600000000000003, "city": "Tatarka"}, {"is_active": true, "rate": 0.856, "city": "Stockholm"}, {"is_active": false, "rate": 0.768, "city": "Nairobi"}, {"is_active": true, "rate": 0.371, "city": "Nairobi"}, {"is_active": false, "rate": 0.363, "city": "Stockholm"}, {"is_active": false, "rate": 0.862, "city": "Summerville"}, {"is_active": true, "rate": 0.846, "city": "Stockholm"}, {"is_active": false, "rate": 0.381, "city": "Stockholm"}, {"is_active": false, "rate": 0.514, "city": "Tatarka"}], "dirty_cols": ["is_active", "rate", "city", "unnamed"], "clean_cols": ["is_active", "rate", "city"], "plan": {"dataset_summary": "14 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"tatarka": "Tatarka", "nairobi": "Nairobi", "stockholm": "Stockholm"}, "rationale": "Unified 3 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"company": "Soylent Corp", "name": "   Frank Lee", "unit": "pound", "status": "Closed Lost", "job_title": "Admin Asst", "city": "Dessau"}, {"company": "Initech ", "name": "  David Wong", "unit": "pound", "status": "lost", "job_title": "Administrative Assistant", "city": "Dessau"}, {"company": "  Acme Inc", "name": "Frank Cruz", "unit": "kg", "status": "Lost   ", "job_title": "VP", "city": "Lianhe"}, {"company": "   Soylent Corp", "name": "David  Khan", "unit": "kilograms", "status": "open", "job_title": "Administrative Assistant", "city": "Dessau "}, {"company": "Globex", "name": "Bob  Ali", "unit": "Lb", "status": " lost ", "job_title": "snr engineer", "city": "Lianhe"}, {"company": "Hooli", "name": "Sara Khan", "unit": "   kg", "status": "open", "job_title": "snr engineer", "city": "Nunspeet"}, {"company": "Acme Inc", "name": "Sara  Adams", "unit": "LB", "status": "Open", "job_title": "Vice President", "city": "Lianhe"}, {"company": " Vehement ", "name": "Karl Khan", "unit": "pounds", "status": "won", "job_title": "v..p", "city": "dessau   "}, {"company": "  Globex", "name": "Bob Novak", "unit": "  ml", "status": "Won", "job_title": "Admn", "city": " DESSAU "}, {"company": "Initech", "name": "Alice Moore", "unit": "kilogram", "status": "new", "job_title": "Senior Engineer", "city": "Lianhe"}, {"company": "Umbrella", "name": "Carol Lee  ", "unit": "kg", "status": "new", "job_title": " vice pres", "city": "Dessau"}, {"company": "Hooli", "name": "Eve Ali", "unit": "Ml", "status": "closedwon", "job_title": "admin assistant", "city": "Lianhe   "}, {"company": "   Cyberdyne", "name": "Lena Lee", "unit": " Lb ", "status": "lost", "job_title": "Vice President", "city": "lianhe"}, {"company": "Soylent Corp", "name": "Karl Lee", "unit": "mL ", "status": "closed-won", "job_title": "Admin Assistant", "city": "nunspeet"}, {"company": "Wonka Co", "name": " Bob Reyes", "unit": "Kg", "status": "won", "job_title": "administrative assistant", "city": "Lianhe"}, {"company": "Stark Industries ", "name": "Lena Johnson", "unit": "mL", "status": "Lost", "job_title": "administrative assistant", "city": "Lianhe"}, {"company": "  Initech", "name": "Grace  Cruz", "unit": "Lbs", "status": "Closed Lost", "job_title": "Sr Engineer", "city": "dessau"}, {"company": "Globex   ", "name": "  Alice Moore", "unit": "ml", "status": "   new", "job_title": "Admin Asst", "city": "Nunspeet"}, {"company": "Wonka Co", "name": "Karl Khan", "unit": "kg", "status": "OPEN", "job_title": "sr.  engineer", "city": "Lianhe"}, {"company": "Hooli   ", "name": "Judy Petrov", "unit": "KG", "status": " open", "job_title": "SNR ENGINEER", "city": "Lianhe"}, {"company": "Soylent Corp", "name": "  Frank Diaz", "unit": "ml", "status": "lost", "job_title": "seniro eng", "city": "Dessau"}, {"company": "Cyberdyne", "name": "Karl Cruz", "unit": "POUNDS", "status": "Won", "job_title": "Sr Engineer", "city": "nunspeet"}, {"company": "   Stark Industries", "name": "Carol Smith", "unit": " Milliliter ", "status": "new", "job_title": "Admin Asst", "city": "Lianhe"}, {"company": "Wonka Co  ", "name": "Karl  Cruz", "unit": "ML", "status": "WON", "job_title": "Vice Pres", "city": "Dessau"}, {"company": "  Vehement", "name": "Mona  Khan", "unit": "pound   ", "status": "lost", "job_title": "Sr Engineer", "city": "lianhe"}, {"company": "Vehement", "name": "Alice Petrov", "unit": "mls", "status": "lost", "job_title": "Admin Asst", "city": "Lianhe"}, {"company": " Umbrella ", "name": "Judy Johnson", "unit": "kg", "status": "closed lost", "job_title": "Sr Engineer", "city": "DESSAU  "}, {"company": "Globex", "name": "David Cruz  ", "unit": "kg", "status": "  won", "job_title": "admi  nassistant", "city": "Lianhe"}, {"company": "Umbrella", "name": "Lena Wong", "unit": "kilograms", "status": "Won  ", "job_title": "vp", "city": "dessau"}, {"company": " Stark Industries", "name": "Mona Petrov", "unit": "lbs", "status": "Lost", "job_title": "v.p.", "city": "Lianhe"}, {"company": "Acme  Inc", "name": "Alice Wong", "unit": "kg", "status": "oepn", "job_title": "Administrative Assistant", "city": "LIANHE  "}, {"company": "Vehement", "name": "Frank Petrov", "unit": "lbs", "status": "opn   ", "job_title": "Sr. Engineer", "city": "Lianhe"}, {"company": "", "name": "", "unit": "", "status": "", "job_title": "", "city": ""}], "clean": [{"company": "Soylent Corp", "name": "Frank Lee", "unit": "lb", "status": "Lost", "job_title": "Administrative Assistant", "city": "Dessau"}, {"company": "Initech", "name": "David Wong", "unit": "lb", "status": "Lost", "job_title": "Administrative Assistant", "city": "Dessau"}, {"company": "Acme Inc", "name": "Frank Cruz", "unit": "kg", "status": "Lost", "job_title": "Vice President", "city": "Lianhe"}, {"company": "Soylent Corp", "name": "David Khan", "unit": "kg", "status": "Open", "job_title": "Administrative Assistant", "city": "Dessau"}, {"company": "Globex", "name": "Bob Ali", "unit": "lb", "status": "Lost", "job_title": "Senior Engineer", "city": "Lianhe"}, {"company": "Hooli", "name": "Sara Khan", "unit": "kg", "status": "Open", "job_title": "Senior Engineer", "city": "Nunspeet"}, {"company": "Acme Inc", "name": "Sara Adams", "unit": "lb", "status": "Open", "job_title": "Vice President", "city": "Lianhe"}, {"company": "Vehement", "name": "Karl Khan", "unit": "lb", "status": "Won", "job_title": "Vice President", "city": "Dessau"}, {"company": "Globex", "name": "Bob Novak", "unit": "mL", "status": "Won", "job_title": "Administrative Assistant", "city": "Dessau"}, {"company": "Initech", "name": "Alice Moore", "unit": "kg", "status": "Open", "job_title": "Senior Engineer", "city": "Lianhe"}, {"company": "Umbrella", "name": "Carol Lee", "unit": "kg", "status": "Open", "job_title": "Vice President", "city": "Dessau"}, {"company": "Hooli", "name": "Eve Ali", "unit": "mL", "status": "Won", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Cyberdyne", "name": "Lena Lee", "unit": "lb", "status": "Lost", "job_title": "Vice President", "city": "Lianhe"}, {"company": "Soylent Corp", "name": "Karl Lee", "unit": "mL", "status": "Won", "job_title": "Administrative Assistant", "city": "Nunspeet"}, {"company": "Wonka Co", "name": "Bob Reyes", "unit": "kg", "status": "Won", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Stark Industries", "name": "Lena Johnson", "unit": "mL", "status": "Lost", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Initech", "name": "Grace Cruz", "unit": "lb", "status": "Lost", "job_title": "Senior Engineer", "city": "Dessau"}, {"company": "Globex", "name": "Alice Moore", "unit": "mL", "status": "Open", "job_title": "Administrative Assistant", "city": "Nunspeet"}, {"company": "Wonka Co", "name": "Karl Khan", "unit": "kg", "status": "Open", "job_title": "Senior Engineer", "city": "Lianhe"}, {"company": "Hooli", "name": "Judy Petrov", "unit": "kg", "status": "Open", "job_title": "Senior Engineer", "city": "Lianhe"}, {"company": "Soylent Corp", "name": "Frank Diaz", "unit": "mL", "status": "Lost", "job_title": "Senior Engineer", "city": "Dessau"}, {"company": "Cyberdyne", "name": "Karl Cruz", "unit": "lb", "status": "Won", "job_title": "Senior Engineer", "city": "Nunspeet"}, {"company": "Stark Industries", "name": "Carol Smith", "unit": "mL", "status": "Open", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Wonka Co", "name": "Karl Cruz", "unit": "mL", "status": "Won", "job_title": "Vice President", "city": "Dessau"}, {"company": "Vehement", "name": "Mona Khan", "unit": "lb", "status": "Lost", "job_title": "Senior Engineer", "city": "Lianhe"}, {"company": "Vehement", "name": "Alice Petrov", "unit": "mL", "status": "Lost", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Umbrella", "name": "Judy Johnson", "unit": "kg", "status": "Lost", "job_title": "Senior Engineer", "city": "Dessau"}, {"company": "Globex", "name": "David Cruz", "unit": "kg", "status": "Won", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Umbrella", "name": "Lena Wong", "unit": "kg", "status": "Won", "job_title": "Vice President", "city": "Dessau"}, {"company": "Stark Industries", "name": "Mona Petrov", "unit": "lb", "status": "Lost", "job_title": "Vice President", "city": "Lianhe"}, {"company": "Acme Inc", "name": "Alice Wong", "unit": "kg", "status": "Open", "job_title": "Administrative Assistant", "city": "Lianhe"}, {"company": "Vehement", "name": "Frank Petrov", "unit": "lb", "status": "Open", "job_title": "Senior Engineer", "city": "Lianhe"}], "dirty_cols": ["company", "name", "unit", "status", "job_title", "city"], "clean_cols": ["company", "name", "unit", "status", "job_title", "city"], "plan": {"dataset_summary": "33 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pound": "lb", "kilograms": "kg", "Lb": "lb", "LB": "lb", "pounds": "lb", "ml": "mL", "kilogram": "kg", "Ml": "mL", "Kg": "kg", "Lbs": "lb", "KG": "kg", "POUNDS": "lb", "Milliliter": "mL", "ML": "mL", "mls": "mL", "lbs": "lb"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Closed Lost": "Lost", "lost": "Lost", "open": "Open", "won": "Won", "new": "Open", "closedwon": "Won", "closed-won": "Won", "OPEN": "Open", "WON": "Won", "closed lost": "Lost", "oepn": "Open", "opn": "Open"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Admin Asst": "Administrative Assistant", "VP": "Vice President", "snr engineer": "Senior Engineer", "v..p": "Vice President", "Admn": "Administrative Assistant", "vice pres": "Vice President", "admin assistant": "Administrative Assistant", "Admin Assistant": "Administrative Assistant", "administrative assistant": "Administrative Assistant", "Sr Engineer": "Senior Engineer", "sr. engineer": "Senior Engineer", "SNR ENGINEER": "Senior Engineer", "seniro eng": "Senior Engineer", "Vice Pres": "Vice President", "admi nassistant": "Administrative Assistant", "vp": "Vice President", "v.p.": "Vice President", "Sr. Engineer": "Senior Engineer"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"dessau": "Dessau", "DESSAU": "Dessau", "lianhe": "Lianhe", "nunspeet": "Nunspeet", "LIANHE": "Lianhe"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"email": "ffcbd@example.com", "city": " Badessa ", "currency": "CLP", "job_title": "CEO", "rate": "94.0%", "extra": ""}, {"email": "PKLF@EXAMPLE.COM", "city": "Maba", "currency": "CHILEAN PESO", "job_title": "CEO", "rate": "10.3%", "extra": ""}, {"email": "jimp@corp.io ", "city": "Badessa", "currency": " gtq ", "job_title": "Snr Engineer", "rate": "31.0%", "extra": ""}, {"email": "dbbik@test.org  ", "city": "MANILA  ", "currency": "GTQ  ", "job_title": "Senior Eng", "rate": "75.8%", "extra": ""}, {"email": "CHFKJ@TEST.ORG", "city": "Riverside", "currency": "chiilean peso", "job_title": "Chief Executive Officer  ", "rate": "18.5%", "extra": ""}, {"email": "NDLOHEL@TEST.ORG ", "city": "manila", "currency": "CLP", "job_title": " C.E.O.", "rate": "97.2%", "extra": ""}, {"email": "PFLNDK@MAIL.COM", "city": "Riverside", "currency": "CLP", "job_title": "Senior Engineer", "rate": "77.8%", "extra": ""}, {"email": " NJAHL@TEST.ORG ", "city": "riverside", "currency": "Quetzal", "job_title": "Sr Engineer", "rate": "5.2%", "extra": ""}, {"email": "HGKG@TEST.ORG ", "city": "badessa", "currency": "ars", "job_title": "CEO", "rate": "10.3%", "extra": ""}, {"email": "NMPKOHB@EXAMPLE.COM", "city": "Manila", "currency": "Chilean Peso", "job_title": "   chief executive officer", "rate": "11.0%", "extra": ""}, {"email": "BNGP@MAIL.COM", "city": "rivezside", "currency": "Quetzal", "job_title": "Chief Executive Officer", "rate": "67.9%", "extra": ""}, {"email": "eajb@test.org   ", "city": " Manila ", "currency": "chilean  peso", "job_title": "Senioor Eng", "rate": "67.7%", "extra": ""}, {"email": " CJNCAHO@MAIL.COM ", "city": "RIVEERSIDE", "currency": "Argentine Peso", "job_title": "   Sr. Engineer", "rate": "22.9%", "extra": ""}, {"email": "cmpf@mail.com", "city": "MABA", "currency": "quetzal", "job_title": "SENIOR ENG", "rate": "22.9%", "extra": ""}, {"email": "ahjph@mail.com", "city": "manila", "currency": "clp", "job_title": "C.E.O.", "rate": "11.1%", "extra": ""}, {"email": "  JOGA@TEST.ORG", "city": "   riverside", "currency": "Chilean Peso", "job_title": "C.E.O.", "rate": "39.1%", "extra": ""}, {"email": "pjadoj@mail.com", "city": "MABA", "currency": "Qpetzal", "job_title": " ceo ", "rate": "84.2%", "extra": ""}, {"email": " HPPAE@EXAMPLE.COM ", "city": "Riverside", "currency": "Chilean Piso", "job_title": "chief executive officer", "rate": "83.8%", "extra": ""}, {"email": " DJKJFKA@MAIL.COM", "city": " manila", "currency": "Quetzal", "job_title": "chief executive officer", "rate": "66.2%", "extra": ""}, {"email": "aipklm@corp.io", "city": "badessa", "currency": "argentine peso", "job_title": "SR ENGINEER ", "rate": "73.0%", "extra": ""}, {"email": " GEMJDFN@TEST.ORG ", "city": " MANILA ", "currency": "  Chilean Peso", "job_title": "Chief Executive Officer", "rate": "46.6%", "extra": ""}, {"email": "  PDMEKAD@TEST.ORG", "city": "maba", "currency": "   QUETZAL", "job_title": "CEO", "rate": "53.9%", "extra": ""}, {"email": "laoie@mail.com", "city": "Maba", "currency": "   chilean peso", "job_title": "Sr. Engineer", "rate": "46.1%", "extra": ""}, {"email": "  gjhlmgj@corp.io", "city": "Badessa", "currency": "   Argentine Peso", "job_title": "Chief Executive  Officer", "rate": "76.6%", "extra": ""}, {"email": "keni@example.com  ", "city": " Manila", "currency": "Argentine Peso", "job_title": "C.E.O.", "rate": "94.5%", "extra": ""}, {"email": "FKKD@CORP.IO", "city": "Riverside", "currency": "   ars", "job_title": "Senior Engineer", "rate": "66.7%", "extra": ""}, {"email": "   OFGIBB@MAIL.COM", "city": "Manila", "currency": "Chilean Peso", "job_title": "ceo", "rate": "50.6%", "extra": ""}, {"email": "gcgm@corp.io", "city": "Maba", "currency": "Argentine Peso", "job_title": "CEO", "rate": "45.0%", "extra": ""}, {"email": "lnoeodf@mail.com", "city": "Badessa", "currency": "clp", "job_title": "c.e.o.", "rate": "81.4%", "extra": ""}, {"email": "DHHHPBJ@CORP.IO   ", "city": "bdessa", "currency": "Quetzal", "job_title": "sr engineer   ", "rate": "55.9%", "extra": ""}, {"email": "gdbcb@example.com", "city": "Maba", "currency": "quetzal", "job_title": " c.e.o. ", "rate": "86.3%", "extra": ""}, {"email": "FHKEJ@EXAMPLE.COM", "city": "Mbaa", "currency": "Chilean Peso", "job_title": "CHIEF EXECUTIVE OFFICER", "rate": "18.1%", "extra": ""}, {"email": " cceh@test.org", "city": " Riverside", "currency": "Argentin ePeso", "job_title": "senior  eng", "rate": "0.5%", "extra": ""}, {"email": "  PDMEKAD@TEST.ORG", "city": "maba", "currency": "   QUETZAL", "job_title": "CEO", "rate": "53.9%", "extra": ""}], "clean": [{"email": "ffcbd@example.com", "city": "Badessa", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.94}, {"email": "pklf@example.com", "city": "Maba", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.10300000000000001}, {"email": "jimp@corp.io", "city": "Badessa", "currency": "GTQ", "job_title": "Senior Engineer", "rate": 0.31}, {"email": "dbbik@test.org", "city": "Manila", "currency": "GTQ", "job_title": "Senior Engineer", "rate": 0.758}, {"email": "chfkj@test.org", "city": "Riverside", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.185}, {"email": "ndlohel@test.org", "city": "Manila", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.972}, {"email": "pflndk@mail.com", "city": "Riverside", "currency": "CLP", "job_title": "Senior Engineer", "rate": 0.778}, {"email": "njahl@test.org", "city": "Riverside", "currency": "GTQ", "job_title": "Senior Engineer", "rate": 0.052000000000000005}, {"email": "hgkg@test.org", "city": "Badessa", "currency": "ARS", "job_title": "Chief Executive Officer", "rate": 0.10300000000000001}, {"email": "nmpkohb@example.com", "city": "Manila", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.11}, {"email": "bngp@mail.com", "city": "Riverside", "currency": "GTQ", "job_title": "Chief Executive Officer", "rate": 0.679}, {"email": "eajb@test.org", "city": "Manila", "currency": "CLP", "job_title": "Senior Engineer", "rate": 0.677}, {"email": "cjncaho@mail.com", "city": "Riverside", "currency": "ARS", "job_title": "Senior Engineer", "rate": 0.22899999999999998}, {"email": "cmpf@mail.com", "city": "Maba", "currency": "GTQ", "job_title": "Senior Engineer", "rate": 0.22899999999999998}, {"email": "ahjph@mail.com", "city": "Manila", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.111}, {"email": "joga@test.org", "city": "Riverside", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.391}, {"email": "pjadoj@mail.com", "city": "Maba", "currency": "GTQ", "job_title": "Chief Executive Officer", "rate": 0.8420000000000001}, {"email": "hppae@example.com", "city": "Riverside", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.838}, {"email": "djkjfka@mail.com", "city": "Manila", "currency": "GTQ", "job_title": "Chief Executive Officer", "rate": 0.662}, {"email": "aipklm@corp.io", "city": "Badessa", "currency": "ARS", "job_title": "Senior Engineer", "rate": 0.73}, {"email": "gemjdfn@test.org", "city": "Manila", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.466}, {"email": "pdmekad@test.org", "city": "Maba", "currency": "GTQ", "job_title": "Chief Executive Officer", "rate": 0.539}, {"email": "laoie@mail.com", "city": "Maba", "currency": "CLP", "job_title": "Senior Engineer", "rate": 0.461}, {"email": "gjhlmgj@corp.io", "city": "Badessa", "currency": "ARS", "job_title": "Chief Executive Officer", "rate": 0.7659999999999999}, {"email": "keni@example.com", "city": "Manila", "currency": "ARS", "job_title": "Chief Executive Officer", "rate": 0.945}, {"email": "fkkd@corp.io", "city": "Riverside", "currency": "ARS", "job_title": "Senior Engineer", "rate": 0.667}, {"email": "ofgibb@mail.com", "city": "Manila", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.506}, {"email": "gcgm@corp.io", "city": "Maba", "currency": "ARS", "job_title": "Chief Executive Officer", "rate": 0.45}, {"email": "lnoeodf@mail.com", "city": "Badessa", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.8140000000000001}, {"email": "dhhhpbj@corp.io", "city": "Badessa", "currency": "GTQ", "job_title": "Senior Engineer", "rate": 0.5589999999999999}, {"email": "gdbcb@example.com", "city": "Maba", "currency": "GTQ", "job_title": "Chief Executive Officer", "rate": 0.863}, {"email": "fhkej@example.com", "city": "Maba", "currency": "CLP", "job_title": "Chief Executive Officer", "rate": 0.18100000000000002}, {"email": "cceh@test.org", "city": "Riverside", "currency": "ARS", "job_title": "Senior Engineer", "rate": 0.005}], "dirty_cols": ["email", "city", "currency", "job_title", "rate", "extra"], "clean_cols": ["email", "city", "currency", "job_title", "rate"], "plan": {"dataset_summary": "34 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MANILA": "Manila", "manila": "Manila", "riverside": "Riverside", "badessa": "Badessa", "rivezside": "Riverside", "RIVEERSIDE": "Riverside", "MABA": "Maba", "maba": "Maba", "bdessa": "Badessa", "Mbaa": "Maba"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CHILEAN PESO": "CLP", "gtq": "GTQ", "chiilean peso": "CLP", "Quetzal": "GTQ", "ars": "ARS", "Chilean Peso": "CLP", "chilean peso": "CLP", "Argentine Peso": "ARS", "quetzal": "GTQ", "clp": "CLP", "Qpetzal": "GTQ", "Chilean Piso": "CLP", "argentine peso": "ARS", "QUETZAL": "GTQ", "Argentin ePeso": "ARS"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CEO": "Chief Executive Officer", "Snr Engineer": "Senior Engineer", "Senior Eng": "Senior Engineer", "C.E.O.": "Chief Executive Officer", "Sr Engineer": "Senior Engineer", "chief executive officer": "Chief Executive Officer", "Senioor Eng": "Senior Engineer", "Sr. Engineer": "Senior Engineer", "SENIOR ENG": "Senior Engineer", "ceo": "Chief Executive Officer", "SR ENGINEER": "Senior Engineer", "c.e.o.": "Chief Executive Officer", "sr engineer": "Senior Engineer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "senior eng": "Senior Engineer"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"industry": "Real Estate", "is_active": "0", "company": "Wonka Co", "status": "late", "job_title": "   Chief Executive Officer"}, {"industry": " Education ", "is_active": "Yes", "company": "Hooli", "status": "refund", "job_title": " chief executive officer"}, {"industry": "Medical", "is_active": "TRUE", "company": "Vehement", "status": " overdue ", "job_title": "Rep"}, {"industry": "Industrial", "is_active": "T", "company": "Hooli", "status": " late", "job_title": "ceo"}, {"industry": "health care", "is_active": "Yes", "company": " Cyberdyne", "status": "late", "job_title": " Vice President"}, {"industry": "Medcial", "is_active": "1", "company": " Initech", "status": "refnuded", "job_title": "  SALES REPRESENTATIVE"}, {"industry": "HEALTH", "is_active": "Yes", "company": " Initech ", "status": "   Refund", "job_title": "dir."}, {"industry": "Mfg", "is_active": "FALSE", "company": "Stark Industries", "status": "refunded", "job_title": "director"}, {"industry": "education", "is_active": "false", "company": "Initech", "status": "refund", "job_title": "DIR"}, {"industry": " Mfg ", "is_active": "No", "company": "Hooli ", "status": "refunded", "job_title": "Dir"}, {"industry": "mfg", "is_active": "Y", "company": "Hooli", "status": "Late", "job_title": "REP"}, {"industry": "", "is_active": "", "company": "", "status": "", "job_title": ""}], "clean": [{"industry": "Real Estate", "is_active": false, "company": "Wonka Co", "status": "Overdue", "job_title": "Chief Executive Officer"}, {"industry": "Education", "is_active": true, "company": "Hooli", "status": "Refunded", "job_title": "Chief Executive Officer"}, {"industry": "Healthcare", "is_active": true, "company": "Vehement", "status": "Overdue", "job_title": "Sales Representative"}, {"industry": "Manufacturing", "is_active": true, "company": "Hooli", "status": "Overdue", "job_title": "Chief Executive Officer"}, {"industry": "Healthcare", "is_active": true, "company": "Cyberdyne", "status": "Overdue", "job_title": "Vice President"}, {"industry": "Healthcare", "is_active": true, "company": "Initech", "status": "Refunded", "job_title": "Sales Representative"}, {"industry": "Healthcare", "is_active": true, "company": "Initech", "status": "Refunded", "job_title": "Director"}, {"industry": "Manufacturing", "is_active": false, "company": "Stark Industries", "status": "Refunded", "job_title": "Director"}, {"industry": "Education", "is_active": false, "company": "Initech", "status": "Refunded", "job_title": "Director"}, {"industry": "Manufacturing", "is_active": false, "company": "Hooli", "status": "Refunded", "job_title": "Director"}, {"industry": "Manufacturing", "is_active": true, "company": "Hooli", "status": "Overdue", "job_title": "Sales Representative"}], "dirty_cols": ["industry", "is_active", "company", "status", "job_title"], "clean_cols": ["industry", "is_active", "company", "status", "job_title"], "plan": {"dataset_summary": "12 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Medical": "Healthcare", "Industrial": "Manufacturing", "health care": "Healthcare", "Medcial": "Healthcare", "HEALTH": "Healthcare", "Mfg": "Manufacturing", "education": "Education", "mfg": "Manufacturing"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"late": "Overdue", "refund": "Refunded", "overdue": "Overdue", "refnuded": "Refunded", "Refund": "Refunded", "refunded": "Refunded", "Late": "Overdue"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"chief executive officer": "Chief Executive Officer", "Rep": "Sales Representative", "ceo": "Chief Executive Officer", "SALES REPRESENTATIVE": "Sales Representative", "dir.": "Director", "director": "Director", "DIR": "Director", "Dir": "Director", "REP": "Sales Representative"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "3/16/2023", "status": "churned", "unit": "  kilometers", "currency": "Yuan Renminbi", "company": "Soylent Corp", "job_title": "PM", "unnamed": ""}, {"signup_date": "24 Apr 2023", "status": "Churned", "unit": " kilometers", "currency": " CNY ", "company": "Initech", "job_title": "Director", "unnamed": ""}, {"signup_date": "12 May 2023", "status": "ACTIVE", "unit": "kilometer", "currency": " cny", "company": " Initech ", "job_title": "Product Manager", "unnamed": ""}, {"signup_date": "25 Jan 2023", "status": "TRIAL", "unit": " KILOMMETER", "currency": "DOP", "company": "Vehement", "job_title": "diirector", "unnamed": ""}, {"signup_date": "2023-03-25", "status": "active", "unit": " Cm", "currency": "   domniican peso", "company": "Acme Inc", "job_title": "PRODUCT MANAGER", "unnamed": ""}, {"signup_date": "22 Jul 2023", "status": " Cancelled ", "unit": "cm", "currency": "  CNY", "company": "   Stark Industries", "job_title": "Chief Executive Officer", "unnamed": ""}, {"signup_date": "45097", "status": "trial", "unit": "   centimeters", "currency": "dop", "company": "Stark Industries", "job_title": "C.E.O.", "unnamed": ""}, {"signup_date": "7/14/2023", "status": "Churned", "unit": "cm", "currency": "DOP", "company": "Umbrella", "job_title": "   chief executive officer", "unnamed": ""}, {"signup_date": "45209", "status": "active", "unit": "kilometer", "currency": "DOP", "company": "   Soylent Corp", "job_title": "CEO", "unnamed": ""}, {"signup_date": "7/27/2023", "status": "active   ", "unit": "cm", "currency": "DOP", "company": "Wonka  Co", "job_title": "DIR.", "unnamed": ""}, {"signup_date": "45231", "status": "CANCELLED", "unit": "  centimeters", "currency": "CNY", "company": "Soylent  Corp", "job_title": "Prod Mgr", "unnamed": ""}, {"signup_date": "45036", "status": "trial", "unit": "cm", "currency": "DOP", "company": "Umbrella", "job_title": "DIRECTOR", "unnamed": ""}, {"signup_date": "45170", "status": "  paused", "unit": "Kilometer ", "currency": "yuan renminbi", "company": "Soylent Corp", "job_title": "Prod Manager", "unnamed": ""}, {"signup_date": "11/28/2023", "status": "churned", "unit": "   cm", "currency": "CNY", "company": "   Wonka Co", "job_title": "Chief Executive Officer", "unnamed": ""}, {"signup_date": "45017", "status": "acitve", "unit": " centimeiers ", "currency": "yuan renminbi", "company": "Acme Inc ", "job_title": "Prod Manager", "unnamed": ""}, {"signup_date": "45170", "status": "  paused", "unit": "Kilometer ", "currency": "yuan renminbi", "company": "Soylent Corp", "job_title": "Prod Manager", "unnamed": ""}, {"signup_date": "22 Jul 2023", "status": " Cancelled ", "unit": "cm", "currency": "  CNY", "company": "   Stark Industries", "job_title": "Chief Executive Officer", "unnamed": ""}, {"signup_date": "", "status": "", "unit": "", "currency": "", "company": "", "job_title": "", "unnamed": ""}], "clean": [{"signup_date": "2023-03-16", "status": "Churned", "unit": "km", "currency": "CNY", "company": "Soylent Corp", "job_title": "Product Manager"}, {"signup_date": "2023-04-24", "status": "Churned", "unit": "km", "currency": "CNY", "company": "Initech", "job_title": "Director"}, {"signup_date": "2023-05-12", "status": "Active", "unit": "km", "currency": "CNY", "company": "Initech", "job_title": "Product Manager"}, {"signup_date": "2023-01-25", "status": "Trial", "unit": "km", "currency": "DOP", "company": "Vehement", "job_title": "Director"}, {"signup_date": "2023-03-25", "status": "Active", "unit": "cm", "currency": "DOP", "company": "Acme Inc", "job_title": "Product Manager"}, {"signup_date": "2023-07-22", "status": "Churned", "unit": "cm", "currency": "CNY", "company": "Stark Industries", "job_title": "Chief Executive Officer"}, {"signup_date": "2023-06-20", "status": "Trial", "unit": "cm", "currency": "DOP", "company": "Stark Industries", "job_title": "Chief Executive Officer"}, {"signup_date": "2023-07-14", "status": "Churned", "unit": "cm", "currency": "DOP", "company": "Umbrella", "job_title": "Chief Executive Officer"}, {"signup_date": "2023-10-10", "status": "Active", "unit": "km", "currency": "DOP", "company": "Soylent Corp", "job_title": "Chief Executive Officer"}, {"signup_date": "2023-07-27", "status": "Active", "unit": "cm", "currency": "DOP", "company": "Wonka Co", "job_title": "Director"}, {"signup_date": "2023-11-01", "status": "Churned", "unit": "cm", "currency": "CNY", "company": "Soylent Corp", "job_title": "Product Manager"}, {"signup_date": "2023-04-20", "status": "Trial", "unit": "cm", "currency": "DOP", "company": "Umbrella", "job_title": "Director"}, {"signup_date": "2023-09-01", "status": "Paused", "unit": "km", "currency": "CNY", "company": "Soylent Corp", "job_title": "Product Manager"}, {"signup_date": "2023-11-28", "status": "Churned", "unit": "cm", "currency": "CNY", "company": "Wonka Co", "job_title": "Chief Executive Officer"}, {"signup_date": "2023-04-01", "status": "Active", "unit": "cm", "currency": "CNY", "company": "Acme Inc", "job_title": "Product Manager"}], "dirty_cols": ["signup_date", "status", "unit", "currency", "company", "job_title", "unnamed"], "clean_cols": ["signup_date", "status", "unit", "currency", "company", "job_title"], "plan": {"dataset_summary": "18 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"churned": "Churned", "ACTIVE": "Active", "TRIAL": "Trial", "active": "Active", "Cancelled": "Churned", "trial": "Trial", "CANCELLED": "Churned", "paused": "Paused", "acitve": "Active"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilometers": "km", "kilometer": "km", "KILOMMETER": "km", "Cm": "cm", "centimeters": "cm", "Kilometer": "km", "centimeiers": "cm"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Yuan Renminbi": "CNY", "cny": "CNY", "domniican peso": "DOP", "dop": "DOP", "yuan renminbi": "CNY"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PM": "Product Manager", "diirector": "Director", "PRODUCT MANAGER": "Product Manager", "C.E.O.": "Chief Executive Officer", "chief executive officer": "Chief Executive Officer", "CEO": "Chief Executive Officer", "DIR.": "Director", "Prod Mgr": "Product Manager", "DIRECTOR": "Director", "Prod Manager": "Product Manager"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"industry": "REALESTATE", "country": "arg", "signup_date": "2023-11-06", "status": "Closed Lost", "unit": "liter", "state": "WY", "is_active": "TRUE"}, {"industry": "PROPERTY", "country": "GRD", "signup_date": "12/3/2023", "status": "in-progress", "unit": "Cm", "state": "wy", "is_active": "TRUE"}, {"industry": "banking   ", "country": "Independent State of Samoa", "signup_date": "16 Jul 2023", "status": "Closed Won", "unit": "cm", "state": " wy ", "is_active": "F"}, {"industry": "Realestate", "country": "GD", "signup_date": "2023-12-18", "status": "closed won", "unit": "centimeter", "state": "KANSAS", "is_active": "N"}, {"industry": " Fintech", "country": "Argentine  Republic", "signup_date": "45047", "status": "   clpsed lost", "unit": "CM", "state": "Kansas", "is_active": "false"}, {"industry": "REAL ESTATE", "country": "  WSM", "signup_date": "6/24/2023", "status": "Lhst  ", "unit": "CENTIMETERS", "state": "ks", "is_active": "FALSE"}, {"industry": "Fintech", "country": "Grenada", "signup_date": "2023-12-17", "status": "wip", "unit": "litre", "state": " WY", "is_active": "T"}, {"industry": "FINANCIAL SERVICES", "country": "gd", "signup_date": "12 Nov 2023", "status": "closed lost", "unit": "centimeter", "state": "Wyoming", "is_active": "F"}, {"industry": "proptech", "country": "AR", "signup_date": "3/5/2023", "status": "in progress", "unit": "cm", "state": "wy", "is_active": "N"}, {"industry": "realestate", "country": "grd", "signup_date": "2023-07-09", "status": "in-progrress", "unit": "cm", "state": " kansas ", "is_active": "Y"}, {"industry": "banking", "country": "GRENADA   ", "signup_date": "9/23/2023", "status": "lost", "unit": "CENTIMETER  ", "state": "  wyoming", "is_active": "T"}, {"industry": "Real Estate", "country": "INDEPENDENT STATE OF SAMOA", "signup_date": "2023-03-24", "status": "ongoing", "unit": "l", "state": "KS", "is_active": "false"}, {"industry": "FINANCIAL SERVICES", "country": "ws", "signup_date": "2023-11-18", "status": "wip", "unit": "cm", "state": "KS", "is_active": "false"}, {"industry": "Finteeh", "country": "AR", "signup_date": "45044", "status": "lost", "unit": "litre", "state": "Wy", "is_active": "N"}, {"industry": "banking", "country": "Grenadda", "signup_date": "45279", "status": "won", "unit": "LITRE", "state": "kansas", "is_active": "false"}, {"industry": "   PropTech", "country": "ar", "signup_date": "7/17/2023", "status": "closed lost", "unit": "CM", "state": "Kansas", "is_active": "0"}, {"industry": "  finserv", "country": "Argentine Republic", "signup_date": "2023-07-24", "status": "closed-won", "unit": "cm", "state": " WY", "is_active": "0"}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "45112", "status": "wip ", "unit": "liters", "state": "  ks", "is_active": "F"}, {"industry": "Financial Services", "country": "samoa   ", "signup_date": "2023-05-13", "status": "Ongoing", "unit": "cm", "state": "Kansas", "is_active": "Y"}, {"industry": "Real Estate", "country": "Argentine Republic", "signup_date": "3/21/2023", "status": "in pfogress", "unit": " liter ", "state": "wy", "is_active": "TRUE"}, {"industry": "Property", "country": "WS", "signup_date": "2023-01-15", "status": "  won", "unit": "l ", "state": "WY  ", "is_active": "Yes"}, {"industry": " proptech ", "country": "Arggentina", "signup_date": "2023-12-15", "status": "Closed Lost", "unit": "L", "state": "Ks", "is_active": "No"}, {"industry": "RealEstate", "country": "WS", "signup_date": "1/18/2023", "status": "Lsot", "unit": "centimeter", "state": "KS  ", "is_active": "N"}, {"industry": "   proptech", "country": "Independent  State of Samoa", "signup_date": "12/22/2023", "status": "won", "unit": "liter", "state": "KS", "is_active": "false"}, {"industry": "Property", "country": "SAAMOA", "signup_date": "2023-10-15", "status": "Closed Lost", "unit": " CENTIMETERS ", "state": " Wyoming ", "is_active": "T"}, {"industry": "Banking", "country": " Ar ", "signup_date": "45195", "status": "closd lost   ", "unit": "centimeter", "state": "Kansas", "is_active": "No"}, {"industry": "finance", "country": "Wsm  ", "signup_date": "45109", "status": "Closed Lost", "unit": "Cm", "state": "wyoming", "is_active": "FALSE"}, {"industry": "   FINSERV", "country": "wsm", "signup_date": "45197", "status": "   Closed Lost", "unit": " L ", "state": "  Ks", "is_active": "No"}, {"industry": "Property", "country": "GRD", "signup_date": "2/3/2023", "status": "Lost", "unit": "centimetre", "state": "Kansas", "is_active": "TRUE"}, {"industry": "Proerty", "country": "INDEPENDENT STATE OF SAMOA", "signup_date": "27 Sep 2023", "status": "CLOSED WON", "unit": " cm ", "state": " wy ", "is_active": "TRUE"}, {"industry": " Property ", "country": "argentine republic", "signup_date": "2023-09-26", "status": "inprogress", "unit": "cm", "state": "WY", "is_active": "true"}, {"industry": "fintech", "country": "Grenada", "signup_date": "12 May 2023", "status": "IN PROGRESS", "unit": "liter", "state": "Wyoming ", "is_active": "Yes"}, {"industry": " PropTech", "country": "Saqoa", "signup_date": "45079", "status": "In Progress", "unit": "liters", "state": "Wyoming", "is_active": "T"}, {"industry": "Property", "country": "GRD ", "signup_date": "14 Dec 2023", "status": " won ", "unit": "CENTIMETERS   ", "state": "KS", "is_active": "Y"}, {"industry": "realestate", "country": "GD", "signup_date": "45179", "status": "Won", "unit": "centimeter", "state": "WY", "is_active": "Yes"}, {"industry": " Banking ", "country": "AR", "signup_date": "17 Apr 2023", "status": "IN PROGRESS", "unit": "L", "state": "Wyoming", "is_active": "0"}, {"industry": "property", "country": "GD", "signup_date": "20 Aug 2023", "status": "lost", "unit": "centimeters", "state": "wy", "is_active": "TRUE"}, {"industry": "property", "country": "Grenada", "signup_date": "11 Dec 2023", "status": "closed lost", "unit": "liter", "state": "wy", "is_active": "N"}, {"industry": "PropTecch", "country": "Independent State of Samoa", "signup_date": "12/4/2023", "status": "wip", "unit": "liters", "state": " Knasas ", "is_active": "1"}, {"industry": "Property ", "country": "ARGENTINA", "signup_date": "5 Jul 2023", "status": "Cloosed Lost", "unit": "centimeter", "state": "ks", "is_active": "TRUE"}, {"industry": "finance", "country": "Grd", "signup_date": "45200", "status": "Closed Won", "unit": "L", "state": "KANSAS", "is_active": "true"}, {"industry": "   FinServ", "country": "  GD", "signup_date": "5/24/2023", "status": "WON", "unit": "centimeters", "state": "  Kansas", "is_active": "N"}, {"industry": "REAL EDTATE", "country": "grd", "signup_date": "2023-03-01", "status": "WIP", "unit": "centimeters", "state": "Kansas", "is_active": "T"}, {"industry": "Real Estate", "country": " WSM", "signup_date": "2023-10-28", "status": "Lost", "unit": "centimeters", "state": "KANSAS", "is_active": "true"}, {"industry": "REALESTATE", "country": "Grenada", "signup_date": "45057", "status": "ongoing", "unit": "Cm", "state": "ks", "is_active": "T"}, {"industry": "Realestate", "country": "WSM", "signup_date": "2023-05-28", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": "Y"}, {"industry": "Banking", "country": "Ar", "signup_date": "26 Nov 2023", "status": "closed-won", "unit": "CENTIMETER", "state": "Wy", "is_active": "1"}, {"industry": "BANKING", "country": " GRD ", "signup_date": "44993", "status": "won", "unit": "LITER", "state": "Kansas", "is_active": "F"}, {"industry": "finance", "country": "GD", "signup_date": "4/27/2023", "status": "Lost", "unit": "L", "state": "WY", "is_active": "1"}, {"industry": "fintech", "country": "ARG", "signup_date": "10/15/2023", "status": "in-progress", "unit": "litre", "state": "Kansas", "is_active": "T"}, {"industry": "FINANCIAL SERVICES", "country": "WS", "signup_date": "45062", "status": "   ClosedWon", "unit": "Centimeter", "state": "wy", "is_active": "N"}, {"industry": "Banking ", "country": "WS", "signup_date": "2023-03-10", "status": "Closed Lost", "unit": "centimeter", "state": "Wyoming   ", "is_active": "T"}, {"industry": "property", "country": "grenada", "signup_date": "12 Apr 2023", "status": "in-progress", "unit": "   liteer", "state": "Wyoming", "is_active": "Yes"}, {"industry": "Finance", "country": "Independent State of Samoa", "signup_date": "28 Aug 2023", "status": "Wip", "unit": "Cm", "state": "Kansas", "is_active": "Y"}, {"industry": "RealEstate", "country": "   Argentina", "signup_date": "18 Jun 2023", "status": "   Wip", "unit": "cm", "state": "Ks  ", "is_active": "Y"}, {"industry": "Banking", "country": "grd", "signup_date": "23 Nov 2023", "status": "WIP", "unit": "litre", "state": "WY", "is_active": "FALSE"}, {"industry": "", "country": "", "signup_date": "", "status": "", "unit": "", "state": "", "is_active": ""}], "clean": [{"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-11-06", "status": "Lost", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-12-03", "status": "In Progress", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-07-16", "status": "Won", "unit": "cm", "state": "Wyoming", "is_active": false}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-12-18", "status": "Won", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-05-01", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-06-24", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-12-17", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-11-12", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": false}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-03-05", "status": "In Progress", "unit": "cm", "state": "Wyoming", "is_active": false}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-07-09", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-09-23", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-03-24", "status": "In Progress", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-11-18", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-04-28", "status": "Lost", "unit": "L", "state": "Wyoming", "is_active": false}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-12-19", "status": "Won", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-07-17", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-07-24", "status": "Won", "unit": "cm", "state": "Wyoming", "is_active": false}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-07-05", "status": "In Progress", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-05-13", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-03-21", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-01-15", "status": "Won", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-12-15", "status": "Lost", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-01-18", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-12-22", "status": "Won", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-10-15", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-09-26", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-07-02", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": false}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-09-28", "status": "Lost", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-02-03", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-09-27", "status": "Won", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-09-26", "status": "In Progress", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-05-12", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-06-02", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-12-14", "status": "Won", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-09-10", "status": "Won", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-04-17", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": false}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-08-20", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-12-11", "status": "Lost", "unit": "L", "state": "Wyoming", "is_active": false}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-12-04", "status": "In Progress", "unit": "L", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-07-05", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-10-01", "status": "Won", "unit": "L", "state": "Kansas", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-05-24", "status": "Won", "unit": "cm", "state": "Kansas", "is_active": false}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-03-01", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-10-28", "status": "Lost", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-05-11", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Samoa", "signup_date": "2023-05-28", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-11-26", "status": "Won", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-03-08", "status": "Won", "unit": "L", "state": "Kansas", "is_active": false}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-04-27", "status": "Lost", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Argentina", "signup_date": "2023-10-15", "status": "In Progress", "unit": "L", "state": "Kansas", "is_active": true}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-05-16", "status": "Won", "unit": "cm", "state": "Wyoming", "is_active": false}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-03-10", "status": "Lost", "unit": "cm", "state": "Wyoming", "is_active": true}, {"industry": "Real Estate", "country": "Grenada", "signup_date": "2023-04-12", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": true}, {"industry": "Financial Services", "country": "Samoa", "signup_date": "2023-08-28", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Real Estate", "country": "Argentina", "signup_date": "2023-06-18", "status": "In Progress", "unit": "cm", "state": "Kansas", "is_active": true}, {"industry": "Financial Services", "country": "Grenada", "signup_date": "2023-11-23", "status": "In Progress", "unit": "L", "state": "Wyoming", "is_active": false}], "dirty_cols": ["industry", "country", "signup_date", "status", "unit", "state", "is_active"], "clean_cols": ["industry", "country", "signup_date", "status", "unit", "state", "is_active"], "plan": {"dataset_summary": "57 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"REALESTATE": "Real Estate", "PROPERTY": "Real Estate", "banking": "Financial Services", "Realestate": "Real Estate", "Fintech": "Financial Services", "REAL ESTATE": "Real Estate", "FINANCIAL SERVICES": "Financial Services", "proptech": "Real Estate", "realestate": "Real Estate", "Finteeh": "Financial Services", "PropTech": "Real Estate", "finserv": "Financial Services", "Property": "Real Estate", "RealEstate": "Real Estate", "Banking": "Financial Services", "finance": "Financial Services", "FINSERV": "Financial Services", "Proerty": "Real Estate", "fintech": "Financial Services", "property": "Real Estate", "PropTecch": "Real Estate", "FinServ": "Financial Services", "REAL EDTATE": "Real Estate", "BANKING": "Financial Services", "Finance": "Financial Services"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"arg": "Argentina", "GRD": "Grenada", "Independent State of Samoa": "Samoa", "GD": "Grenada", "Argentine Republic": "Argentina", "WSM": "Samoa", "gd": "Grenada", "AR": "Argentina", "grd": "Grenada", "GRENADA": "Grenada", "INDEPENDENT STATE OF SAMOA": "Samoa", "ws": "Samoa", "Grenadda": "Grenada", "ar": "Argentina", "samoa": "Samoa", "WS": "Samoa", "Arggentina": "Argentina", "SAAMOA": "Samoa", "Ar": "Argentina", "Wsm": "Samoa", "wsm": "Samoa", "argentine republic": "Argentina", "Saqoa": "Samoa", "ARGENTINA": "Argentina", "Grd": "Grenada", "ARG": "Argentina", "grenada": "Grenada"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Closed Lost": "Lost", "in-progress": "In Progress", "Closed Won": "Won", "closed won": "Won", "clpsed lost": "Lost", "Lhst": "Lost", "wip": "In Progress", "closed lost": "Lost", "in progress": "In Progress", "in-progrress": "In Progress", "lost": "Lost", "ongoing": "In Progress", "won": "Won", "closed-won": "Won", "Ongoing": "In Progress", "in pfogress": "In Progress", "Lsot": "Lost", "closd lost": "Lost", "CLOSED WON": "Won", "inprogress": "In Progress", "IN PROGRESS": "In Progress", "Cloosed Lost": "Lost", "WON": "Won", "WIP": "In Progress", "ClosedWon": "Won", "Wip": "In Progress"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liter": "L", "Cm": "cm", "centimeter": "cm", "CM": "cm", "CENTIMETERS": "cm", "litre": "L", "CENTIMETER": "cm", "l": "L", "LITRE": "L", "liters": "L", "centimetre": "cm", "centimeters": "cm", "LITER": "L", "Centimeter": "cm", "liteer": "L"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WY": "Wyoming", "wy": "Wyoming", "KANSAS": "Kansas", "ks": "Kansas", "kansas": "Kansas", "wyoming": "Wyoming", "KS": "Kansas", "Wy": "Wyoming", "Ks": "Kansas", "Knasas": "Kansas"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"signup_date": "8/8/2023", "industry": "property   ", "rate": "69.6%", "job_title": "Account Executive", "amount": "4.954,18", "currency": "ALL"}, {"signup_date": "45022", "industry": "industrial", "rate": "42.3%", "job_title": "Vice  Pres", "amount": "7.080,46", "currency": "azn"}, {"signup_date": "6/18/2023", "industry": " mfg", "rate": "65.8%", "job_title": "ACCT EXEC", "amount": "1.932,58", "currency": "  all"}, {"signup_date": "45258", "industry": " proptech ", "rate": "13.0%", "job_title": "Vice Pers", "amount": "$3,820.89", "currency": "azn"}, {"signup_date": "2023-06-03", "industry": "Health Care", "rate": "8.3%", "job_title": "vice pres   ", "amount": "8.544,20", "currency": "AZN"}, {"signup_date": "45000", "industry": " INDUSTRIAL ", "rate": "52.0%", "job_title": "Senior Eng", "amount": "$8,594.19", "currency": "Lek"}, {"signup_date": "7 Feb 2023", "industry": "Healthcare", "rate": "55.4%", "job_title": "Senior Engineer", "amount": "8.444,84", "currency": "all"}, {"signup_date": "23 Dec 2023", "industry": "Finace", "rate": "67.8%", "job_title": "senior engineer", "amount": "$2,463.40", "currency": "ALL"}, {"signup_date": "45172", "industry": "Heklth", "rate": "63.2%", "job_title": "ae", "amount": "$2,756.06", "currency": "   ALL"}, {"signup_date": "2023-09-27", "industry": "real estate", "rate": "56.0%", "job_title": "VP..", "amount": "7.911,49", "currency": "   All"}, {"signup_date": "3/2/2023", "industry": "Realestate", "rate": "25.9%", "job_title": "account executive", "amount": "8.062,62", "currency": "all"}, {"signup_date": "28 Nov 2023", "industry": "Finance", "rate": "64.6%", "job_title": "  Senior Engineer", "amount": "1.470,50", "currency": "Azerbaijan Manat"}, {"signup_date": "2023-03-12", "industry": "Industrial", "rate": "7.8%", "job_title": "ACCT EXEC", "amount": "7.559,73", "currency": "AZN"}, {"signup_date": "3/7/2023", "industry": "Fintech", "rate": "4.0%", "job_title": "Accz Exec", "amount": "5.909,84", "currency": "  AZERBAIJAN MANAT"}, {"signup_date": "14 Jun 2023", "industry": "Health Cane", "rate": "39.1%", "job_title": "vice president", "amount": "#N/A", "currency": "all"}, {"signup_date": "44938", "industry": " manufacturing ", "rate": "36.2%", "job_title": "Account Executive", "amount": "$609.60", "currency": " Lek "}, {"signup_date": "22 Jan 2023", "industry": "FINDNCE", "rate": "15.1%", "job_title": "acct exec", "amount": "4.524,65", "currency": "azerbaijan manat"}, {"signup_date": "8/18/2023", "industry": "Fintech", "rate": "28.0%", "job_title": "acct exec", "amount": "$5,345.89", "currency": "Azn"}, {"signup_date": "44928", "industry": "Manufacturing", "rate": "16.0%", "job_title": "  Sr. Engineer", "amount": "5.372,96", "currency": "ALL"}, {"signup_date": "4/9/2023", "industry": "healthcare", "rate": "64.1%", "job_title": " vice president", "amount": "$1,385.57", "currency": "all"}, {"signup_date": "45200", "industry": "MEDICAL", "rate": "80.4%", "job_title": "AE", "amount": "1.282,91", "currency": "ALL"}, {"signup_date": "2023-12-02", "industry": "Property", "rate": "43.4%", "job_title": "  ACCT EXEC", "amount": "$4,267.82", "currency": "Azerbaijan Manat"}, {"signup_date": "9/1/2023", "industry": " healtth care", "rate": "24.0%", "job_title": "account executive", "amount": "$2,421.73", "currency": "ALL"}, {"signup_date": "2 Sep 2023", "industry": "financial services", "rate": "47.8%", "job_title": " V.P.", "amount": "$2,461.50", "currency": " ALL"}, {"signup_date": "21 Jul 2023", "industry": "MEDCIAL", "rate": "98.6%", "job_title": "Senior Eng", "amount": "7.125,46", "currency": " AZN "}, {"signup_date": "7/14/2023", "industry": "Real Estate", "rate": "31.0%", "job_title": "VP", "amount": "449,32", "currency": "AZN"}, {"signup_date": "24 Feb 2023", "industry": "Mfg", "rate": "60.9%", "job_title": "Senior Engineer", "amount": "$7,138.42", "currency": "Azerbaijan Manat"}, {"signup_date": "2023-09-14", "industry": "financial services", "rate": "96.7%", "job_title": "Acct Exec", "amount": "$8,334.38", "currency": "AZN"}, {"signup_date": "6/14/2023", "industry": "Prbperty", "rate": "26.0%", "job_title": "V.P.", "amount": "$6,532.82", "currency": "ALL"}, {"signup_date": "9 Mar 2023", "industry": "Fintech", "rate": "14.3%", "job_title": "Snr Enginer", "amount": "6.557,14", "currency": "LEK"}, {"signup_date": "2023-09-20", "industry": "Finserv", "rate": "68.6%", "job_title": "senioreng", "amount": "502,42", "currency": "All"}, {"signup_date": "2023-03-05", "industry": "FINANCIAL SERVICES", "rate": "56.5%", "job_title": "Senior Engineer", "amount": "$1,750.15", "currency": " AZN"}, {"signup_date": "2023-05-09", "industry": "banking", "rate": "26.7%", "job_title": "VP", "amount": "5.065,36", "currency": "LEK"}, {"signup_date": "45112", "industry": "Finance", "rate": "15.5%", "job_title": "account executive", "amount": "$8,159.68", "currency": "Azerbaijan Manat"}, {"signup_date": "44934", "industry": "banking", "rate": "63.5%", "job_title": "Account Executive", "amount": "N/A", "currency": "  Azerbaijan Manat"}, {"signup_date": "2023-05-03", "industry": "realestate", "rate": "42.7%", "job_title": "senior eng", "amount": "4.929,76", "currency": "Lek"}, {"signup_date": "45019", "industry": "Industrial", "rate": "56.2%", "job_title": "Account Executive", "amount": "5.977,22", "currency": "Azerbaijan Manat"}, {"signup_date": "44952", "industry": "Mfg", "rate": "80.9%", "job_title": "Ae", "amount": "8.257,30", "currency": "AZN"}, {"signup_date": "45035", "industry": " PropTech ", "rate": "16.2%", "job_title": "V.P.", "amount": "8.452,93", "currency": "AZN"}, {"signup_date": "2023-11-23", "industry": "Property   ", "rate": "27.1%", "job_title": "ae", "amount": "1.577,38", "currency": "ALL"}, {"signup_date": "20 Dec 2023", "industry": "manufacturing", "rate": "16.7%", "job_title": "Senior Engineer", "amount": "523,02", "currency": "Azerbaijan  Manat"}, {"signup_date": "2023-10-09", "industry": "Manufacturing", "rate": "1.8%", "job_title": "  ACCOUNT EXECUTIVE", "amount": "$6,427.42", "currency": "AZERBAIJAN MANAT"}, {"signup_date": "2023-11-25", "industry": "Real Estate", "rate": "76.3%", "job_title": " AE ", "amount": "3.057,51", "currency": "Azn"}, {"signup_date": "10/11/2023", "industry": "Health CCare", "rate": "54.0%", "job_title": "AccthExec", "amount": "6.071,10", "currency": "AZN"}, {"signup_date": "5 Jan 2023", "industry": "mfg", "rate": "6.4%", "job_title": "SR ENGINEER", "amount": "699,01", "currency": "lek"}, {"signup_date": "19 Jun 2023", "industry": "Medical", "rate": "65.2%", "job_title": " Vp ", "amount": "$8,868.93", "currency": "AZN"}, {"signup_date": "8/15/2023", "industry": "Banking", "rate": "7.9%", "job_title": "V.P.", "amount": "6.843,25", "currency": "Azerbaijan  Manat"}, {"signup_date": "45093", "industry": "Fitnech", "rate": "66.1%", "job_title": "V.P.", "amount": "null", "currency": "Lek"}, {"signup_date": "45255", "industry": "manufacturing", "rate": "84.7%", "job_title": "Accoount Executive", "amount": "na", "currency": "Lek"}, {"signup_date": "45006", "industry": "  Industrial", "rate": "61.0%", "job_title": "Vice Pres", "amount": "2.237,57", "currency": "Azerbaijan Manat"}, {"signup_date": "", "industry": "", "rate": "", "job_title": "", "amount": "", "currency": ""}], "clean": [{"signup_date": "2023-08-08", "industry": "Real Estate", "rate": 0.696, "job_title": "Account Executive", "amount": 4954.18, "currency": "ALL"}, {"signup_date": "2023-04-06", "industry": "Manufacturing", "rate": 0.423, "job_title": "Vice President", "amount": 7080.46, "currency": "AZN"}, {"signup_date": "2023-06-18", "industry": "Manufacturing", "rate": 0.6579999999999999, "job_title": "Account Executive", "amount": 1932.58, "currency": "ALL"}, {"signup_date": "2023-11-28", "industry": "Real Estate", "rate": 0.13, "job_title": "Vice President", "amount": 3820.89, "currency": "AZN"}, {"signup_date": "2023-06-03", "industry": "Healthcare", "rate": 0.083, "job_title": "Vice President", "amount": 8544.2, "currency": "AZN"}, {"signup_date": "2023-03-15", "industry": "Manufacturing", "rate": 0.52, "job_title": "Senior Engineer", "amount": 8594.19, "currency": "ALL"}, {"signup_date": "2023-02-07", "industry": "Healthcare", "rate": 0.5539999999999999, "job_title": "Senior Engineer", "amount": 8444.84, "currency": "ALL"}, {"signup_date": "2023-12-23", "industry": "Financial Services", "rate": 0.6779999999999999, "job_title": "Senior Engineer", "amount": 2463.4, "currency": "ALL"}, {"signup_date": "2023-09-03", "industry": "Healthcare", "rate": 0.632, "job_title": "Account Executive", "amount": 2756.06, "currency": "ALL"}, {"signup_date": "2023-09-27", "industry": "Real Estate", "rate": 0.56, "job_title": "Vice President", "amount": 7911.49, "currency": "ALL"}, {"signup_date": "2023-03-02", "industry": "Real Estate", "rate": 0.259, "job_title": "Account Executive", "amount": 8062.62, "currency": "ALL"}, {"signup_date": "2023-11-28", "industry": "Financial Services", "rate": 0.6459999999999999, "job_title": "Senior Engineer", "amount": 1470.5, "currency": "AZN"}, {"signup_date": "2023-03-12", "industry": "Manufacturing", "rate": 0.078, "job_title": "Account Executive", "amount": 7559.73, "currency": "AZN"}, {"signup_date": "2023-03-07", "industry": "Financial Services", "rate": 0.04, "job_title": "Account Executive", "amount": 5909.84, "currency": "AZN"}, {"signup_date": "2023-06-14", "industry": "Healthcare", "rate": 0.391, "job_title": "Vice President", "amount": NaN, "currency": "ALL"}, {"signup_date": "2023-01-12", "industry": "Manufacturing", "rate": 0.36200000000000004, "job_title": "Account Executive", "amount": 609.6, "currency": "ALL"}, {"signup_date": "2023-01-22", "industry": "Financial Services", "rate": 0.151, "job_title": "Account Executive", "amount": 4524.65, "currency": "AZN"}, {"signup_date": "2023-08-18", "industry": "Financial Services", "rate": 0.28, "job_title": "Account Executive", "amount": 5345.89, "currency": "AZN"}, {"signup_date": "2023-01-02", "industry": "Manufacturing", "rate": 0.16, "job_title": "Senior Engineer", "amount": 5372.96, "currency": "ALL"}, {"signup_date": "2023-04-09", "industry": "Healthcare", "rate": 0.6409999999999999, "job_title": "Vice President", "amount": 1385.57, "currency": "ALL"}, {"signup_date": "2023-10-01", "industry": "Healthcare", "rate": 0.804, "job_title": "Account Executive", "amount": 1282.91, "currency": "ALL"}, {"signup_date": "2023-12-02", "industry": "Real Estate", "rate": 0.434, "job_title": "Account Executive", "amount": 4267.82, "currency": "AZN"}, {"signup_date": "2023-09-01", "industry": "Healthcare", "rate": 0.24, "job_title": "Account Executive", "amount": 2421.73, "currency": "ALL"}, {"signup_date": "2023-09-02", "industry": "Financial Services", "rate": 0.478, "job_title": "Vice President", "amount": 2461.5, "currency": "ALL"}, {"signup_date": "2023-07-21", "industry": "Healthcare", "rate": 0.986, "job_title": "Senior Engineer", "amount": 7125.46, "currency": "AZN"}, {"signup_date": "2023-07-14", "industry": "Real Estate", "rate": 0.31, "job_title": "Vice President", "amount": 449.32, "currency": "AZN"}, {"signup_date": "2023-02-24", "industry": "Manufacturing", "rate": 0.609, "job_title": "Senior Engineer", "amount": 7138.42, "currency": "AZN"}, {"signup_date": "2023-09-14", "industry": "Financial Services", "rate": 0.9670000000000001, "job_title": "Account Executive", "amount": 8334.38, "currency": "AZN"}, {"signup_date": "2023-06-14", "industry": "Real Estate", "rate": 0.26, "job_title": "Vice President", "amount": 6532.82, "currency": "ALL"}, {"signup_date": "2023-03-09", "industry": "Financial Services", "rate": 0.14300000000000002, "job_title": "Senior Engineer", "amount": 6557.14, "currency": "ALL"}, {"signup_date": "2023-09-20", "industry": "Financial Services", "rate": 0.6859999999999999, "job_title": "Senior Engineer", "amount": 502.42, "currency": "ALL"}, {"signup_date": "2023-03-05", "industry": "Financial Services", "rate": 0.565, "job_title": "Senior Engineer", "amount": 1750.15, "currency": "AZN"}, {"signup_date": "2023-05-09", "industry": "Financial Services", "rate": 0.267, "job_title": "Vice President", "amount": 5065.36, "currency": "ALL"}, {"signup_date": "2023-07-05", "industry": "Financial Services", "rate": 0.155, "job_title": "Account Executive", "amount": 8159.68, "currency": "AZN"}, {"signup_date": "2023-01-08", "industry": "Financial Services", "rate": 0.635, "job_title": "Account Executive", "amount": NaN, "currency": "AZN"}, {"signup_date": "2023-05-03", "industry": "Real Estate", "rate": 0.42700000000000005, "job_title": "Senior Engineer", "amount": 4929.76, "currency": "ALL"}, {"signup_date": "2023-04-03", "industry": "Manufacturing", "rate": 0.562, "job_title": "Account Executive", "amount": 5977.22, "currency": "AZN"}, {"signup_date": "2023-01-26", "industry": "Manufacturing", "rate": 0.809, "job_title": "Account Executive", "amount": 8257.3, "currency": "AZN"}, {"signup_date": "2023-04-19", "industry": "Real Estate", "rate": 0.162, "job_title": "Vice President", "amount": 8452.93, "currency": "AZN"}, {"signup_date": "2023-11-23", "industry": "Real Estate", "rate": 0.271, "job_title": "Account Executive", "amount": 1577.38, "currency": "ALL"}, {"signup_date": "2023-12-20", "industry": "Manufacturing", "rate": 0.16699999999999998, "job_title": "Senior Engineer", "amount": 523.02, "currency": "AZN"}, {"signup_date": "2023-10-09", "industry": "Manufacturing", "rate": 0.018000000000000002, "job_title": "Account Executive", "amount": 6427.42, "currency": "AZN"}, {"signup_date": "2023-11-25", "industry": "Real Estate", "rate": 0.763, "job_title": "Account Executive", "amount": 3057.51, "currency": "AZN"}, {"signup_date": "2023-10-11", "industry": "Healthcare", "rate": 0.54, "job_title": "Account Executive", "amount": 6071.1, "currency": "AZN"}, {"signup_date": "2023-01-05", "industry": "Manufacturing", "rate": 0.064, "job_title": "Senior Engineer", "amount": 699.01, "currency": "ALL"}, {"signup_date": "2023-06-19", "industry": "Healthcare", "rate": 0.652, "job_title": "Vice President", "amount": 8868.93, "currency": "AZN"}, {"signup_date": "2023-08-15", "industry": "Financial Services", "rate": 0.079, "job_title": "Vice President", "amount": 6843.25, "currency": "AZN"}, {"signup_date": "2023-06-16", "industry": "Financial Services", "rate": 0.6609999999999999, "job_title": "Vice President", "amount": NaN, "currency": "ALL"}, {"signup_date": "2023-11-25", "industry": "Manufacturing", "rate": 0.847, "job_title": "Account Executive", "amount": NaN, "currency": "ALL"}, {"signup_date": "2023-03-21", "industry": "Manufacturing", "rate": 0.61, "job_title": "Vice President", "amount": 2237.57, "currency": "AZN"}], "dirty_cols": ["signup_date", "industry", "rate", "job_title", "amount", "currency"], "clean_cols": ["signup_date", "industry", "rate", "job_title", "amount", "currency"], "plan": {"dataset_summary": "51 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"property": "Real Estate", "industrial": "Manufacturing", "mfg": "Manufacturing", "proptech": "Real Estate", "Health Care": "Healthcare", "INDUSTRIAL": "Manufacturing", "Finace": "Financial Services", "Heklth": "Healthcare", "real estate": "Real Estate", "Realestate": "Real Estate", "Finance": "Financial Services", "Industrial": "Manufacturing", "Fintech": "Financial Services", "Health Cane": "Healthcare", "manufacturing": "Manufacturing", "FINDNCE": "Financial Services", "healthcare": "Healthcare", "MEDICAL": "Healthcare", "Property": "Real Estate", "healtth care": "Healthcare", "financial services": "Financial Services", "MEDCIAL": "Healthcare", "Mfg": "Manufacturing", "Prbperty": "Real Estate", "Finserv": "Financial Services", "FINANCIAL SERVICES": "Financial Services", "banking": "Financial Services", "realestate": "Real Estate", "PropTech": "Real Estate", "Health CCare": "Healthcare", "Medical": "Healthcare", "Banking": "Financial Services", "Fitnech": "Financial Services"}, "rationale": "Unified 33 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Vice  Pres": "Vice President", "ACCT EXEC": "Account Executive", "Vice Pers": "Vice President", "vice pres": "Vice President", "Senior Eng": "Senior Engineer", "senior engineer": "Senior Engineer", "ae": "Account Executive", "VP..": "Vice President", "account executive": "Account Executive", "Accz Exec": "Account Executive", "vice president": "Vice President", "acct exec": "Account Executive", "Sr. Engineer": "Senior Engineer", "AE": "Account Executive", "V.P.": "Vice President", "VP": "Vice President", "Acct Exec": "Account Executive", "Snr Enginer": "Senior Engineer", "senioreng": "Senior Engineer", "senior eng": "Senior Engineer", "Ae": "Account Executive", "ACCOUNT EXECUTIVE": "Account Executive", "AccthExec": "Account Executive", "SR ENGINEER": "Senior Engineer", "Vp": "Vice President", "Accoount Executive": "Account Executive", "Vice Pres": "Vice President"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"azn": "AZN", "all": "ALL", "Lek": "ALL", "All": "ALL", "Azerbaijan Manat": "AZN", "AZERBAIJAN MANAT": "AZN", "azerbaijan manat": "AZN", "Azn": "AZN", "LEK": "ALL", "lek": "ALL"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "45056", "state": "United States Minor Outlying Islands", "currency": "falkland islands pound", "department": " Legal", "is_active": "TRUE", "notes2": ""}, {"signup_date": "20 Aug 2023", "state": "alaska", "currency": "falkland islands pound", "department": " IT ", "is_active": "Y", "notes2": ""}, {"signup_date": "10/6/2023", "state": "Ohio", "currency": "GYD", "department": "Legal", "is_active": "N", "notes2": ""}, {"signup_date": "7/11/2023", "state": "   United States Minor Outlying Islands", "currency": "GYD", "department": " Eng", "is_active": "1", "notes2": ""}, {"signup_date": "45058", "state": "Mihigan", "currency": "FKP", "department": "Legal & Compliance", "is_active": "0", "notes2": ""}, {"signup_date": "12/24/2023", "state": "MI", "currency": "FKP", "department": "Information Technology", "is_active": "1", "notes2": ""}, {"signup_date": "7/20/2023", "state": "UM", "currency": "GYD", "department": "Dev   ", "is_active": "1", "notes2": ""}, {"signup_date": "22 Aug 2023", "state": "um", "currency": "Guyana Dollar", "department": "it", "is_active": "F", "notes2": ""}, {"signup_date": "22 Jun 2023", "state": "WI ", "currency": "Guyana Dollar", "department": "IT", "is_active": "false", "notes2": ""}, {"signup_date": "2023-10-02", "state": "um", "currency": "FALKLAND ISLANDS POUND", "department": "legal", "is_active": "0", "notes2": ""}, {"signup_date": "2023-12-25", "state": " WI ", "currency": "GYD", "department": "Legal & Compliance", "is_active": "F", "notes2": ""}, {"signup_date": "5 Nov 2023", "state": "OH", "currency": "FKP", "department": "IT", "is_active": "F", "notes2": ""}, {"signup_date": "45044", "state": "OHIO  ", "currency": "FKP", "department": "I.T.", "is_active": "false", "notes2": ""}, {"signup_date": "45268", "state": " OH ", "currency": "FKP", "department": " I.T. ", "is_active": "T", "notes2": ""}, {"signup_date": "2023-07-05", "state": "alaska", "currency": "guyana dollar", "department": " Information Technology", "is_active": "false", "notes2": ""}, {"signup_date": "5/10/2023", "state": "um", "currency": "fkp", "department": "Information Technology", "is_active": "false", "notes2": ""}, {"signup_date": "8/13/2023", "state": "wisconsin", "currency": "guyana dollar", "department": "IT", "is_active": "0", "notes2": ""}, {"signup_date": "2 Mar 2023", "state": " wi ", "currency": "GYD", "department": "LEGAL & COMPLIANCE", "is_active": "TRUE", "notes2": ""}, {"signup_date": "2023-06-21", "state": "OH", "currency": " gyd ", "department": "Engineering", "is_active": "TRUE", "notes2": ""}, {"signup_date": "2023-04-07", "state": "MI  ", "currency": "  gyd", "department": "Dev", "is_active": "Yes", "notes2": ""}, {"signup_date": "5/10/2023", "state": "um", "currency": "fkp", "department": "Information Technology", "is_active": "false", "notes2": ""}], "clean": [{"signup_date": "2023-05-10", "state": "United States Minor Outlying Islands", "currency": "FKP", "department": "Legal", "is_active": true}, {"signup_date": "2023-08-20", "state": "Alaska", "currency": "FKP", "department": "Information Technology", "is_active": true}, {"signup_date": "2023-10-06", "state": "Ohio", "currency": "GYD", "department": "Legal", "is_active": false}, {"signup_date": "2023-07-11", "state": "United States Minor Outlying Islands", "currency": "GYD", "department": "Engineering", "is_active": true}, {"signup_date": "2023-05-12", "state": "Michigan", "currency": "FKP", "department": "Legal", "is_active": false}, {"signup_date": "2023-12-24", "state": "Michigan", "currency": "FKP", "department": "Information Technology", "is_active": true}, {"signup_date": "2023-07-20", "state": "United States Minor Outlying Islands", "currency": "GYD", "department": "Engineering", "is_active": true}, {"signup_date": "2023-08-22", "state": "United States Minor Outlying Islands", "currency": "GYD", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-06-22", "state": "Wisconsin", "currency": "GYD", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-10-02", "state": "United States Minor Outlying Islands", "currency": "FKP", "department": "Legal", "is_active": false}, {"signup_date": "2023-12-25", "state": "Wisconsin", "currency": "GYD", "department": "Legal", "is_active": false}, {"signup_date": "2023-11-05", "state": "Ohio", "currency": "FKP", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-04-28", "state": "Ohio", "currency": "FKP", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-12-08", "state": "Ohio", "currency": "FKP", "department": "Information Technology", "is_active": true}, {"signup_date": "2023-07-05", "state": "Alaska", "currency": "GYD", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-05-10", "state": "United States Minor Outlying Islands", "currency": "FKP", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-08-13", "state": "Wisconsin", "currency": "GYD", "department": "Information Technology", "is_active": false}, {"signup_date": "2023-03-02", "state": "Wisconsin", "currency": "GYD", "department": "Legal", "is_active": true}, {"signup_date": "2023-06-21", "state": "Ohio", "currency": "GYD", "department": "Engineering", "is_active": true}, {"signup_date": "2023-04-07", "state": "Michigan", "currency": "GYD", "department": "Engineering", "is_active": true}], "dirty_cols": ["signup_date", "state", "currency", "department", "is_active", "notes2"], "clean_cols": ["signup_date", "state", "currency", "department", "is_active"], "plan": {"dataset_summary": "21 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"alaska": "Alaska", "Mihigan": "Michigan", "MI": "Michigan", "UM": "United States Minor Outlying Islands", "um": "United States Minor Outlying Islands", "WI": "Wisconsin", "OH": "Ohio", "OHIO": "Ohio", "wisconsin": "Wisconsin", "wi": "Wisconsin"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"falkland islands pound": "FKP", "Guyana Dollar": "GYD", "FALKLAND ISLANDS POUND": "FKP", "guyana dollar": "GYD", "fkp": "FKP", "gyd": "GYD"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "Eng": "Engineering", "Legal & Compliance": "Legal", "Dev": "Engineering", "it": "Information Technology", "legal": "Legal", "I.T.": "Information Technology", "LEGAL & COMPLIANCE": "Legal"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"industry": "Education", "status": "   med", "email": "  FNAMOJD@CORP.IO", "department": "Customer Support", "unit": "l", "extra": ""}, {"industry": "realestate", "status": "Low", "email": "ogfj@example.com", "department": "People", "unit": " INCHES ", "extra": ""}, {"industry": "PropTech", "status": "medium", "email": "  kfibm@test.org", "department": "HR", "unit": "kilometers", "extra": ""}, {"industry": "Industrial", "status": " low ", "email": "bpmpc@test.org", "department": "CS", "unit": "liters", "extra": ""}, {"industry": "Manufacturing", "status": "  M", "email": "  lnamnme@corp.io", "department": "sales   ", "unit": "kliometers", "extra": ""}, {"industry": "Industgial", "status": "m", "email": "FPDGGA@TEST.ORG", "department": " People", "unit": "ML", "extra": ""}, {"industry": "RealEstate", "status": "L", "email": "iccklc@mail.com", "department": "Human Resources  ", "unit": "MILLILITER", "extra": ""}, {"industry": "Manufacturing", "status": "P3", "email": "fboncdp@test.org", "department": "Sales", "unit": "inches", "extra": ""}, {"industry": "education", "status": "p3", "email": "JBPM@MAIL.COM ", "department": "HR", "unit": "l", "extra": ""}, {"industry": "Manufacturing", "status": "p1", "email": "MEKMB@TEST.ORG", "department": "Sales", "unit": "LITER", "extra": ""}, {"industry": "Mfg", "status": "medium", "email": "EMOKAC@MAIL.COM", "department": "   Human Resources", "unit": "KM  ", "extra": ""}, {"industry": " Property ", "status": "P1", "email": "FJCFJ@TEST.ORG  ", "department": "Human Resources", "unit": "l", "extra": ""}, {"industry": "Real Estate", "status": "MEDIUM", "email": "NALKO@EXAMPLE.COM   ", "department": "PEOPLE", "unit": "L", "extra": ""}, {"industry": "education", "status": "L", "email": "ednee@test.org", "department": "sales", "unit": "kilomter", "extra": ""}, {"industry": "industrial  ", "status": "LOW", "email": "BIMADHG@EXAMPLE.COM", "department": "human resources", "unit": "MILLILITER  ", "extra": ""}, {"industry": "Education", "status": "  Low", "email": "ADMN@MAIL.COM", "department": "Support", "unit": " Km ", "extra": ""}, {"industry": "education", "status": " low ", "email": "KDHE@EXAMPLE.COM", "department": "CS  ", "unit": " L ", "extra": ""}, {"industry": "PropTech", "status": "P3", "email": "ABNCEG@TEST.ORG", "department": "Saels", "unit": "KILOMETERS  ", "extra": ""}, {"industry": "Real Estate", "status": " low ", "email": "jmhk@example.com", "department": "Sales", "unit": "km", "extra": ""}, {"industry": "real estate", "status": "low", "email": "LGBN@EXAMPLE.COM", "department": "SALES", "unit": "ml", "extra": ""}], "clean": [{"industry": "Education", "status": "Medium", "email": "fnamojd@corp.io", "department": "Customer Support", "unit": "L"}, {"industry": "Real Estate", "status": "Low", "email": "ogfj@example.com", "department": "Human Resources", "unit": "in"}, {"industry": "Real Estate", "status": "Medium", "email": "kfibm@test.org", "department": "Human Resources", "unit": "km"}, {"industry": "Manufacturing", "status": "Low", "email": "bpmpc@test.org", "department": "Customer Support", "unit": "L"}, {"industry": "Manufacturing", "status": "Medium", "email": "lnamnme@corp.io", "department": "Sales", "unit": "km"}, {"industry": "Manufacturing", "status": "Medium", "email": "fpdgga@test.org", "department": "Human Resources", "unit": "mL"}, {"industry": "Real Estate", "status": "Low", "email": "iccklc@mail.com", "department": "Human Resources", "unit": "mL"}, {"industry": "Manufacturing", "status": "Low", "email": "fboncdp@test.org", "department": "Sales", "unit": "in"}, {"industry": "Education", "status": "Low", "email": "jbpm@mail.com", "department": "Human Resources", "unit": "L"}, {"industry": "Manufacturing", "status": "High", "email": "mekmb@test.org", "department": "Sales", "unit": "L"}, {"industry": "Manufacturing", "status": "Medium", "email": "emokac@mail.com", "department": "Human Resources", "unit": "km"}, {"industry": "Real Estate", "status": "High", "email": "fjcfj@test.org", "department": "Human Resources", "unit": "L"}, {"industry": "Real Estate", "status": "Medium", "email": "nalko@example.com", "department": "Human Resources", "unit": "L"}, {"industry": "Education", "status": "Low", "email": "ednee@test.org", "department": "Sales", "unit": "km"}, {"industry": "Manufacturing", "status": "Low", "email": "bimadhg@example.com", "department": "Human Resources", "unit": "mL"}, {"industry": "Education", "status": "Low", "email": "admn@mail.com", "department": "Customer Support", "unit": "km"}, {"industry": "Education", "status": "Low", "email": "kdhe@example.com", "department": "Customer Support", "unit": "L"}, {"industry": "Real Estate", "status": "Low", "email": "abnceg@test.org", "department": "Sales", "unit": "km"}, {"industry": "Real Estate", "status": "Low", "email": "jmhk@example.com", "department": "Sales", "unit": "km"}, {"industry": "Real Estate", "status": "Low", "email": "lgbn@example.com", "department": "Sales", "unit": "mL"}], "dirty_cols": ["industry", "status", "email", "department", "unit", "extra"], "clean_cols": ["industry", "status", "email", "department", "unit"], "plan": {"dataset_summary": "20 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"realestate": "Real Estate", "PropTech": "Real Estate", "Industrial": "Manufacturing", "Industgial": "Manufacturing", "RealEstate": "Real Estate", "education": "Education", "Mfg": "Manufacturing", "Property": "Real Estate", "industrial": "Manufacturing", "real estate": "Real Estate"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"med": "Medium", "medium": "Medium", "low": "Low", "M": "Medium", "m": "Medium", "L": "Low", "P3": "Low", "p3": "Low", "p1": "High", "P1": "High", "MEDIUM": "Medium", "LOW": "Low"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"People": "Human Resources", "HR": "Human Resources", "CS": "Customer Support", "sales": "Sales", "PEOPLE": "Human Resources", "human resources": "Human Resources", "Support": "Customer Support", "Saels": "Sales", "SALES": "Sales"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"l": "L", "INCHES": "in", "kilometers": "km", "liters": "L", "kliometers": "km", "ML": "mL", "MILLILITER": "mL", "inches": "in", "LITER": "L", "KM": "km", "kilomter": "km", "Km": "km", "KILOMETERS": "km", "ml": "mL"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "7/4/2023", "department": "Engineering", "phone": "5820845403", "notes2": ""}, {"signup_date": "3/16/2023", "department": "HR", "phone": "534.577.9458", "notes2": ""}, {"signup_date": "2023-03-07", "department": "R&D", "phone": "506.095.2088", "notes2": ""}, {"signup_date": "2023-02-10", "department": "engineering", "phone": "525.055.2380", "notes2": ""}, {"signup_date": "6/8/2023", "department": "CS", "phone": "576.496.6860", "notes2": ""}, {"signup_date": "11/25/2023", "department": "   HR", "phone": "5194156433", "notes2": ""}, {"signup_date": "45196", "department": "  R&D", "phone": "599.818.4727", "notes2": ""}, {"signup_date": "2023-11-15", "department": "peopel", "phone": "(523)610-2140", "notes2": ""}, {"signup_date": "45175", "department": "cs", "phone": "539-357-6196", "notes2": ""}, {"signup_date": "44966", "department": "PEOPLE", "phone": "558.726.9366", "notes2": ""}, {"signup_date": "45266", "department": " Pople ", "phone": "575.862.1164", "notes2": ""}, {"signup_date": "12 Jun 2023", "department": "R&D", "phone": "585-618-1557", "notes2": ""}, {"signup_date": "2023-08-09", "department": "LEGAL", "phone": "5016973156", "notes2": ""}, {"signup_date": "45129", "department": "Human Resourcse", "phone": "(537)573-5452", "notes2": ""}, {"signup_date": "2023-06-13", "department": "HR", "phone": "556.502.9521", "notes2": ""}, {"signup_date": "7/18/2023", "department": "legal & compliance", "phone": "597-739-6001", "notes2": ""}, {"signup_date": "10 Jul 2023", "department": "Dev", "phone": "(517)925-6819", "notes2": ""}, {"signup_date": "2/9/2023", "department": "Engineering", "phone": "5196282033", "notes2": ""}], "clean": [{"signup_date": "2023-07-04", "department": "Engineering", "phone": "(582) 084-5403"}, {"signup_date": "2023-03-16", "department": "Human Resources", "phone": "(534) 577-9458"}, {"signup_date": "2023-03-07", "department": "Engineering", "phone": "(506) 095-2088"}, {"signup_date": "2023-02-10", "department": "Engineering", "phone": "(525) 055-2380"}, {"signup_date": "2023-06-08", "department": "Customer Support", "phone": "(576) 496-6860"}, {"signup_date": "2023-11-25", "department": "Human Resources", "phone": "(519) 415-6433"}, {"signup_date": "2023-09-27", "department": "Engineering", "phone": "(599) 818-4727"}, {"signup_date": "2023-11-15", "department": "Human Resources", "phone": "(523) 610-2140"}, {"signup_date": "2023-09-06", "department": "Customer Support", "phone": "(539) 357-6196"}, {"signup_date": "2023-02-09", "department": "Human Resources", "phone": "(558) 726-9366"}, {"signup_date": "2023-12-06", "department": "Human Resources", "phone": "(575) 862-1164"}, {"signup_date": "2023-06-12", "department": "Engineering", "phone": "(585) 618-1557"}, {"signup_date": "2023-08-09", "department": "Legal", "phone": "(501) 697-3156"}, {"signup_date": "2023-07-22", "department": "Human Resources", "phone": "(537) 573-5452"}, {"signup_date": "2023-06-13", "department": "Human Resources", "phone": "(556) 502-9521"}, {"signup_date": "2023-07-18", "department": "Legal", "phone": "(597) 739-6001"}, {"signup_date": "2023-07-10", "department": "Engineering", "phone": "(517) 925-6819"}, {"signup_date": "2023-02-09", "department": "Engineering", "phone": "(519) 628-2033"}], "dirty_cols": ["signup_date", "department", "phone", "notes2"], "clean_cols": ["signup_date", "department", "phone"], "plan": {"dataset_summary": "18 rows × 4 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HR": "Human Resources", "R&D": "Engineering", "engineering": "Engineering", "CS": "Customer Support", "peopel": "Human Resources", "cs": "Customer Support", "PEOPLE": "Human Resources", "Pople": "Human Resources", "LEGAL": "Legal", "Human Resourcse": "Human Resources", "legal & compliance": "Legal", "Dev": "Engineering"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"industry": "retail & e-cmomerce", "department": "operations", "status": "L", "is_active": "0", "amount": "$4,852.88", "extra": ""}, {"industry": "Health Care", "department": "Mktg", "status": "medium", "is_active": "true", "amount": "7.606,03", "extra": ""}, {"industry": "medical", "department": "Finance   ", "status": "  p1", "is_active": "T", "amount": "$149.84", "extra": ""}, {"industry": "Telceommunications", "department": "Ops", "status": "medum", "is_active": "No", "amount": "8.584,92", "extra": ""}, {"industry": "healthcare", "department": "Operations", "status": "L", "is_active": "Y", "amount": "7.750,57", "extra": ""}, {"industry": "Telecom", "department": "Marketing", "status": "High", "is_active": "FALSE", "amount": "8.164,23", "extra": ""}, {"industry": "HEALTHCARE", "department": "Growth", "status": "HIGH", "is_active": "No", "amount": "2.438,38", "extra": ""}, {"industry": "Telecommunications", "department": "FIN", "status": "LOW", "is_active": "T", "amount": "$5,757.96", "extra": ""}, {"industry": "Comms", "department": "Finnce  ", "status": "l ", "is_active": "F", "amount": "$3,874.95", "extra": ""}, {"industry": " ed", "department": "Finance ", "status": " low ", "is_active": "T", "amount": "$4,505.46", "extra": ""}, {"industry": "Comms", "department": "Operations", "status": "P2", "is_active": "N", "amount": "8.644,93", "extra": ""}, {"industry": "RETAIL", "department": "FINANCE", "status": "H", "is_active": "TRUE", "amount": "$8,686.43", "extra": ""}, {"industry": "EdTech", "department": " MKTG ", "status": "low", "is_active": "1", "amount": "-", "extra": ""}, {"industry": "education", "department": "Operations", "status": "P2", "is_active": "Yes", "amount": "2.272,31", "extra": ""}, {"industry": "medical", "department": "accounting", "status": "L", "is_active": "TRUE", "amount": "null", "extra": ""}, {"industry": " education ", "department": "Fin  ", "status": "  medium", "is_active": "FALSE", "amount": "TBD", "extra": ""}, {"industry": "Telecommunications", "department": "   operations", "status": "p3", "is_active": "Y", "amount": "$5,833.54", "extra": ""}, {"industry": "retail", "department": "accounting", "status": "   P2", "is_active": "0", "amount": "1.394,12", "extra": ""}, {"industry": "Retail & E-commerce", "department": "   Mkt", "status": "M ", "is_active": "Yes", "amount": "$4,865.79", "extra": ""}, {"industry": "retail   ", "department": " Operations ", "status": "medium  ", "is_active": "1", "amount": "2.978,46", "extra": ""}, {"industry": "Telceommunications", "department": "Ops", "status": "medum", "is_active": "No", "amount": "8.584,92", "extra": ""}, {"industry": "", "department": "", "status": "", "is_active": "", "amount": "", "extra": ""}], "clean": [{"industry": "Retail", "department": "Operations", "status": "Low", "is_active": false, "amount": 4852.88}, {"industry": "Healthcare", "department": "Marketing", "status": "Medium", "is_active": true, "amount": 7606.03}, {"industry": "Healthcare", "department": "Finance", "status": "High", "is_active": true, "amount": 149.84}, {"industry": "Telecommunications", "department": "Operations", "status": "Medium", "is_active": false, "amount": 8584.92}, {"industry": "Healthcare", "department": "Operations", "status": "Low", "is_active": true, "amount": 7750.57}, {"industry": "Telecommunications", "department": "Marketing", "status": "High", "is_active": false, "amount": 8164.23}, {"industry": "Healthcare", "department": "Marketing", "status": "High", "is_active": false, "amount": 2438.38}, {"industry": "Telecommunications", "department": "Finance", "status": "Low", "is_active": true, "amount": 5757.96}, {"industry": "Telecommunications", "department": "Finance", "status": "Low", "is_active": false, "amount": 3874.95}, {"industry": "Education", "department": "Finance", "status": "Low", "is_active": true, "amount": 4505.46}, {"industry": "Telecommunications", "department": "Operations", "status": "Medium", "is_active": false, "amount": 8644.93}, {"industry": "Retail", "department": "Finance", "status": "High", "is_active": true, "amount": 8686.43}, {"industry": "Education", "department": "Marketing", "status": "Low", "is_active": true, "amount": NaN}, {"industry": "Education", "department": "Operations", "status": "Medium", "is_active": true, "amount": 2272.31}, {"industry": "Healthcare", "department": "Finance", "status": "Low", "is_active": true, "amount": NaN}, {"industry": "Education", "department": "Finance", "status": "Medium", "is_active": false, "amount": NaN}, {"industry": "Telecommunications", "department": "Operations", "status": "Low", "is_active": true, "amount": 5833.54}, {"industry": "Retail", "department": "Finance", "status": "Medium", "is_active": false, "amount": 1394.12}, {"industry": "Retail", "department": "Marketing", "status": "Medium", "is_active": true, "amount": 4865.79}, {"industry": "Retail", "department": "Operations", "status": "Medium", "is_active": true, "amount": 2978.46}], "dirty_cols": ["industry", "department", "status", "is_active", "amount", "extra"], "clean_cols": ["industry", "department", "status", "is_active", "amount"], "plan": {"dataset_summary": "22 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"retail & e-cmomerce": "Retail", "Health Care": "Healthcare", "medical": "Healthcare", "Telceommunications": "Telecommunications", "healthcare": "Healthcare", "Telecom": "Telecommunications", "HEALTHCARE": "Healthcare", "Comms": "Telecommunications", "ed": "Education", "RETAIL": "Retail", "EdTech": "Education", "education": "Education", "retail": "Retail", "Retail & E-commerce": "Retail"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"operations": "Operations", "Mktg": "Marketing", "Ops": "Operations", "Growth": "Marketing", "FIN": "Finance", "Finnce": "Finance", "FINANCE": "Finance", "MKTG": "Marketing", "accounting": "Finance", "Fin": "Finance", "Mkt": "Marketing"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"L": "Low", "medium": "Medium", "p1": "High", "medum": "Medium", "HIGH": "High", "LOW": "Low", "l": "Low", "low": "Low", "P2": "Medium", "H": "High", "p3": "Low", "M": "Medium"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": []}}
{"dirty": [{"city": "kasungu", "country": "Guadeloupe", "state": "OR", "phone": "566.591.5650", "status": "med ", "name": "Carol Johnson", "job_title": "Rep", "company": " Cyberdyne "}, {"city": "kasungu", "country": "Guadelloupe", "state": "  MICHIGAN", "phone": "(577)187-3592", "status": "   med", "name": " Eve Park", "job_title": "  AE", "company": "Hooli"}, {"city": "  Kasungu", "country": "Cameroon", "state": "MI", "phone": "580.342.2349", "status": "Hgh", "name": " Eve Fischer", "job_title": "Rep", "company": "Globex"}, {"city": "bukit tambun", "country": "PM", "state": "maryland", "phone": "589.609.7610", "status": "HIGH", "name": "Frank Novak   ", "job_title": "Senior Engineer", "company": " Hooli "}, {"city": "Bukix Tambun", "country": "cameroon", "state": "MICHIGAN", "phone": "579-461-4632", "status": "P3", "name": "Bob Moore", "job_title": " Sales Rep", "company": "Cyberdyne"}, {"city": "Bukit Tambun", "country": "Saint Pierre And Miquelon", "state": "md", "phone": "5083005906", "status": "MEDIUM", "name": "  Lena Adams", "job_title": "engineering mgr", "company": " Vehement "}, {"city": " Kasungu ", "country": "cmr", "state": "Oregon", "phone": "5915566268", "status": "L", "name": "Karl Fischer   ", "job_title": "AE", "company": "Wonka Co"}, {"city": "KASUNGU", "country": "Guadeloupe", "state": "Md", "phone": "588.640.6868", "status": "med", "name": "   David Lee", "job_title": "Sales Rep", "company": "Initech"}, {"city": "Kasungu", "country": "me", "state": "oregon", "phone": "(570)526-2603", "status": "  High", "name": "Mona  Lee", "job_title": "acct exec", "company": "Stark Industries"}, {"city": "kasungu", "country": "republic of cameroon", "state": "MD", "phone": "549-135-5299", "status": "P3", "name": "Mona Smith", "job_title": "SALES RUP", "company": "Soylent Corp   "}, {"city": "kasungu", "country": "GP", "state": "   michigan", "phone": "587-973-2028", "status": " high", "name": "Bob Adams", "job_title": "eng manager", "company": "Initech "}, {"city": "Kasungu   ", "country": "Gp", "state": "Maryland", "phone": "(538)985-8552", "status": "P3", "name": " Ivan Novak", "job_title": "Engineering Mgr", "company": "Soylent  Corp"}, {"city": "Kasngu", "country": "GP", "state": "MARYLAND", "phone": "5079422936", "status": "MEIUM", "name": "  Omar Lee", "job_title": "Engineering  Manager", "company": "Soylent  Corp"}, {"city": "bukit tambun", "country": "  Cameoon", "state": "Michigan", "phone": "565-200-1136", "status": " med ", "name": "Karl Park", "job_title": "sr. engineer", "company": "Wonka Co"}, {"city": "KASUNGU", "country": "me", "state": "MI", "phone": "(585)364-4961", "status": "med", "name": "Lena  Moore", "job_title": "ACCOUNT EXECUTIVE", "company": "  Hooli"}, {"city": "Bukit Tambun", "country": "Me", "state": "oregon", "phone": "(535)635-8115", "status": "  High", "name": "Sara Wong", "job_title": "Sales Rep", "company": "  Initech"}, {"city": "Bukit Tambun", "country": "ME", "state": "MI", "phone": "5196297210", "status": "h", "name": "Omar Reyes   ", "job_title": "Engineering Manager ", "company": "Stark Industries"}, {"city": "Kasngu", "country": "   SAINT PIERRE AND MIQUELON", "state": " maryland", "phone": "592-041-6484", "status": "H", "name": "Sara Cruz", "job_title": "ENG MANAGER", "company": "Globex"}, {"city": "Kasungu  ", "country": "CMR", "state": "Michigan", "phone": "(596)772-3337", "status": " p1", "name": "David Khan", "job_title": "eng mgr", "company": "Cyberdyne"}, {"city": "Bukit  Tambun", "country": " CM", "state": "Or", "phone": "5154461943", "status": " High ", "name": "Frank Smith  ", "job_title": "Rep", "company": "   Wonka Co"}, {"city": "BUKIT TAMBUN", "country": "spm", "state": "Maryland", "phone": "5401735881", "status": "medium", "name": "Ivan Petrov   ", "job_title": "sales rep", "company": "Stark Industries "}, {"city": "Kasungu", "country": "CMR", "state": "Maryland", "phone": "542.958.4418", "status": "H", "name": "Lena Cruz", "job_title": "   Sales Rep", "company": " Umbrella"}, {"city": "Kasungu", "country": "spm", "state": "Michigan", "phone": "513-709-6893", "status": "High ", "name": "Lena Lee", "job_title": "engineering mgr", "company": "Wonka Co"}, {"city": "   Kasungu", "country": "CM", "state": "Oregon", "phone": "5831967356", "status": "hgih", "name": "  Omar Moore", "job_title": "Sales Representative", "company": "Soylent Corp"}, {"city": " Kasungu ", "country": " PM ", "state": "   Michigan", "phone": "582-333-6768", "status": "P1", "name": "Priya Lee ", "job_title": "senior engineer", "company": "  Hooli"}, {"city": "perugia  ", "country": "MNE", "state": "  Michigan", "phone": "582.009.0969", "status": "P2", "name": "Frank Moore", "job_title": "Rep", "company": " Hooli "}, {"city": "Kasungu", "country": " cameroon ", "state": "MD", "phone": "568-136-3971", "status": "high", "name": "Lena Smith", "job_title": "sales representative", "company": "Hooli"}, {"city": "bukit tambun", "country": "ME", "state": " MI ", "phone": "587-765-1860", "status": "HIGH", "name": "Priya Reyes  ", "job_title": "Eng Manager", "company": "Wonka Co"}, {"city": "KASUNGU", "country": "GLP", "state": "Maryland", "phone": "5522057685", "status": "   l", "name": "Judy  Reyes", "job_title": "   Senior Eng", "company": "Soylent  Corp"}, {"city": "perugia", "country": "mne", "state": "or", "phone": "5456296320", "status": "P3", "name": "Heidi Johnson  ", "job_title": "senior engineer", "company": "Soylent Corp   "}, {"city": "ksungu", "country": "ME", "state": "OR", "phone": "5506806261", "status": "H ", "name": "Frank Diaz", "job_title": " SENIOR ENGINEER", "company": "Cyberdyne"}, {"city": "Kasungu", "country": "CMR", "state": "MI", "phone": "548-118-4214", "status": "low", "name": "Lena Park", "job_title": "acct exec ", "company": "Acme Inc"}, {"city": "Kasungu", "country": "glp", "state": "Md", "phone": "588.825.8802", "status": "  med", "name": "Eve  Khan", "job_title": "Acct Exec", "company": "Hooli"}, {"city": "Kasungu", "country": "  CMR", "state": " or", "phone": "578-787-0033", "status": "high", "name": " Omar Diaz", "job_title": "Acct Exec", "company": " Globex "}, {"city": "PERUGIA", "country": "Glp", "state": "Oregon", "phone": "(511)377-5788", "status": "LOW", "name": "Judy Novak  ", "job_title": "eng  manager", "company": "Vehement"}, {"city": "kasunpu", "country": "GLP", "state": "md", "phone": "5282126139", "status": "l", "name": "Heidi Fischer", "job_title": "Senior Eng", "company": "Cyberdyne"}, {"city": "Bukit Tambun", "country": "PM", "state": "MICHIGAN", "phone": "(564)892-9932", "status": "low", "name": " David Fischer", "job_title": "Sales Rep", "company": "   Globex"}, {"city": "pecugia", "country": "MNE", "state": "md", "phone": "537.341.6964", "status": "medium ", "name": "Heidi Smith", "job_title": "Engineering Mgr", "company": "Cyberdyne"}, {"city": "Kasungu", "country": "cm", "state": "Michigan", "phone": "543.838.1562", "status": " L ", "name": "Sara Diaz", "job_title": "account executive", "company": "Stark  Industries"}, {"city": "BUKIT TAMBUN", "country": "CMR", "state": " oregon", "phone": "510-143-2449", "status": "  P3", "name": "Grace Smith", "job_title": "Eng  Mgr", "company": "Umbrella"}, {"city": "   Perugia", "country": "me", "state": "   or", "phone": "566-656-8090", "status": "P1", "name": "Eve  Adams", "job_title": "Acount Executive", "company": "Soylent Corp"}, {"city": "Perugia ", "country": "Saint Pierre and  Miquelon", "state": "Michigan", "phone": "592-151-2107", "status": "High", "name": "David  Cruz", "job_title": "Senior Eng ", "company": "Acme Inc  "}, {"city": "Kasungu", "country": "CMR", "state": "MI", "phone": "548-118-4214", "status": "low", "name": "Lena Park", "job_title": "acct exec ", "company": "Acme Inc"}, {"city": "", "country": "", "state": "", "phone": "", "status": "", "name": "", "job_title": "", "company": ""}], "clean": [{"city": "Kasungu", "country": "Guadeloupe", "state": "Oregon", "phone": "(566) 591-5650", "status": "Medium", "name": "Carol Johnson", "job_title": "Sales Representative", "company": "Cyberdyne"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Michigan", "phone": "(577) 187-3592", "status": "Medium", "name": "Eve Park", "job_title": "Account Executive", "company": "Hooli"}, {"city": "Kasungu", "country": "Cameroon", "state": "Michigan", "phone": "(580) 342-2349", "status": "High", "name": "Eve Fischer", "job_title": "Sales Representative", "company": "Globex"}, {"city": "Bukit Tambun", "country": "Saint Pierre and Miquelon", "state": "Maryland", "phone": "(589) 609-7610", "status": "High", "name": "Frank Novak", "job_title": "Senior Engineer", "company": "Hooli"}, {"city": "Bukit Tambun", "country": "Cameroon", "state": "Michigan", "phone": "(579) 461-4632", "status": "Low", "name": "Bob Moore", "job_title": "Sales Representative", "company": "Cyberdyne"}, {"city": "Bukit Tambun", "country": "Saint Pierre and Miquelon", "state": "Maryland", "phone": "(508) 300-5906", "status": "Medium", "name": "Lena Adams", "job_title": "Engineering Manager", "company": "Vehement"}, {"city": "Kasungu", "country": "Cameroon", "state": "Oregon", "phone": "(591) 556-6268", "status": "Low", "name": "Karl Fischer", "job_title": "Account Executive", "company": "Wonka Co"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Maryland", "phone": "(588) 640-6868", "status": "Medium", "name": "David Lee", "job_title": "Sales Representative", "company": "Initech"}, {"city": "Kasungu", "country": "Montenegro", "state": "Oregon", "phone": "(570) 526-2603", "status": "High", "name": "Mona Lee", "job_title": "Account Executive", "company": "Stark Industries"}, {"city": "Kasungu", "country": "Cameroon", "state": "Maryland", "phone": "(549) 135-5299", "status": "Low", "name": "Mona Smith", "job_title": "Sales Representative", "company": "Soylent Corp"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Michigan", "phone": "(587) 973-2028", "status": "High", "name": "Bob Adams", "job_title": "Engineering Manager", "company": "Initech"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Maryland", "phone": "(538) 985-8552", "status": "Low", "name": "Ivan Novak", "job_title": "Engineering Manager", "company": "Soylent Corp"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Maryland", "phone": "(507) 942-2936", "status": "Medium", "name": "Omar Lee", "job_title": "Engineering Manager", "company": "Soylent Corp"}, {"city": "Bukit Tambun", "country": "Cameroon", "state": "Michigan", "phone": "(565) 200-1136", "status": "Medium", "name": "Karl Park", "job_title": "Senior Engineer", "company": "Wonka Co"}, {"city": "Kasungu", "country": "Montenegro", "state": "Michigan", "phone": "(585) 364-4961", "status": "Medium", "name": "Lena Moore", "job_title": "Account Executive", "company": "Hooli"}, {"city": "Bukit Tambun", "country": "Montenegro", "state": "Oregon", "phone": "(535) 635-8115", "status": "High", "name": "Sara Wong", "job_title": "Sales Representative", "company": "Initech"}, {"city": "Bukit Tambun", "country": "Montenegro", "state": "Michigan", "phone": "(519) 629-7210", "status": "High", "name": "Omar Reyes", "job_title": "Engineering Manager", "company": "Stark Industries"}, {"city": "Kasungu", "country": "Saint Pierre and Miquelon", "state": "Maryland", "phone": "(592) 041-6484", "status": "High", "name": "Sara Cruz", "job_title": "Engineering Manager", "company": "Globex"}, {"city": "Kasungu", "country": "Cameroon", "state": "Michigan", "phone": "(596) 772-3337", "status": "High", "name": "David Khan", "job_title": "Engineering Manager", "company": "Cyberdyne"}, {"city": "Bukit Tambun", "country": "Cameroon", "state": "Oregon", "phone": "(515) 446-1943", "status": "High", "name": "Frank Smith", "job_title": "Sales Representative", "company": "Wonka Co"}, {"city": "Bukit Tambun", "country": "Saint Pierre and Miquelon", "state": "Maryland", "phone": "(540) 173-5881", "status": "Medium", "name": "Ivan Petrov", "job_title": "Sales Representative", "company": "Stark Industries"}, {"city": "Kasungu", "country": "Cameroon", "state": "Maryland", "phone": "(542) 958-4418", "status": "High", "name": "Lena Cruz", "job_title": "Sales Representative", "company": "Umbrella"}, {"city": "Kasungu", "country": "Saint Pierre and Miquelon", "state": "Michigan", "phone": "(513) 709-6893", "status": "High", "name": "Lena Lee", "job_title": "Engineering Manager", "company": "Wonka Co"}, {"city": "Kasungu", "country": "Cameroon", "state": "Oregon", "phone": "(583) 196-7356", "status": "High", "name": "Omar Moore", "job_title": "Sales Representative", "company": "Soylent Corp"}, {"city": "Kasungu", "country": "Saint Pierre and Miquelon", "state": "Michigan", "phone": "(582) 333-6768", "status": "High", "name": "Priya Lee", "job_title": "Senior Engineer", "company": "Hooli"}, {"city": "Perugia", "country": "Montenegro", "state": "Michigan", "phone": "(582) 009-0969", "status": "Medium", "name": "Frank Moore", "job_title": "Sales Representative", "company": "Hooli"}, {"city": "Kasungu", "country": "Cameroon", "state": "Maryland", "phone": "(568) 136-3971", "status": "High", "name": "Lena Smith", "job_title": "Sales Representative", "company": "Hooli"}, {"city": "Bukit Tambun", "country": "Montenegro", "state": "Michigan", "phone": "(587) 765-1860", "status": "High", "name": "Priya Reyes", "job_title": "Engineering Manager", "company": "Wonka Co"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Maryland", "phone": "(552) 205-7685", "status": "Low", "name": "Judy Reyes", "job_title": "Senior Engineer", "company": "Soylent Corp"}, {"city": "Perugia", "country": "Montenegro", "state": "Oregon", "phone": "(545) 629-6320", "status": "Low", "name": "Heidi Johnson", "job_title": "Senior Engineer", "company": "Soylent Corp"}, {"city": "Kasungu", "country": "Montenegro", "state": "Oregon", "phone": "(550) 680-6261", "status": "High", "name": "Frank Diaz", "job_title": "Senior Engineer", "company": "Cyberdyne"}, {"city": "Kasungu", "country": "Cameroon", "state": "Michigan", "phone": "(548) 118-4214", "status": "Low", "name": "Lena Park", "job_title": "Account Executive", "company": "Acme Inc"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Maryland", "phone": "(588) 825-8802", "status": "Medium", "name": "Eve Khan", "job_title": "Account Executive", "company": "Hooli"}, {"city": "Kasungu", "country": "Cameroon", "state": "Oregon", "phone": "(578) 787-0033", "status": "High", "name": "Omar Diaz", "job_title": "Account Executive", "company": "Globex"}, {"city": "Perugia", "country": "Guadeloupe", "state": "Oregon", "phone": "(511) 377-5788", "status": "Low", "name": "Judy Novak", "job_title": "Engineering Manager", "company": "Vehement"}, {"city": "Kasungu", "country": "Guadeloupe", "state": "Maryland", "phone": "(528) 212-6139", "status": "Low", "name": "Heidi Fischer", "job_title": "Senior Engineer", "company": "Cyberdyne"}, {"city": "Bukit Tambun", "country": "Saint Pierre and Miquelon", "state": "Michigan", "phone": "(564) 892-9932", "status": "Low", "name": "David Fischer", "job_title": "Sales Representative", "company": "Globex"}, {"city": "Perugia", "country": "Montenegro", "state": "Maryland", "phone": "(537) 341-6964", "status": "Medium", "name": "Heidi Smith", "job_title": "Engineering Manager", "company": "Cyberdyne"}, {"city": "Kasungu", "country": "Cameroon", "state": "Michigan", "phone": "(543) 838-1562", "status": "Low", "name": "Sara Diaz", "job_title": "Account Executive", "company": "Stark Industries"}, {"city": "Bukit Tambun", "country": "Cameroon", "state": "Oregon", "phone": "(510) 143-2449", "status": "Low", "name": "Grace Smith", "job_title": "Engineering Manager", "company": "Umbrella"}, {"city": "Perugia", "country": "Montenegro", "state": "Oregon", "phone": "(566) 656-8090", "status": "High", "name": "Eve Adams", "job_title": "Account Executive", "company": "Soylent Corp"}, {"city": "Perugia", "country": "Saint Pierre and Miquelon", "state": "Michigan", "phone": "(592) 151-2107", "status": "High", "name": "David Cruz", "job_title": "Senior Engineer", "company": "Acme Inc"}], "dirty_cols": ["city", "country", "state", "phone", "status", "name", "job_title", "company"], "clean_cols": ["city", "country", "state", "phone", "status", "name", "job_title", "company"], "plan": {"dataset_summary": "44 rows × 8 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kasungu": "Kasungu", "bukit tambun": "Bukit Tambun", "Bukix Tambun": "Bukit Tambun", "KASUNGU": "Kasungu", "Kasngu": "Kasungu", "BUKIT TAMBUN": "Bukit Tambun", "perugia": "Perugia", "ksungu": "Kasungu", "PERUGIA": "Perugia", "kasunpu": "Kasungu", "pecugia": "Perugia"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Guadelloupe": "Guadeloupe", "PM": "Saint Pierre and Miquelon", "cameroon": "Cameroon", "Saint Pierre And Miquelon": "Saint Pierre and Miquelon", "cmr": "Cameroon", "me": "Montenegro", "republic of cameroon": "Cameroon", "GP": "Guadeloupe", "Gp": "Guadeloupe", "Cameoon": "Cameroon", "Me": "Montenegro", "ME": "Montenegro", "SAINT PIERRE AND MIQUELON": "Saint Pierre and Miquelon", "CMR": "Cameroon", "CM": "Cameroon", "spm": "Saint Pierre and Miquelon", "MNE": "Montenegro", "GLP": "Guadeloupe", "mne": "Montenegro", "glp": "Guadeloupe", "Glp": "Guadeloupe", "cm": "Cameroon"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OR": "Oregon", "MICHIGAN": "Michigan", "MI": "Michigan", "maryland": "Maryland", "md": "Maryland", "Md": "Maryland", "oregon": "Oregon", "MD": "Maryland", "michigan": "Michigan", "MARYLAND": "Maryland", "Or": "Oregon", "or": "Oregon"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"med": "Medium", "Hgh": "High", "HIGH": "High", "P3": "Low", "MEDIUM": "Medium", "L": "Low", "high": "High", "MEIUM": "Medium", "h": "High", "H": "High", "p1": "High", "medium": "Medium", "hgih": "High", "P1": "High", "P2": "Medium", "l": "Low", "low": "Low", "LOW": "Low"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Rep": "Sales Representative", "AE": "Account Executive", "Sales Rep": "Sales Representative", "engineering mgr": "Engineering Manager", "acct exec": "Account Executive", "SALES RUP": "Sales Representative", "eng manager": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "sr. engineer": "Senior Engineer", "ACCOUNT EXECUTIVE": "Account Executive", "ENG MANAGER": "Engineering Manager", "eng mgr": "Engineering Manager", "sales rep": "Sales Representative", "senior engineer": "Senior Engineer", "sales representative": "Sales Representative", "Eng Manager": "Engineering Manager", "Senior Eng": "Senior Engineer", "SENIOR ENGINEER": "Senior Engineer", "Acct Exec": "Account Executive", "account executive": "Account Executive", "Eng Mgr": "Engineering Manager", "Acount Executive": "Account Executive"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"email": " HEKA@MAIL.COM ", "amount": "$8,291.68", "status": "cancelled", "state": "al", "job_title": "Engineering  Mgr", "company": "Cyberdyne  ", "unit": "miles", "is_active": "F"}, {"email": "GJKCBO@EXAMPLE.COM", "amount": "$4,708.02", "status": "Trial", "state": "utah", "job_title": "Dir", "company": "Acme Inc", "unit": "   Mi", "is_active": "T"}, {"email": " pbigib@mail.com ", "amount": "$5,692.69", "status": "CANCELLED", "state": "ar", "job_title": "eng  manager", "company": "Umbrella", "unit": "MI", "is_active": "false"}, {"email": "  NAEDLFJ@EXAMPLE.COM", "amount": "$633.65", "status": " canceled ", "state": "TX", "job_title": "eng mgr", "company": "Initech", "unit": "kg", "is_active": "F"}, {"email": "ENLGPGA@TEST.ORG  ", "amount": "491,44", "status": "Trail", "state": "Texas", "job_title": "Director  ", "company": "Stark Industries", "unit": "kilogram  ", "is_active": "0"}, {"email": "ODLD@EXAMPLE.COM  ", "amount": "5.483,43", "status": "Churned", "state": "Utaah", "job_title": " director ", "company": "Hooli", "unit": "MILES", "is_active": "Y"}, {"email": " HLCEGPI@CORP.IO", "amount": "8.537,08", "status": "   TKIAL", "state": "tx", "job_title": "CEO", "company": "Umbrella", "unit": "mi", "is_active": "N"}, {"email": " nbfk@example.com ", "amount": "$1,317.20", "status": "cancelled", "state": "NC", "job_title": "   Eng Manager", "company": "Stark Industries", "unit": "Kg", "is_active": "0"}, {"email": "KGHEFE@CORP.IO   ", "amount": "1.513,04", "status": " cancelled ", "state": "AR  ", "job_title": "Eng Mgr", "company": "Wonka  Co", "unit": "kilograms", "is_active": "N"}, {"email": "MPLM@EXAMPLE.COM", "amount": "$7,236.20", "status": "trial  ", "state": "al", "job_title": "Dir.", "company": "Soylent Corp", "unit": "mi", "is_active": "T"}, {"email": "PAFKOHH@MAIL.COM", "amount": "$2,307.99", "status": " tbial ", "state": "  UTTAH", "job_title": "Dir   ", "company": "Globex", "unit": "   kilogram", "is_active": "No"}, {"email": " KNOAIEH@MAIL.COM ", "amount": "#N/A", "status": "trial   ", "state": "Alabava", "job_title": "dir", "company": "   Acme Inc", "unit": "Mi", "is_active": "false"}, {"email": "NNGMAN@TEST.ORG", "amount": "7.512,45", "status": "free trial", "state": "Texas", "job_title": "Ceo", "company": "Acme Inc   ", "unit": "Kg", "is_active": "F"}, {"email": "  jokil@corp.io", "amount": "7.093,13", "status": "TRIAL", "state": "AR", "job_title": "CHIEF EXECUTIVE OFFICER", "company": "Vehement   ", "unit": "kiograms", "is_active": "true"}, {"email": " NKCL@MAIL.COM ", "amount": "8.362,64", "status": "canceled", "state": "al", "job_title": "  Eng Mgr", "company": " Cyberdyne ", "unit": "kiolgram   ", "is_active": "T"}, {"email": "gnmhp@corp.io", "amount": "1.576,12", "status": "CHHURNED", "state": "Al", "job_title": "eng mgr", "company": "Soylent Corp", "unit": "mi ", "is_active": "Y"}, {"email": "PAIBGEK@EXAMPLE.COM", "amount": "$1,957.80", "status": "TRIAL", "state": "TX", "job_title": "   CEO", "company": " Initech ", "unit": "miles  ", "is_active": "Yes"}, {"email": " kbfk@corp.io ", "amount": "735,41", "status": "cancelled", "state": "ar", "job_title": " ENG MGR", "company": "Vehement", "unit": "KG", "is_active": "false"}, {"email": "flllob@test.org", "amount": "4.770,66", "status": "CANCELED", "state": "AR", "job_title": "DIR", "company": "Umbrella   ", "unit": "kilograms", "is_active": "Y"}, {"email": "NOFCNHL@MAIL.COM", "amount": "$6,460.30", "status": "trial", "state": " AL", "job_title": "  CEO", "company": "Acme Inc  ", "unit": " Kg ", "is_active": "FALSE"}, {"email": "pmdak@corp.io", "amount": "$1,858.83", "status": "Cancelled", "state": "Arkansas", "job_title": "Chief Executive Officer", "company": "Soylent  Corp", "unit": "kg", "is_active": "true"}, {"email": " DLHB@MAIL.COM ", "amount": "$5,256.65", "status": " free trial", "state": "Utah", "job_title": "Eng Mgr", "company": "Soylent Corp", "unit": "Mi", "is_active": "1"}, {"email": "BHJOOII@CORP.IO", "amount": "$4,317.53", "status": "Churnned", "state": "NC", "job_title": "   engineering manager", "company": "  Acme Inc", "unit": "Mi", "is_active": "TRUE"}, {"email": "HCDPIOJ@MAIL.COM", "amount": "2.530,54", "status": "trial", "state": "utah", "job_title": "Dir.", "company": "Wonka  Co", "unit": "  kg", "is_active": "No"}, {"email": "plpdi@test.org", "amount": "na", "status": "Thial", "state": "AR", "job_title": " C.E.O. ", "company": "  Hooli", "unit": "mile", "is_active": "1"}, {"email": "  KGHN@MAIL.COM", "amount": "7.384,78", "status": "Trial", "state": " Arkansas ", "job_title": "eng mgr  ", "company": "Acme Inc", "unit": "kg", "is_active": "Yes"}, {"email": "gechhi@example.com", "amount": "3.428,15", "status": "CANCCELED", "state": "   utah", "job_title": "ceo", "company": "Stark Industries ", "unit": "KG", "is_active": "false"}, {"email": "  beeacdg@mail.com", "amount": "$6,535.20", "status": " TRILL ", "state": "Utah", "job_title": "Dir", "company": "Vehement", "unit": " kg ", "is_active": "TRUE"}, {"email": "BIIHG@EXAMPLE.COM", "amount": "7.900,67", "status": "  TRIAL", "state": "ALABAMA", "job_title": "C.E.O.", "company": "Globex", "unit": "KG", "is_active": "false"}, {"email": "hmioaj@corp.io", "amount": "-100", "status": "free  trial", "state": " ut ", "job_title": "Dir.", "company": "Hooli", "unit": "kilograms", "is_active": "Y"}, {"email": " eobim@mail.com", "amount": "8.694,43", "status": "free trial", "state": "AR", "job_title": "director", "company": "Wonka Co", "unit": "miles", "is_active": "1"}, {"email": "pkdcgm@example.com", "amount": "$3,922.44", "status": "Trial", "state": "utah ", "job_title": "Chief  Executive Officer", "company": "Acme Inc", "unit": "miles", "is_active": "No"}, {"email": "DOEINF@TEST.ORG", "amount": "$7,084.78", "status": "free trial", "state": "north craolina", "job_title": "Eng Manager", "company": "Wonka Co", "unit": "kilograms", "is_active": "TRUE"}, {"email": "egoii@example.com", "amount": "6.506,44", "status": "Free Tral", "state": "Alabama", "job_title": "engineering mgr", "company": "Soylent Corp ", "unit": "miles", "is_active": "F"}, {"email": "ACHBB@CORP.IO  ", "amount": "8.327,31", "status": "churnde", "state": "Texas", "job_title": "DIRECTOR", "company": " Cyberdyne ", "unit": "KILOGRAM", "is_active": "N"}, {"email": " BEDGCC@EXAMPLE.COM ", "amount": "$113.82", "status": " CANCELED", "state": "Txeas", "job_title": "engineering mgr", "company": "Cyberdyne", "unit": "   mi", "is_active": "FALSE"}, {"email": " leoimn@test.org", "amount": "$892.14", "status": "Triaal", "state": "north carolina", "job_title": "Chief Executive Officer", "company": "Wonka Co", "unit": "kg", "is_active": "1"}, {"email": "   FDMHCEA@MAIL.COM", "amount": "-", "status": "  TRIAL", "state": "al", "job_title": "Chief  Executive Officer", "company": "Acme Inc   ", "unit": "KG ", "is_active": "Y"}, {"email": "nkgi@example.com", "amount": "4.943,60", "status": "Trial", "state": "North Carolina", "job_title": "ENGINEERING MANAGER", "company": "Soylent Corp  ", "unit": "kg", "is_active": "N"}, {"email": " jcdf@corp.io ", "amount": "$1,347.83", "status": "churned", "state": " tx ", "job_title": "Dir", "company": "Initech", "unit": " kilogram", "is_active": "FALSE"}, {"email": "IIBC@MAIL.COM", "amount": "N/A", "status": "Free Trial", "state": "north carolina", "job_title": "Chief Executive Officer", "company": "Umbrella", "unit": "miles", "is_active": "FALSE"}, {"email": "amdc@example.com   ", "amount": "$3,022.27", "status": "cancelled", "state": "TEXAS", "job_title": "Dir.", "company": "Wonka Co", "unit": "kilogram", "is_active": "Yes"}, {"email": "BIFB@MAIL.COM", "amount": "$2,070.42", "status": "churned", "state": "Ar", "job_title": "C.E.O.", "company": "Initech", "unit": " kg", "is_active": "Yes"}, {"email": " HLCEGPI@CORP.IO", "amount": "8.537,08", "status": "   TKIAL", "state": "tx", "job_title": "CEO", "company": "Umbrella", "unit": "mi", "is_active": "N"}], "clean": [{"email": "heka@mail.com", "amount": 8291.68, "status": "Churned", "state": "Alabama", "job_title": "Engineering Manager", "company": "Cyberdyne", "unit": "mi", "is_active": false}, {"email": "gjkcbo@example.com", "amount": 4708.02, "status": "Trial", "state": "Utah", "job_title": "Director", "company": "Acme Inc", "unit": "mi", "is_active": true}, {"email": "pbigib@mail.com", "amount": 5692.69, "status": "Churned", "state": "Arkansas", "job_title": "Engineering Manager", "company": "Umbrella", "unit": "mi", "is_active": false}, {"email": "naedlfj@example.com", "amount": 633.65, "status": "Churned", "state": "Texas", "job_title": "Engineering Manager", "company": "Initech", "unit": "kg", "is_active": false}, {"email": "enlgpga@test.org", "amount": 491.44, "status": "Trial", "state": "Texas", "job_title": "Director", "company": "Stark Industries", "unit": "kg", "is_active": false}, {"email": "odld@example.com", "amount": 5483.43, "status": "Churned", "state": "Utah", "job_title": "Director", "company": "Hooli", "unit": "mi", "is_active": true}, {"email": "hlcegpi@corp.io", "amount": 8537.08, "status": "Trial", "state": "Texas", "job_title": "Chief Executive Officer", "company": "Umbrella", "unit": "mi", "is_active": false}, {"email": "nbfk@example.com", "amount": 1317.2, "status": "Churned", "state": "North Carolina", "job_title": "Engineering Manager", "company": "Stark Industries", "unit": "kg", "is_active": false}, {"email": "kghefe@corp.io", "amount": 1513.04, "status": "Churned", "state": "Arkansas", "job_title": "Engineering Manager", "company": "Wonka Co", "unit": "kg", "is_active": false}, {"email": "mplm@example.com", "amount": 7236.2, "status": "Trial", "state": "Alabama", "job_title": "Director", "company": "Soylent Corp", "unit": "mi", "is_active": true}, {"email": "pafkohh@mail.com", "amount": 2307.99, "status": "Trial", "state": "Utah", "job_title": "Director", "company": "Globex", "unit": "kg", "is_active": false}, {"email": "knoaieh@mail.com", "amount": NaN, "status": "Trial", "state": "Alabama", "job_title": "Director", "company": "Acme Inc", "unit": "mi", "is_active": false}, {"email": "nngman@test.org", "amount": 7512.45, "status": "Trial", "state": "Texas", "job_title": "Chief Executive Officer", "company": "Acme Inc", "unit": "kg", "is_active": false}, {"email": "jokil@corp.io", "amount": 7093.13, "status": "Trial", "state": "Arkansas", "job_title": "Chief Executive Officer", "company": "Vehement", "unit": "kg", "is_active": true}, {"email": "nkcl@mail.com", "amount": 8362.64, "status": "Churned", "state": "Alabama", "job_title": "Engineering Manager", "company": "Cyberdyne", "unit": "kg", "is_active": true}, {"email": "gnmhp@corp.io", "amount": 1576.12, "status": "Churned", "state": "Alabama", "job_title": "Engineering Manager", "company": "Soylent Corp", "unit": "mi", "is_active": true}, {"email": "paibgek@example.com", "amount": 1957.8, "status": "Trial", "state": "Texas", "job_title": "Chief Executive Officer", "company": "Initech", "unit": "mi", "is_active": true}, {"email": "kbfk@corp.io", "amount": 735.41, "status": "Churned", "state": "Arkansas", "job_title": "Engineering Manager", "company": "Vehement", "unit": "kg", "is_active": false}, {"email": "flllob@test.org", "amount": 4770.66, "status": "Churned", "state": "Arkansas", "job_title": "Director", "company": "Umbrella", "unit": "kg", "is_active": true}, {"email": "nofcnhl@mail.com", "amount": 6460.3, "status": "Trial", "state": "Alabama", "job_title": "Chief Executive Officer", "company": "Acme Inc", "unit": "kg", "is_active": false}, {"email": "pmdak@corp.io", "amount": 1858.83, "status": "Churned", "state": "Arkansas", "job_title": "Chief Executive Officer", "company": "Soylent Corp", "unit": "kg", "is_active": true}, {"email": "dlhb@mail.com", "amount": 5256.65, "status": "Trial", "state": "Utah", "job_title": "Engineering Manager", "company": "Soylent Corp", "unit": "mi", "is_active": true}, {"email": "bhjooii@corp.io", "amount": 4317.53, "status": "Churned", "state": "North Carolina", "job_title": "Engineering Manager", "company": "Acme Inc", "unit": "mi", "is_active": true}, {"email": "hcdpioj@mail.com", "amount": 2530.54, "status": "Trial", "state": "Utah", "job_title": "Director", "company": "Wonka Co", "unit": "kg", "is_active": false}, {"email": "plpdi@test.org", "amount": NaN, "status": "Trial", "state": "Arkansas", "job_title": "Chief Executive Officer", "company": "Hooli", "unit": "mi", "is_active": true}, {"email": "kghn@mail.com", "amount": 7384.78, "status": "Trial", "state": "Arkansas", "job_title": "Engineering Manager", "company": "Acme Inc", "unit": "kg", "is_active": true}, {"email": "gechhi@example.com", "amount": 3428.15, "status": "Churned", "state": "Utah", "job_title": "Chief Executive Officer", "company": "Stark Industries", "unit": "kg", "is_active": false}, {"email": "beeacdg@mail.com", "amount": 6535.2, "status": "Trial", "state": "Utah", "job_title": "Director", "company": "Vehement", "unit": "kg", "is_active": true}, {"email": "biihg@example.com", "amount": 7900.67, "status": "Trial", "state": "Alabama", "job_title": "Chief Executive Officer", "company": "Globex", "unit": "kg", "is_active": false}, {"email": "hmioaj@corp.io", "amount": -100.0, "status": "Trial", "state": "Utah", "job_title": "Director", "company": "Hooli", "unit": "kg", "is_active": true}, {"email": "eobim@mail.com", "amount": 8694.43, "status": "Trial", "state": "Arkansas", "job_title": "Director", "company": "Wonka Co", "unit": "mi", "is_active": true}, {"email": "pkdcgm@example.com", "amount": 3922.44, "status": "Trial", "state": "Utah", "job_title": "Chief Executive Officer", "company": "Acme Inc", "unit": "mi", "is_active": false}, {"email": "doeinf@test.org", "amount": 7084.78, "status": "Trial", "state": "North Carolina", "job_title": "Engineering Manager", "company": "Wonka Co", "unit": "kg", "is_active": true}, {"email": "egoii@example.com", "amount": 6506.44, "status": "Trial", "state": "Alabama", "job_title": "Engineering Manager", "company": "Soylent Corp", "unit": "mi", "is_active": false}, {"email": "achbb@corp.io", "amount": 8327.31, "status": "Churned", "state": "Texas", "job_title": "Director", "company": "Cyberdyne", "unit": "kg", "is_active": false}, {"email": "bedgcc@example.com", "amount": 113.82, "status": "Churned", "state": "Texas", "job_title": "Engineering Manager", "company": "Cyberdyne", "unit": "mi", "is_active": false}, {"email": "leoimn@test.org", "amount": 892.14, "status": "Trial", "state": "North Carolina", "job_title": "Chief Executive Officer", "company": "Wonka Co", "unit": "kg", "is_active": true}, {"email": "fdmhcea@mail.com", "amount": NaN, "status": "Trial", "state": "Alabama", "job_title": "Chief Executive Officer", "company": "Acme Inc", "unit": "kg", "is_active": true}, {"email": "nkgi@example.com", "amount": 4943.6, "status": "Trial", "state": "North Carolina", "job_title": "Engineering Manager", "company": "Soylent Corp", "unit": "kg", "is_active": false}, {"email": "jcdf@corp.io", "amount": 1347.83, "status": "Churned", "state": "Texas", "job_title": "Director", "company": "Initech", "unit": "kg", "is_active": false}, {"email": "iibc@mail.com", "amount": NaN, "status": "Trial", "state": "North Carolina", "job_title": "Chief Executive Officer", "company": "Umbrella", "unit": "mi", "is_active": false}, {"email": "amdc@example.com", "amount": 3022.27, "status": "Churned", "state": "Texas", "job_title": "Director", "company": "Wonka Co", "unit": "kg", "is_active": true}, {"email": "bifb@mail.com", "amount": 2070.42, "status": "Churned", "state": "Arkansas", "job_title": "Chief Executive Officer", "company": "Initech", "unit": "kg", "is_active": true}], "dirty_cols": ["email", "amount", "status", "state", "job_title", "company", "unit", "is_active"], "clean_cols": ["email", "amount", "status", "state", "job_title", "company", "unit", "is_active"], "plan": {"dataset_summary": "44 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cancelled": "Churned", "CANCELLED": "Churned", "canceled": "Churned", "Trail": "Trial", "TKIAL": "Trial", "trial": "Trial", "tbial": "Trial", "free trial": "Trial", "TRIAL": "Trial", "CHHURNED": "Churned", "CANCELED": "Churned", "Cancelled": "Churned", "Churnned": "Churned", "Thial": "Trial", "CANCCELED": "Churned", "TRILL": "Trial", "Free Tral": "Trial", "churnde": "Churned", "Triaal": "Trial", "churned": "Churned", "Free Trial": "Trial"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"al": "Alabama", "utah": "Utah", "ar": "Arkansas", "TX": "Texas", "Utaah": "Utah", "tx": "Texas", "NC": "North Carolina", "AR": "Arkansas", "UTTAH": "Utah", "Alabava": "Alabama", "Al": "Alabama", "AL": "Alabama", "ALABAMA": "Alabama", "ut": "Utah", "north craolina": "North Carolina", "Txeas": "Texas", "north carolina": "North Carolina", "TEXAS": "Texas", "Ar": "Arkansas"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Engineering Mgr": "Engineering Manager", "Dir": "Director", "eng manager": "Engineering Manager", "eng mgr": "Engineering Manager", "director": "Director", "CEO": "Chief Executive Officer", "Eng Manager": "Engineering Manager", "Eng Mgr": "Engineering Manager", "Dir.": "Director", "dir": "Director", "Ceo": "Chief Executive Officer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "ENG MGR": "Engineering Manager", "DIR": "Director", "engineering manager": "Engineering Manager", "C.E.O.": "Chief Executive Officer", "ceo": "Chief Executive Officer", "engineering mgr": "Engineering Manager", "DIRECTOR": "Director", "ENGINEERING MANAGER": "Engineering Manager"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miles": "mi", "Mi": "mi", "MI": "mi", "kilogram": "kg", "MILES": "mi", "Kg": "kg", "kilograms": "kg", "kiograms": "kg", "kiolgram": "kg", "KG": "kg", "mile": "mi", "KILOGRAM": "kg"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"signup_date": "1 Aug 2023", "unit": "   cm", "phone": "5830038457", "name": "Priya Moore", "notes2": ""}, {"signup_date": "2023-05-06", "unit": " KILOMETER ", "phone": "575-003-1342", "name": "Ivan Wong", "notes2": ""}, {"signup_date": "45242", "unit": "km", "phone": "(517)902-5845", "name": "Omar Adams", "notes2": ""}, {"signup_date": "2023-06-12", "unit": "  kilometer", "phone": "(535)946-6743", "name": "Karl Petrov", "notes2": ""}, {"signup_date": "45027", "unit": "CENTIMETERS", "phone": "5475555562", "name": "David Smith", "notes2": ""}, {"signup_date": "7/8/2023", "unit": "Cm", "phone": "5597530945", "name": "Heidi Novak", "notes2": ""}, {"signup_date": "2023-03-17", "unit": "Km", "phone": "507-435-4480", "name": "Omar Johnson  ", "notes2": ""}, {"signup_date": "5/15/2023", "unit": "CM", "phone": "511-160-9833", "name": "Priya Lee", "notes2": ""}, {"signup_date": "45131", "unit": "centimeter", "phone": "(585)038-1535", "name": "Karl  Petrov", "notes2": ""}, {"signup_date": "14 May 2023", "unit": "Centimeter", "phone": "5554993145", "name": "Grace  Khan", "notes2": ""}, {"signup_date": "2023-01-05", "unit": "kilometer", "phone": "5759532683", "name": "David Wong", "notes2": ""}, {"signup_date": "45113", "unit": "km", "phone": "5595716303", "name": "Heidi Moore", "notes2": ""}, {"signup_date": "3 May 2023", "unit": "km", "phone": "533.467.4789", "name": " Judy Johnson", "notes2": ""}, {"signup_date": "45131", "unit": "centimeter", "phone": "(585)038-1535", "name": "Karl  Petrov", "notes2": ""}, {"signup_date": "7/8/2023", "unit": "Cm", "phone": "5597530945", "name": "Heidi Novak", "notes2": ""}, {"signup_date": "", "unit": "", "phone": "", "name": "", "notes2": ""}], "clean": [{"signup_date": "2023-08-01", "unit": "cm", "phone": "(583) 003-8457", "name": "Priya Moore"}, {"signup_date": "2023-05-06", "unit": "km", "phone": "(575) 003-1342", "name": "Ivan Wong"}, {"signup_date": "2023-11-12", "unit": "km", "phone": "(517) 902-5845", "name": "Omar Adams"}, {"signup_date": "2023-06-12", "unit": "km", "phone": "(535) 946-6743", "name": "Karl Petrov"}, {"signup_date": "2023-04-11", "unit": "cm", "phone": "(547) 555-5562", "name": "David Smith"}, {"signup_date": "2023-07-08", "unit": "cm", "phone": "(559) 753-0945", "name": "Heidi Novak"}, {"signup_date": "2023-03-17", "unit": "km", "phone": "(507) 435-4480", "name": "Omar Johnson"}, {"signup_date": "2023-05-15", "unit": "cm", "phone": "(511) 160-9833", "name": "Priya Lee"}, {"signup_date": "2023-07-24", "unit": "cm", "phone": "(585) 038-1535", "name": "Karl Petrov"}, {"signup_date": "2023-05-14", "unit": "cm", "phone": "(555) 499-3145", "name": "Grace Khan"}, {"signup_date": "2023-01-05", "unit": "km", "phone": "(575) 953-2683", "name": "David Wong"}, {"signup_date": "2023-07-06", "unit": "km", "phone": "(559) 571-6303", "name": "Heidi Moore"}, {"signup_date": "2023-05-03", "unit": "km", "phone": "(533) 467-4789", "name": "Judy Johnson"}], "dirty_cols": ["signup_date", "unit", "phone", "name", "notes2"], "clean_cols": ["signup_date", "unit", "phone", "name"], "plan": {"dataset_summary": "16 rows × 5 columns. 4 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KILOMETER": "km", "kilometer": "km", "CENTIMETERS": "cm", "Cm": "cm", "Km": "km", "CM": "cm", "centimeter": "cm", "Centimeter": "cm"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"state": "Georgia", "company": "Umbrella", "phone": "5748272472", "col_x": ""}, {"state": "WYOMING", "company": "Vehement", "phone": "(588)858-8112", "col_x": ""}, {"state": "Puerto Rico", "company": "   Globex", "phone": "(589)916-8788", "col_x": ""}, {"state": "GA", "company": "Stark Industries", "phone": "5343709145", "col_x": ""}, {"state": "HI", "company": "Initech", "phone": "552.450.8038", "col_x": ""}, {"state": "   Georgia", "company": "Wonka Co", "phone": "5953361314", "col_x": ""}, {"state": "wyoming", "company": "Wonka Co", "phone": "541.720.0918", "col_x": ""}, {"state": "HAWAII", "company": "Umbrella", "phone": "505.621.1462", "col_x": ""}, {"state": "INDIANA", "company": " Initech ", "phone": "5181232252", "col_x": ""}, {"state": "georgia", "company": "Umbrella", "phone": "(527)616-4151", "col_x": ""}, {"state": "PR", "company": "Soylent  Corp", "phone": "(503)831-8906", "col_x": ""}, {"state": "Puerto Rico", "company": "   Globex", "phone": "(589)916-8788", "col_x": ""}, {"state": "PR", "company": "Soylent  Corp", "phone": "(503)831-8906", "col_x": ""}], "clean": [{"state": "Georgia", "company": "Umbrella", "phone": "(574) 827-2472"}, {"state": "Wyoming", "company": "Vehement", "phone": "(588) 858-8112"}, {"state": "Puerto Rico", "company": "Globex", "phone": "(589) 916-8788"}, {"state": "Georgia", "company": "Stark Industries", "phone": "(534) 370-9145"}, {"state": "Hawaii", "company": "Initech", "phone": "(552) 450-8038"}, {"state": "Georgia", "company": "Wonka Co", "phone": "(595) 336-1314"}, {"state": "Wyoming", "company": "Wonka Co", "phone": "(541) 720-0918"}, {"state": "Hawaii", "company": "Umbrella", "phone": "(505) 621-1462"}, {"state": "Indiana", "company": "Initech", "phone": "(518) 123-2252"}, {"state": "Georgia", "company": "Umbrella", "phone": "(527) 616-4151"}, {"state": "Puerto Rico", "company": "Soylent Corp", "phone": "(503) 831-8906"}], "dirty_cols": ["state", "company", "phone", "col_x"], "clean_cols": ["state", "company", "phone"], "plan": {"dataset_summary": "13 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WYOMING": "Wyoming", "GA": "Georgia", "HI": "Hawaii", "wyoming": "Wyoming", "HAWAII": "Hawaii", "INDIANA": "Indiana", "georgia": "Georgia", "PR": "Puerto Rico"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"city": "   Bueng Kum", "unit": "cm  ", "status": "In-Progress", "is_active": "TRUE", "state": "FL", "rate": "60.1%", "extra": ""}, {"city": "Aston", "unit": " KG ", "status": "closed won", "is_active": "T", "state": "fl", "rate": "86.6%", "extra": ""}, {"city": "   buen gkum", "unit": "kg  ", "status": "WIP", "is_active": "TRUE", "state": "ok ", "rate": "73.1%", "extra": ""}, {"city": "Bueng Kum  ", "unit": "centimeter", "status": "won", "is_active": "TRUE", "state": "FL", "rate": "19.9%", "extra": ""}, {"city": "Aston", "unit": "cm", "status": "  won", "is_active": "F", "state": "Wyoming", "rate": "59.3%", "extra": ""}, {"city": "Aston", "unit": "kilograms", "status": "inprgoress  ", "is_active": "Y", "state": "United States Minor  Outlying Islands", "rate": "31.6%", "extra": ""}, {"city": "bueng kum", "unit": "cm", "status": " lost ", "is_active": "Yes", "state": " WY ", "rate": "66.5%", "extra": ""}, {"city": "aston", "unit": "   centimeter", "status": "CLOSED LOST", "is_active": "0", "state": "fl", "rate": "83.2%", "extra": ""}, {"city": "aston", "unit": "KG", "status": "WIP", "is_active": "TRUE", "state": "WY", "rate": "65.3%", "extra": ""}, {"city": "BUENG  KUM", "unit": "cm", "status": "in progress", "is_active": "T", "state": "Florida", "rate": "53.6%", "extra": ""}, {"city": "bueng kum", "unit": "centimeter", "status": "lost", "is_active": "No", "state": "UM", "rate": "5.8%", "extra": ""}, {"city": "Amton ", "unit": "kg", "status": "closed lost", "is_active": "true", "state": "FL", "rate": "62.7%", "extra": ""}, {"city": "   Bueng Kum", "unit": " CENTIMETER", "status": "In Progress", "is_active": "F", "state": "Wyomyng", "rate": "76.5%", "extra": ""}, {"city": "Bueng Kum", "unit": "KILOGRAM", "status": "Closed Lost", "is_active": "Y", "state": "United States Minor Outlying Islands", "rate": "4.2%", "extra": ""}, {"city": "Bueng Kum", "unit": "Kg", "status": "closed lost", "is_active": "T", "state": "OKLAHOMA", "rate": "44.9%", "extra": ""}, {"city": "Bueng Kum", "unit": "kg", "status": "Won", "is_active": "Y", "state": "ok", "rate": "71.9%", "extra": ""}, {"city": "Bueng Kum", "unit": "kg", "status": "in-progress", "is_active": "0", "state": "WY   ", "rate": "19.7%", "extra": ""}, {"city": "Bueng Kum", "unit": "KILOGRAM", "status": "won", "is_active": "No", "state": "WY", "rate": "38.4%", "extra": ""}, {"city": "ASTON", "unit": "cm", "status": "in progress", "is_active": "FALSE", "state": "United States Minor Outlying Islwnds", "rate": "61.3%", "extra": ""}, {"city": "aston", "unit": "   KG", "status": "Lost", "is_active": "No", "state": " fl ", "rate": "87.2%", "extra": ""}, {"city": "Bueng Kum", "unit": "CM", "status": "Closed Won", "is_active": "T", "state": "Floorida", "rate": "32.3%", "extra": ""}, {"city": "Bueng Kum", "unit": "centimeters", "status": "Closed Lost", "is_active": "Yes", "state": "Wyoming", "rate": "39.3%", "extra": ""}, {"city": "Bueng Kum", "unit": "kg", "status": "   won", "is_active": "F", "state": "FL  ", "rate": "45.1%", "extra": ""}, {"city": "   bueng kum", "unit": "Cm", "status": "lost", "is_active": "TRUE", "state": "  OK", "rate": "68.7%", "extra": ""}, {"city": "Aston", "unit": "KILOGRAM", "status": "won", "is_active": "F", "state": "OK ", "rate": "76.8%", "extra": ""}, {"city": " Aston", "unit": " KG ", "status": "in progress", "is_active": "false", "state": "FLORIDA", "rate": "37.9%", "extra": ""}, {"city": "aston", "unit": "centimeter", "status": "In  Progress", "is_active": "1", "state": "wsoming", "rate": "82.2%", "extra": ""}, {"city": "aston", "unit": "KG", "status": "Won", "is_active": "N", "state": "Wy", "rate": "91.3%", "extra": ""}], "clean": [{"city": "Bueng Kum", "unit": "cm", "status": "In Progress", "is_active": true, "state": "Florida", "rate": 0.601}, {"city": "Aston", "unit": "kg", "status": "Won", "is_active": true, "state": "Florida", "rate": 0.866}, {"city": "Bueng Kum", "unit": "kg", "status": "In Progress", "is_active": true, "state": "Oklahoma", "rate": 0.731}, {"city": "Bueng Kum", "unit": "cm", "status": "Won", "is_active": true, "state": "Florida", "rate": 0.19899999999999998}, {"city": "Aston", "unit": "cm", "status": "Won", "is_active": false, "state": "Wyoming", "rate": 0.593}, {"city": "Aston", "unit": "kg", "status": "In Progress", "is_active": true, "state": "United States Minor Outlying Islands", "rate": 0.316}, {"city": "Bueng Kum", "unit": "cm", "status": "Lost", "is_active": true, "state": "Wyoming", "rate": 0.665}, {"city": "Aston", "unit": "cm", "status": "Lost", "is_active": false, "state": "Florida", "rate": 0.8320000000000001}, {"city": "Aston", "unit": "kg", "status": "In Progress", "is_active": true, "state": "Wyoming", "rate": 0.653}, {"city": "Bueng Kum", "unit": "cm", "status": "In Progress", "is_active": true, "state": "Florida", "rate": 0.536}, {"city": "Bueng Kum", "unit": "cm", "status": "Lost", "is_active": false, "state": "United States Minor Outlying Islands", "rate": 0.057999999999999996}, {"city": "Aston", "unit": "kg", "status": "Lost", "is_active": true, "state": "Florida", "rate": 0.627}, {"city": "Bueng Kum", "unit": "cm", "status": "In Progress", "is_active": false, "state": "Wyoming", "rate": 0.765}, {"city": "Bueng Kum", "unit": "kg", "status": "Lost", "is_active": true, "state": "United States Minor Outlying Islands", "rate": 0.042}, {"city": "Bueng Kum", "unit": "kg", "status": "Lost", "is_active": true, "state": "Oklahoma", "rate": 0.449}, {"city": "Bueng Kum", "unit": "kg", "status": "Won", "is_active": true, "state": "Oklahoma", "rate": 0.7190000000000001}, {"city": "Bueng Kum", "unit": "kg", "status": "In Progress", "is_active": false, "state": "Wyoming", "rate": 0.19699999999999998}, {"city": "Bueng Kum", "unit": "kg", "status": "Won", "is_active": false, "state": "Wyoming", "rate": 0.384}, {"city": "Aston", "unit": "cm", "status": "In Progress", "is_active": false, "state": "United States Minor Outlying Islands", "rate": 0.613}, {"city": "Aston", "unit": "kg", "status": "Lost", "is_active": false, "state": "Florida", "rate": 0.872}, {"city": "Bueng Kum", "unit": "cm", "status": "Won", "is_active": true, "state": "Florida", "rate": 0.32299999999999995}, {"city": "Bueng Kum", "unit": "cm", "status": "Lost", "is_active": true, "state": "Wyoming", "rate": 0.39299999999999996}, {"city": "Bueng Kum", "unit": "kg", "status": "Won", "is_active": false, "state": "Florida", "rate": 0.451}, {"city": "Bueng Kum", "unit": "cm", "status": "Lost", "is_active": true, "state": "Oklahoma", "rate": 0.687}, {"city": "Aston", "unit": "kg", "status": "Won", "is_active": false, "state": "Oklahoma", "rate": 0.768}, {"city": "Aston", "unit": "kg", "status": "In Progress", "is_active": false, "state": "Florida", "rate": 0.379}, {"city": "Aston", "unit": "cm", "status": "In Progress", "is_active": true, "state": "Wyoming", "rate": 0.8220000000000001}, {"city": "Aston", "unit": "kg", "status": "Won", "is_active": false, "state": "Wyoming", "rate": 0.9129999999999999}], "dirty_cols": ["city", "unit", "status", "is_active", "state", "rate", "extra"], "clean_cols": ["city", "unit", "status", "is_active", "state", "rate"], "plan": {"dataset_summary": "28 rows × 7 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"buen gkum": "Bueng Kum", "bueng kum": "Bueng Kum", "aston": "Aston", "BUENG KUM": "Bueng Kum", "Amton": "Aston", "ASTON": "Aston"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KG": "kg", "centimeter": "cm", "kilograms": "kg", "CENTIMETER": "cm", "KILOGRAM": "kg", "Kg": "kg", "CM": "cm", "centimeters": "cm", "Cm": "cm"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"In-Progress": "In Progress", "closed won": "Won", "WIP": "In Progress", "won": "Won", "inprgoress": "In Progress", "lost": "Lost", "CLOSED LOST": "Lost", "in progress": "In Progress", "closed lost": "Lost", "Closed Lost": "Lost", "in-progress": "In Progress", "Closed Won": "Won"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"FL": "Florida", "fl": "Florida", "ok": "Oklahoma", "WY": "Wyoming", "UM": "United States Minor Outlying Islands", "Wyomyng": "Wyoming", "OKLAHOMA": "Oklahoma", "United States Minor Outlying Islwnds": "United States Minor Outlying Islands", "Floorida": "Florida", "OK": "Oklahoma", "FLORIDA": "Florida", "wsoming": "Wyoming", "Wy": "Wyoming"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"state": "ID", "unit": "litre", "email": "PBNBAJ@TEST.ORG", "country": "Republic Of Haiti", "industry": "Tourism", "is_active": "F", "job_title": "  c..t.o."}, {"state": "kentucky", "unit": "centimeter", "email": "neni@mail.com", "country": "Islamic Republic of Afghanistan ", "industry": "Industral", "is_active": "false", "job_title": "vice president"}, {"state": "Idaho", "unit": "l", "email": "OICHP@MAIL.COM ", "country": "HT", "industry": "Hospittality", "is_active": "F", "job_title": "Acct Exec"}, {"state": "Idaho   ", "unit": "cm", "email": " hdbadjd@test.org", "country": "SE", "industry": "Comms", "is_active": "F", "job_title": "V.P."}, {"state": "Idaho", "unit": "CM", "email": "pgjbl@example.com", "country": "SE", "industry": "   retail", "is_active": "TRUE", "job_title": "Acct Exec"}, {"state": "INDIANA", "unit": "cenntimeter", "email": "MJCLOIP@EXAMPLE.COM", "country": " U.S. ", "industry": "telecom", "is_active": "true", "job_title": "EngqMgr"}, {"state": "in", "unit": "LITER", "email": "MHMHDLJ@CORP.IO", "country": "afghanistan", "industry": "Tourism", "is_active": "T", "job_title": "Eng Mgr"}, {"state": "ID", "unit": "Km", "email": " jpmedja@corp.io ", "country": "   af", "industry": "Mfg", "is_active": "Y", "job_title": "CTO "}, {"state": "ID", "unit": "kilometers  ", "email": "DMFEC@CORP.IO", "country": "republic of haiti", "industry": "telco", "is_active": "Yes", "job_title": "AE  "}, {"state": "kentucky", "unit": " cm ", "email": "LNLLGIG@MAIL.COM", "country": "USA", "industry": "manufacturing", "is_active": "0", "job_title": " AE "}, {"state": "KY", "unit": "km", "email": "NGPIBAI@MAIL.COM", "country": "Argentina", "industry": "E-Cmomerce", "is_active": "F", "job_title": "eng mgr"}, {"state": " IN ", "unit": " L ", "email": "BEBHA@EXAMPLE.COM", "country": "   AF", "industry": "Hospitality", "is_active": "FALSE", "job_title": "Ae"}, {"state": " Idaho ", "unit": "KILOMETERS", "email": "PKOB@TEST.ORG", "country": "Sweden", "industry": "retail & e-commerce", "is_active": "false", "job_title": "Chief Technology Officer"}, {"state": "Kentucky", "unit": "cm", "email": "apiidei@example.com", "country": "united states", "industry": "HOTELS & TRAVEL", "is_active": "true", "job_title": "Engineering Mgr"}, {"state": "id", "unit": "centimeter", "email": "kbik@example.com", "country": "Afghanistan", "industry": "COMES", "is_active": "N", "job_title": " VP "}, {"state": "   Idaho", "unit": "L", "email": " DAADCDO@CORP.IO ", "country": "SE", "industry": "   travel", "is_active": "N", "job_title": "Chief Technology Officer"}, {"state": "idaho", "unit": "kilometers", "email": "HPGHE@MAIL.COM", "country": "hiati", "industry": "Hotels & Travel", "is_active": "Y", "job_title": "  v.p."}, {"state": "KENTUCKY", "unit": "LITERS", "email": "npmb@corp.io", "country": "SWE", "industry": "TELECOM ", "is_active": "Y", "job_title": "   engineering manager"}, {"state": "indiana", "unit": "centimeters", "email": "jcoc@test.org", "country": "Kingdom of Sweden   ", "industry": "  Manuffacturing", "is_active": "FALSE", "job_title": "Engineering Mgr"}, {"state": "iniana", "unit": "centimeter  ", "email": " IDPDJI@CORP.IO", "country": "argentine republic", "industry": "Hospitality", "is_active": "FALSE", "job_title": " ae "}, {"state": "ky", "unit": "km", "email": "hdcno@example.com", "country": "Islamic Republic of Afghanistan", "industry": "Telecom", "is_active": "true", "job_title": "  AE"}, {"state": " Idaho", "unit": "Centimeters", "email": "   KMNI@MAIL.COM", "country": "Kingdom of Sweden", "industry": "INDUSTRIAL", "is_active": "0", "job_title": "chief technology officer"}, {"state": "Idaho", "unit": "Centmieters", "email": " IMDOEOE@CORP.IO ", "country": "united states ", "industry": "MFG", "is_active": "TRUE", "job_title": "Vice President"}, {"state": " indiana ", "unit": "km", "email": "LJFDKE@TEST.ORG", "country": "Argentine Republic", "industry": "Hospitality", "is_active": "false", "job_title": " eng manager"}, {"state": "  ID", "unit": "liters", "email": "MKFONJ@CORP.IO", "country": "united states of america", "industry": "TELECOM", "is_active": "Y", "job_title": " vp "}, {"state": "ID", "unit": " kilometers ", "email": "mldoefc@example.com", "country": "AR  ", "industry": "RETAIL", "is_active": "0", "job_title": "eng mgr"}, {"state": "IN ", "unit": " l", "email": " NDAFAG@TEST.ORG ", "country": "argentinee republic   ", "industry": "Industrial", "is_active": "F", "job_title": "VP"}, {"state": "ID", "unit": "centimeters", "email": "MGPL@MAIL.COM", "country": "ARG", "industry": "rteail", "is_active": "Y", "job_title": "   Vice President"}, {"state": "KY", "unit": " liters ", "email": "DPFPJBO@EXAMPLE.COM", "country": "swe", "industry": "Retail", "is_active": "No", "job_title": "Vice President"}, {"state": "Kextucky", "unit": "centimeter", "email": "KHLPIAK@EXAMPLE.COM", "country": "se", "industry": "tourism", "is_active": "No", "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "km", "email": "bnhii@example.com", "country": "United States", "industry": "E-commerce", "is_active": "Yes", "job_title": " Chief Technology Officer"}, {"state": "INDIANA", "unit": "Km", "email": "BODEFPE@CORP.IO", "country": "afghanistan", "industry": "E-commerce", "is_active": "false", "job_title": "chief technology officer"}, {"state": "  Kentucky", "unit": " litre", "email": "gkhc@test.org", "country": "HAITI", "industry": "Telco", "is_active": "TRUE", "job_title": "AE"}, {"state": "id", "unit": " CENTIMETERS ", "email": " KGDFFF@EXAMPLE.COM ", "country": "sweden", "industry": "Telecom   ", "is_active": "FALSE", "job_title": " eng manager"}, {"state": "Indiana", "unit": "Km", "email": " PCEG@MAIL.COM", "country": "AR", "industry": "travel", "is_active": "N", "job_title": "Engineering Manager"}, {"state": "KY", "unit": "KILOMETERS", "email": "PAJMAPD@EXAMPLE.COM", "country": "Kingdom of Sweden  ", "industry": "   INDUSTRIAL", "is_active": "F", "job_title": "Acctt Exec"}, {"state": "IN", "unit": "centimeters", "email": "HGKKHMI@CORP.IO", "country": " United States of America", "industry": "manufacturing", "is_active": "1", "job_title": "VP"}, {"state": "ID", "unit": "l", "email": "KLLJ@MAIL.COM", "country": "Arg", "industry": "retail", "is_active": "1", "job_title": "vice president"}, {"state": "Idaho", "unit": "LITRE", "email": "fbeomb@mail.com", "country": "AFGHANISTAN", "industry": " Hotels & Travel", "is_active": "FALSE", "job_title": "Eng Mgr "}, {"state": "Idaho", "unit": "Centmieters", "email": " IMDOEOE@CORP.IO ", "country": "united states ", "industry": "MFG", "is_active": "TRUE", "job_title": "Vice President"}, {"state": "Kentucky", "unit": "cm", "email": "apiidei@example.com", "country": "united states", "industry": "HOTELS & TRAVEL", "is_active": "true", "job_title": "Engineering Mgr"}], "clean": [{"state": "Idaho", "unit": "L", "email": "pbnbaj@test.org", "country": "Haiti", "industry": "Hospitality", "is_active": false, "job_title": "Chief Technology Officer"}, {"state": "Kentucky", "unit": "cm", "email": "neni@mail.com", "country": "Afghanistan", "industry": "Manufacturing", "is_active": false, "job_title": "Vice President"}, {"state": "Idaho", "unit": "L", "email": "oichp@mail.com", "country": "Haiti", "industry": "Hospitality", "is_active": false, "job_title": "Account Executive"}, {"state": "Idaho", "unit": "cm", "email": "hdbadjd@test.org", "country": "Sweden", "industry": "Telecommunications", "is_active": false, "job_title": "Vice President"}, {"state": "Idaho", "unit": "cm", "email": "pgjbl@example.com", "country": "Sweden", "industry": "Retail", "is_active": true, "job_title": "Account Executive"}, {"state": "Indiana", "unit": "cm", "email": "mjcloip@example.com", "country": "United States", "industry": "Telecommunications", "is_active": true, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "L", "email": "mhmhdlj@corp.io", "country": "Afghanistan", "industry": "Hospitality", "is_active": true, "job_title": "Engineering Manager"}, {"state": "Idaho", "unit": "km", "email": "jpmedja@corp.io", "country": "Afghanistan", "industry": "Manufacturing", "is_active": true, "job_title": "Chief Technology Officer"}, {"state": "Idaho", "unit": "km", "email": "dmfec@corp.io", "country": "Haiti", "industry": "Telecommunications", "is_active": true, "job_title": "Account Executive"}, {"state": "Kentucky", "unit": "cm", "email": "lnllgig@mail.com", "country": "United States", "industry": "Manufacturing", "is_active": false, "job_title": "Account Executive"}, {"state": "Kentucky", "unit": "km", "email": "ngpibai@mail.com", "country": "Argentina", "industry": "Retail", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "L", "email": "bebha@example.com", "country": "Afghanistan", "industry": "Hospitality", "is_active": false, "job_title": "Account Executive"}, {"state": "Idaho", "unit": "km", "email": "pkob@test.org", "country": "Sweden", "industry": "Retail", "is_active": false, "job_title": "Chief Technology Officer"}, {"state": "Kentucky", "unit": "cm", "email": "apiidei@example.com", "country": "United States", "industry": "Hospitality", "is_active": true, "job_title": "Engineering Manager"}, {"state": "Idaho", "unit": "cm", "email": "kbik@example.com", "country": "Afghanistan", "industry": "Telecommunications", "is_active": false, "job_title": "Vice President"}, {"state": "Idaho", "unit": "L", "email": "daadcdo@corp.io", "country": "Sweden", "industry": "Hospitality", "is_active": false, "job_title": "Chief Technology Officer"}, {"state": "Idaho", "unit": "km", "email": "hpghe@mail.com", "country": "Haiti", "industry": "Hospitality", "is_active": true, "job_title": "Vice President"}, {"state": "Kentucky", "unit": "L", "email": "npmb@corp.io", "country": "Sweden", "industry": "Telecommunications", "is_active": true, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "cm", "email": "jcoc@test.org", "country": "Sweden", "industry": "Manufacturing", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "cm", "email": "idpdji@corp.io", "country": "Argentina", "industry": "Hospitality", "is_active": false, "job_title": "Account Executive"}, {"state": "Kentucky", "unit": "km", "email": "hdcno@example.com", "country": "Afghanistan", "industry": "Telecommunications", "is_active": true, "job_title": "Account Executive"}, {"state": "Idaho", "unit": "cm", "email": "kmni@mail.com", "country": "Sweden", "industry": "Manufacturing", "is_active": false, "job_title": "Chief Technology Officer"}, {"state": "Idaho", "unit": "cm", "email": "imdoeoe@corp.io", "country": "United States", "industry": "Manufacturing", "is_active": true, "job_title": "Vice President"}, {"state": "Indiana", "unit": "km", "email": "ljfdke@test.org", "country": "Argentina", "industry": "Hospitality", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Idaho", "unit": "L", "email": "mkfonj@corp.io", "country": "United States", "industry": "Telecommunications", "is_active": true, "job_title": "Vice President"}, {"state": "Idaho", "unit": "km", "email": "mldoefc@example.com", "country": "Argentina", "industry": "Retail", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "L", "email": "ndafag@test.org", "country": "Argentina", "industry": "Manufacturing", "is_active": false, "job_title": "Vice President"}, {"state": "Idaho", "unit": "cm", "email": "mgpl@mail.com", "country": "Argentina", "industry": "Retail", "is_active": true, "job_title": "Vice President"}, {"state": "Kentucky", "unit": "L", "email": "dpfpjbo@example.com", "country": "Sweden", "industry": "Retail", "is_active": false, "job_title": "Vice President"}, {"state": "Kentucky", "unit": "cm", "email": "khlpiak@example.com", "country": "Sweden", "industry": "Hospitality", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "km", "email": "bnhii@example.com", "country": "United States", "industry": "Retail", "is_active": true, "job_title": "Chief Technology Officer"}, {"state": "Indiana", "unit": "km", "email": "bodefpe@corp.io", "country": "Afghanistan", "industry": "Retail", "is_active": false, "job_title": "Chief Technology Officer"}, {"state": "Kentucky", "unit": "L", "email": "gkhc@test.org", "country": "Haiti", "industry": "Telecommunications", "is_active": true, "job_title": "Account Executive"}, {"state": "Idaho", "unit": "cm", "email": "kgdfff@example.com", "country": "Sweden", "industry": "Telecommunications", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Indiana", "unit": "km", "email": "pceg@mail.com", "country": "Argentina", "industry": "Hospitality", "is_active": false, "job_title": "Engineering Manager"}, {"state": "Kentucky", "unit": "km", "email": "pajmapd@example.com", "country": "Sweden", "industry": "Manufacturing", "is_active": false, "job_title": "Account Executive"}, {"state": "Indiana", "unit": "cm", "email": "hgkkhmi@corp.io", "country": "United States", "industry": "Manufacturing", "is_active": true, "job_title": "Vice President"}, {"state": "Idaho", "unit": "L", "email": "kllj@mail.com", "country": "Argentina", "industry": "Retail", "is_active": true, "job_title": "Vice President"}, {"state": "Idaho", "unit": "L", "email": "fbeomb@mail.com", "country": "Afghanistan", "industry": "Hospitality", "is_active": false, "job_title": "Engineering Manager"}], "dirty_cols": ["state", "unit", "email", "country", "industry", "is_active", "job_title"], "clean_cols": ["state", "unit", "email", "country", "industry", "is_active", "job_title"], "plan": {"dataset_summary": "41 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ID": "Idaho", "kentucky": "Kentucky", "INDIANA": "Indiana", "in": "Indiana", "KY": "Kentucky", "IN": "Indiana", "id": "Idaho", "idaho": "Idaho", "KENTUCKY": "Kentucky", "indiana": "Indiana", "iniana": "Indiana", "ky": "Kentucky", "Kextucky": "Kentucky"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"litre": "L", "centimeter": "cm", "l": "L", "CM": "cm", "cenntimeter": "cm", "LITER": "L", "Km": "km", "kilometers": "km", "KILOMETERS": "km", "LITERS": "L", "centimeters": "cm", "Centimeters": "cm", "Centmieters": "cm", "liters": "L", "CENTIMETERS": "cm", "LITRE": "L"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Republic Of Haiti": "Haiti", "Islamic Republic of Afghanistan": "Afghanistan", "HT": "Haiti", "SE": "Sweden", "U.S.": "United States", "afghanistan": "Afghanistan", "af": "Afghanistan", "republic of haiti": "Haiti", "USA": "United States", "AF": "Afghanistan", "united states": "United States", "hiati": "Haiti", "SWE": "Sweden", "Kingdom of Sweden": "Sweden", "argentine republic": "Argentina", "Argentine Republic": "Argentina", "united states of america": "United States", "AR": "Argentina", "argentinee republic": "Argentina", "ARG": "Argentina", "swe": "Sweden", "se": "Sweden", "HAITI": "Haiti", "sweden": "Sweden", "United States of America": "United States", "Arg": "Argentina", "AFGHANISTAN": "Afghanistan"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Tourism": "Hospitality", "Industral": "Manufacturing", "Hospittality": "Hospitality", "Comms": "Telecommunications", "retail": "Retail", "telecom": "Telecommunications", "Mfg": "Manufacturing", "telco": "Telecommunications", "manufacturing": "Manufacturing", "E-Cmomerce": "Retail", "retail & e-commerce": "Retail", "HOTELS & TRAVEL": "Hospitality", "COMES": "Telecommunications", "travel": "Hospitality", "Hotels & Travel": "Hospitality", "TELECOM": "Telecommunications", "Manuffacturing": "Manufacturing", "Telecom": "Telecommunications", "INDUSTRIAL": "Manufacturing", "MFG": "Manufacturing", "RETAIL": "Retail", "Industrial": "Manufacturing", "rteail": "Retail", "tourism": "Hospitality", "E-commerce": "Retail", "Telco": "Telecommunications"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"c..t.o.": "Chief Technology Officer", "vice president": "Vice President", "Acct Exec": "Account Executive", "V.P.": "Vice President", "EngqMgr": "Engineering Manager", "Eng Mgr": "Engineering Manager", "CTO": "Chief Technology Officer", "AE": "Account Executive", "eng mgr": "Engineering Manager", "Ae": "Account Executive", "Engineering Mgr": "Engineering Manager", "VP": "Vice President", "v.p.": "Vice President", "engineering manager": "Engineering Manager", "ae": "Account Executive", "chief technology officer": "Chief Technology Officer", "eng manager": "Engineering Manager", "vp": "Vice President", "Acctt Exec": "Account Executive"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "(577)249-2874", "country": "swiss confederation", "industry": "energy", "email": "OJACBJM@MAIL.COM", "job_title": "Administrative Assistant"}, {"phone": "(510)099-0179", "country": "rw", "industry": "finesrv", "email": "IMBDN@CORP.IO  ", "job_title": "Admin"}, {"phone": "5574056034", "country": "  Switzerland", "industry": "E-commerce", "email": "  FPFENG@EXAMPLE.COM", "job_title": "Admin Asst"}, {"phone": "570.227.0587", "country": "  Suriname", "industry": "finance", "email": "blfejk@corp.io", "job_title": "rep"}, {"phone": "597-072-5286", "country": "Bermuda", "industry": "Baoking", "email": "OLNIIK@MAIL.COM", "job_title": "ADMIN ASST"}, {"phone": "586.913.4188", "country": "republic of suriname  ", "industry": "   financial services", "email": "ggogna@example.com", "job_title": "admin asst"}, {"phone": "524.461.4626", "country": "Rwanda", "industry": "retail", "email": " kdfkk@corp.io ", "job_title": " Addmin "}, {"phone": "545-655-5572", "country": "RW   ", "industry": " Finance ", "email": "  caiggco@corp.io", "job_title": "Admin"}, {"phone": "537.141.9675", "country": "Rwnada", "industry": "OIL & GAS ", "email": " LAAHPDN@EXAMPLE.COM", "job_title": "Sales Representative"}, {"phone": "529.895.6055", "country": "CHE", "industry": "  Health", "email": "  JFJFD@MAIL.COM", "job_title": "Sales Representative"}, {"phone": "576.403.8350", "country": "SWISS CONFEDERATION", "industry": "finance", "email": "JKAO@CORP.IO", "job_title": "Rep"}, {"phone": "533-876-0057", "country": "Bermuda", "industry": "oil & gas", "email": "hbnfcbm@corp.io  ", "job_title": "rep"}, {"phone": "(572)436-4217", "country": "   SUR", "industry": "ecommerce", "email": " ejeg@test.org ", "job_title": "Rep"}, {"phone": "513-517-2564", "country": "RWA", "industry": "uitlities", "email": "MNEMGEJ@CORP.IO  ", "job_title": "Sales Rep"}, {"phone": "502.784.0379", "country": " Rwanda ", "industry": "Health Care", "email": "khkiejj@test.org", "job_title": "Admin   "}, {"phone": "539.312.3859", "country": "bermuda", "industry": "health care", "email": "mhnjclg@corp.io", "job_title": "Sales Rep"}, {"phone": "592-796-1101", "country": " Bermuda ", "industry": "HEALTH", "email": "KOIO@CORP.IO", "job_title": "administrative assistant   "}, {"phone": "(583)487-5122", "country": "sr ", "industry": "Medical", "email": "iedcd@mail.com", "job_title": " SALES REPRESENTATIVE"}, {"phone": "518.148.4826", "country": "rwanda", "industry": "retail", "email": "hgdae@corp.io", "job_title": "Sales Representative"}, {"phone": "558-681-4132", "country": "Sur", "industry": "finance", "email": "jfldon@mail.com", "job_title": "Sales Representative"}, {"phone": "(568)096-5624", "country": "RWA", "industry": "Healthcare", "email": "   FIEIPAH@EXAMPLE.COM", "job_title": "Rep"}, {"phone": "510-570-1873", "country": " RW", "industry": "energy", "email": " OEBMFCB@MAIL.COM", "job_title": "rep"}, {"phone": "(547)704-7372", "country": "BMU", "industry": "Fnance   ", "email": "lhkgio@example.com", "job_title": " REP "}, {"phone": "5348024309", "country": "RW", "industry": "Energy", "email": "EBFOH@TEST.ORG", "job_title": "Rep"}, {"phone": "596-352-9019", "country": "CHE", "industry": "Healt Care", "email": "EOHDIPI@MAIL.COM", "job_title": "Rep"}, {"phone": "(583)212-8554", "country": "bmu", "industry": " Retail ", "email": "lnlgah@test.org", "job_title": "ADMIN ASSISTANT"}, {"phone": "515.850.8351", "country": "Bremuda", "industry": "Retail", "email": " cpclbck@corp.io ", "job_title": "ADMINISTRATIVE ASSISTANT"}, {"phone": "529.895.6055", "country": "CHE", "industry": "  Health", "email": "  JFJFD@MAIL.COM", "job_title": "Sales Representative"}], "clean": [{"phone": "(577) 249-2874", "country": "Switzerland", "industry": "Energy", "email": "ojacbjm@mail.com", "job_title": "Administrative Assistant"}, {"phone": "(510) 099-0179", "country": "Rwanda", "industry": "Financial Services", "email": "imbdn@corp.io", "job_title": "Administrative Assistant"}, {"phone": "(557) 405-6034", "country": "Switzerland", "industry": "Retail", "email": "fpfeng@example.com", "job_title": "Administrative Assistant"}, {"phone": "(570) 227-0587", "country": "Suriname", "industry": "Financial Services", "email": "blfejk@corp.io", "job_title": "Sales Representative"}, {"phone": "(597) 072-5286", "country": "Bermuda", "industry": "Financial Services", "email": "olniik@mail.com", "job_title": "Administrative Assistant"}, {"phone": "(586) 913-4188", "country": "Suriname", "industry": "Financial Services", "email": "ggogna@example.com", "job_title": "Administrative Assistant"}, {"phone": "(524) 461-4626", "country": "Rwanda", "industry": "Retail", "email": "kdfkk@corp.io", "job_title": "Administrative Assistant"}, {"phone": "(545) 655-5572", "country": "Rwanda", "industry": "Financial Services", "email": "caiggco@corp.io", "job_title": "Administrative Assistant"}, {"phone": "(537) 141-9675", "country": "Rwanda", "industry": "Energy", "email": "laahpdn@example.com", "job_title": "Sales Representative"}, {"phone": "(529) 895-6055", "country": "Switzerland", "industry": "Healthcare", "email": "jfjfd@mail.com", "job_title": "Sales Representative"}, {"phone": "(576) 403-8350", "country": "Switzerland", "industry": "Financial Services", "email": "jkao@corp.io", "job_title": "Sales Representative"}, {"phone": "(533) 876-0057", "country": "Bermuda", "industry": "Energy", "email": "hbnfcbm@corp.io", "job_title": "Sales Representative"}, {"phone": "(572) 436-4217", "country": "Suriname", "industry": "Retail", "email": "ejeg@test.org", "job_title": "Sales Representative"}, {"phone": "(513) 517-2564", "country": "Rwanda", "industry": "Energy", "email": "mnemgej@corp.io", "job_title": "Sales Representative"}, {"phone": "(502) 784-0379", "country": "Rwanda", "industry": "Healthcare", "email": "khkiejj@test.org", "job_title": "Administrative Assistant"}, {"phone": "(539) 312-3859", "country": "Bermuda", "industry": "Healthcare", "email": "mhnjclg@corp.io", "job_title": "Sales Representative"}, {"phone": "(592) 796-1101", "country": "Bermuda", "industry": "Healthcare", "email": "koio@corp.io", "job_title": "Administrative Assistant"}, {"phone": "(583) 487-5122", "country": "Suriname", "industry": "Healthcare", "email": "iedcd@mail.com", "job_title": "Sales Representative"}, {"phone": "(518) 148-4826", "country": "Rwanda", "industry": "Retail", "email": "hgdae@corp.io", "job_title": "Sales Representative"}, {"phone": "(558) 681-4132", "country": "Suriname", "industry": "Financial Services", "email": "jfldon@mail.com", "job_title": "Sales Representative"}, {"phone": "(568) 096-5624", "country": "Rwanda", "industry": "Healthcare", "email": "fieipah@example.com", "job_title": "Sales Representative"}, {"phone": "(510) 570-1873", "country": "Rwanda", "industry": "Energy", "email": "oebmfcb@mail.com", "job_title": "Sales Representative"}, {"phone": "(547) 704-7372", "country": "Bermuda", "industry": "Financial Services", "email": "lhkgio@example.com", "job_title": "Sales Representative"}, {"phone": "(534) 802-4309", "country": "Rwanda", "industry": "Energy", "email": "ebfoh@test.org", "job_title": "Sales Representative"}, {"phone": "(596) 352-9019", "country": "Switzerland", "industry": "Healthcare", "email": "eohdipi@mail.com", "job_title": "Sales Representative"}, {"phone": "(583) 212-8554", "country": "Bermuda", "industry": "Retail", "email": "lnlgah@test.org", "job_title": "Administrative Assistant"}, {"phone": "(515) 850-8351", "country": "Bermuda", "industry": "Retail", "email": "cpclbck@corp.io", "job_title": "Administrative Assistant"}], "dirty_cols": ["phone", "country", "industry", "email", "job_title"], "clean_cols": ["phone", "country", "industry", "email", "job_title"], "plan": {"dataset_summary": "28 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"swiss confederation": "Switzerland", "rw": "Rwanda", "republic of suriname": "Suriname", "RW": "Rwanda", "Rwnada": "Rwanda", "CHE": "Switzerland", "SWISS CONFEDERATION": "Switzerland", "SUR": "Suriname", "RWA": "Rwanda", "bermuda": "Bermuda", "sr": "Suriname", "rwanda": "Rwanda", "Sur": "Suriname", "BMU": "Bermuda", "bmu": "Bermuda", "Bremuda": "Bermuda"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"energy": "Energy", "finesrv": "Financial Services", "E-commerce": "Retail", "finance": "Financial Services", "Baoking": "Financial Services", "financial services": "Financial Services", "retail": "Retail", "Finance": "Financial Services", "OIL & GAS": "Energy", "Health": "Healthcare", "oil & gas": "Energy", "ecommerce": "Retail", "uitlities": "Energy", "Health Care": "Healthcare", "health care": "Healthcare", "HEALTH": "Healthcare", "Medical": "Healthcare", "Fnance": "Financial Services", "Healt Care": "Healthcare"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Admin": "Administrative Assistant", "Admin Asst": "Administrative Assistant", "rep": "Sales Representative", "ADMIN ASST": "Administrative Assistant", "admin asst": "Administrative Assistant", "Addmin": "Administrative Assistant", "Rep": "Sales Representative", "Sales Rep": "Sales Representative", "administrative assistant": "Administrative Assistant", "SALES REPRESENTATIVE": "Sales Representative", "REP": "Sales Representative", "ADMIN ASSISTANT": "Administrative Assistant", "ADMINISTRATIVE ASSISTANT": "Administrative Assistant"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "45253", "name": "Priya Petrov ", "email": "   DONA@EXAMPLE.COM", "industry": "Ed", "col_x": ""}, {"signup_date": "45146", "name": "Priya Cruz", "email": "klkelg@example.com  ", "industry": "TELO   ", "col_x": ""}, {"signup_date": "2023-08-14", "name": "Carol Adams  ", "email": "PPKHFB@MAIL.COM", "industry": "telecommuiications ", "col_x": ""}, {"signup_date": "45268", "name": "   Lena Lee", "email": "feiio@example.com", "industry": "TELCO", "col_x": ""}, {"signup_date": "8 Feb 2023", "name": "Carol  Reyes", "email": " IDBFH@MAIL.COM ", "industry": "Telecom", "col_x": ""}, {"signup_date": "2023-05-01", "name": "Judy Fischer", "email": "GBLE@EXAMPLE.COM", "industry": "EdTech", "col_x": ""}, {"signup_date": "2023-10-18", "name": "Alice Diaz", "email": "JHAGJNJ@CORP.IO", "industry": "telco", "col_x": ""}, {"signup_date": "3/16/2023", "name": "Lena Cruz", "email": "ihcdn@test.org", "industry": "Ed", "col_x": ""}, {"signup_date": "24 Jul 2023", "name": "Bob Brandt", "email": " HLFKAOE@MAIL.COM ", "industry": "education", "col_x": ""}, {"signup_date": "2023-11-02", "name": "Bob Novak", "email": "IADENO@CORP.IO", "industry": "Education", "col_x": ""}, {"signup_date": "9/16/2023", "name": "David  Fischer", "email": "pcjhip@example.com", "industry": "educltion", "col_x": ""}, {"signup_date": "12 Apr 2023", "name": "Carol Park", "email": "jjeojf@corp.io  ", "industry": "Telecommunications", "col_x": ""}, {"signup_date": "12/19/2023", "name": "Priya Smith", "email": " iinbhp@corp.io", "industry": "Telecommunications", "col_x": ""}, {"signup_date": "8 Feb 2023", "name": "Carol  Reyes", "email": " IDBFH@MAIL.COM ", "industry": "Telecom", "col_x": ""}, {"signup_date": "24 Jul 2023", "name": "Bob Brandt", "email": " HLFKAOE@MAIL.COM ", "industry": "education", "col_x": ""}, {"signup_date": "", "name": "", "email": "", "industry": "", "col_x": ""}], "clean": [{"signup_date": "2023-11-23", "name": "Priya Petrov", "email": "dona@example.com", "industry": "Education"}, {"signup_date": "2023-08-08", "name": "Priya Cruz", "email": "klkelg@example.com", "industry": "Telecommunications"}, {"signup_date": "2023-08-14", "name": "Carol Adams", "email": "ppkhfb@mail.com", "industry": "Telecommunications"}, {"signup_date": "2023-12-08", "name": "Lena Lee", "email": "feiio@example.com", "industry": "Telecommunications"}, {"signup_date": "2023-02-08", "name": "Carol Reyes", "email": "idbfh@mail.com", "industry": "Telecommunications"}, {"signup_date": "2023-05-01", "name": "Judy Fischer", "email": "gble@example.com", "industry": "Education"}, {"signup_date": "2023-10-18", "name": "Alice Diaz", "email": "jhagjnj@corp.io", "industry": "Telecommunications"}, {"signup_date": "2023-03-16", "name": "Lena Cruz", "email": "ihcdn@test.org", "industry": "Education"}, {"signup_date": "2023-07-24", "name": "Bob Brandt", "email": "hlfkaoe@mail.com", "industry": "Education"}, {"signup_date": "2023-11-02", "name": "Bob Novak", "email": "iadeno@corp.io", "industry": "Education"}, {"signup_date": "2023-09-16", "name": "David Fischer", "email": "pcjhip@example.com", "industry": "Education"}, {"signup_date": "2023-04-12", "name": "Carol Park", "email": "jjeojf@corp.io", "industry": "Telecommunications"}, {"signup_date": "2023-12-19", "name": "Priya Smith", "email": "iinbhp@corp.io", "industry": "Telecommunications"}], "dirty_cols": ["signup_date", "name", "email", "industry", "col_x"], "clean_cols": ["signup_date", "name", "email", "industry"], "plan": {"dataset_summary": "16 rows × 5 columns. 4 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ed": "Education", "TELO": "Telecommunications", "telecommuiications": "Telecommunications", "TELCO": "Telecommunications", "Telecom": "Telecommunications", "EdTech": "Education", "telco": "Telecommunications", "education": "Education", "educltion": "Education"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "-", "department": "Cs", "email": "KOIAP@TEST.ORG", "is_active": "1", "job_title": "prod mgr", "rate": "7.0%", "notes2": ""}, {"amount": "$1,342.02", "department": "IT", "email": " BAAH@MAIL.COM ", "is_active": "Y", "job_title": "   ae", "rate": "77.8%", "notes2": ""}, {"amount": "$3,722.13", "department": "Support", "email": "PKMGCN@TEST.ORG", "is_active": "0", "job_title": "  CEO", "rate": "21.3%", "notes2": ""}, {"amount": "4.465,73", "department": "operations", "email": "iohjh@corp.io", "is_active": "0", "job_title": "CEO", "rate": "89.5%", "notes2": ""}, {"amount": "2.644,57", "department": "Growth", "email": "bcpcg@test.org", "is_active": "FALSE", "job_title": "PRODUCT MANAGER", "rate": "83.5%", "notes2": ""}, {"amount": "1.464,04", "department": "Marketing  ", "email": " JLKDM@CORP.IO ", "is_active": "Yes", "job_title": "Product Manager", "rate": "75.9%", "notes2": ""}, {"amount": "3.669,29", "department": "Ops", "email": "PEHMP@MAIL.COM", "is_active": "TRUE", "job_title": "chief executive offficer", "rate": "86.2%", "notes2": ""}, {"amount": "8.649,22", "department": "it", "email": "CMPK@CORP.IO", "is_active": "false", "job_title": "Chief Executive Officer", "rate": "88.5%", "notes2": ""}, {"amount": "8.022,30", "department": "operations", "email": "EAOHBF@CORP.IO", "is_active": "N", "job_title": "Prod Manager", "rate": "11.8%", "notes2": ""}, {"amount": "$4,980.80", "department": "i.tt.", "email": " mdiejj@corp.io", "is_active": "true", "job_title": "PM", "rate": "90.2%", "notes2": ""}, {"amount": "$6,226.15", "department": "Support", "email": "hpic@example.com   ", "is_active": "1", "job_title": "Chief Executive Officer", "rate": "85.2%", "notes2": ""}, {"amount": "?", "department": " support ", "email": "bgcpia@mail.com", "is_active": "false", "job_title": "PROD MANAGER ", "rate": "89.8%", "notes2": ""}, {"amount": "8.835,60", "department": " i.. ", "email": "OLBFKJ@TEST.ORG", "is_active": "false", "job_title": "Chief Executive Officer", "rate": "5.2%", "notes2": ""}, {"amount": "$3,967.14", "department": "IT", "email": "MHGEFLI@MAIL.COM", "is_active": "Yes", "job_title": "Acct Exec", "rate": "92.8%", "notes2": ""}, {"amount": "$6,456.26", "department": "information technology", "email": "nkoemmm@test.org", "is_active": "No", "job_title": "AE", "rate": "38.1%", "notes2": ""}, {"amount": "5.806,05", "department": "   Operations", "email": "PFNM@CORP.IO", "is_active": "Y", "job_title": "PM", "rate": "91.6%", "notes2": ""}, {"amount": "4.959,38", "department": "CS", "email": " HKCOJP@MAIL.COM ", "is_active": "FALSE", "job_title": "  chief executive officer", "rate": "93.2%", "notes2": ""}, {"amount": "9999999", "department": "CUSTOMER SUPPORT", "email": "ECMN@MAIL.COM", "is_active": "Y", "job_title": "Chiief Executive Officer", "rate": "85.2%", "notes2": ""}, {"amount": "N/A", "department": "i.t.", "email": " GKME@TEST.ORG", "is_active": "false", "job_title": "CEO", "rate": "93.0%", "notes2": ""}, {"amount": "$6,487.65", "department": "CS", "email": "njja@example.com", "is_active": "1", "job_title": "   PM", "rate": "92.6%", "notes2": ""}, {"amount": "$3,423.24", "department": "maketing", "email": "MNHFB@MAIL.COM", "is_active": "TRUE", "job_title": "CEO", "rate": "39.7%", "notes2": ""}, {"amount": "6.357,12", "department": "SUPPORT", "email": "APKPK@TEST.ORG", "is_active": "FALSE", "job_title": "PM", "rate": "36.2%", "notes2": ""}, {"amount": "$1,182.90", "department": "Cust Support", "email": "   JCAA@MAIL.COM", "is_active": "T", "job_title": "acct exec", "rate": "6.1%", "notes2": ""}, {"amount": "7.691,09", "department": "OPERATIONS", "email": "HMNBKMD@MAIL.COM", "is_active": "TRUE", "job_title": " CEO ", "rate": "19.4%", "notes2": ""}, {"amount": "?", "department": "I.T.  ", "email": "gnffgg@example.com", "is_active": "Y", "job_title": "account executive", "rate": "18.0%", "notes2": ""}, {"amount": "?", "department": "OPREATIONS", "email": "mndcf@example.com", "is_active": "true", "job_title": "prod manager", "rate": "33.6%", "notes2": ""}, {"amount": "3.975,54", "department": " mkt ", "email": " ggbbj@test.org", "is_active": "0", "job_title": "CEO", "rate": "87.0%", "notes2": ""}, {"amount": "$1,615.96", "department": "IT", "email": "okoe@example.com", "is_active": "Yes", "job_title": "Acct  Exec", "rate": "92.1%", "notes2": ""}, {"amount": "$4,403.32", "department": "Ops", "email": "INGGKJG@CORP.IO", "is_active": "true", "job_title": " account executive", "rate": "41.6%", "notes2": ""}, {"amount": "$1,834.90", "department": "Gromth  ", "email": "nnejln@example.com   ", "is_active": "1", "job_title": "Chief Executive Officer", "rate": "96.5%", "notes2": ""}, {"amount": "$3,123.48", "department": "CS  ", "email": "  nhkeh@mail.com", "is_active": "true", "job_title": "   CEO", "rate": "97.1%", "notes2": ""}, {"amount": "$5,898.43", "department": " ops ", "email": "DMDA@TEST.ORG  ", "is_active": "T", "job_title": "acct exec", "rate": "36.7%", "notes2": ""}, {"amount": "8.643,47", "department": "cust support", "email": " GKBNJ@CORP.IO ", "is_active": "false", "job_title": "CEO", "rate": "7.6%", "notes2": ""}, {"amount": "$3,347.74", "department": "Ops", "email": "nclok@mail.com ", "is_active": "Yes", "job_title": "C.E.O.", "rate": "94.5%", "notes2": ""}, {"amount": "N/A", "department": "IT", "email": "BILNP@TEST.ORG  ", "is_active": "false", "job_title": "acct exec", "rate": "92.0%", "notes2": ""}, {"amount": "8.592,29", "department": "Cust Support", "email": "OPIO@EXAMPLE.COM", "is_active": "FALSE", "job_title": "pm", "rate": "36.7%", "notes2": ""}, {"amount": "7.184,95", "department": "CS", "email": "PCMIKI@MAIL.COM", "is_active": "false", "job_title": "CEO", "rate": "39.8%", "notes2": ""}, {"amount": "$3,722.13", "department": "Support", "email": "PKMGCN@TEST.ORG", "is_active": "0", "job_title": "  CEO", "rate": "21.3%", "notes2": ""}, {"amount": "8.643,47", "department": "cust support", "email": " GKBNJ@CORP.IO ", "is_active": "false", "job_title": "CEO", "rate": "7.6%", "notes2": ""}], "clean": [{"amount": NaN, "department": "Customer Support", "email": "koiap@test.org", "is_active": true, "job_title": "Product Manager", "rate": 0.07}, {"amount": 1342.02, "department": "Information Technology", "email": "baah@mail.com", "is_active": true, "job_title": "Account Executive", "rate": 0.778}, {"amount": 3722.13, "department": "Customer Support", "email": "pkmgcn@test.org", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.213}, {"amount": 4465.73, "department": "Operations", "email": "iohjh@corp.io", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.895}, {"amount": 2644.57, "department": "Marketing", "email": "bcpcg@test.org", "is_active": false, "job_title": "Product Manager", "rate": 0.835}, {"amount": 1464.04, "department": "Marketing", "email": "jlkdm@corp.io", "is_active": true, "job_title": "Product Manager", "rate": 0.759}, {"amount": 3669.29, "department": "Operations", "email": "pehmp@mail.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.862}, {"amount": 8649.22, "department": "Information Technology", "email": "cmpk@corp.io", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.885}, {"amount": 8022.3, "department": "Operations", "email": "eaohbf@corp.io", "is_active": false, "job_title": "Product Manager", "rate": 0.11800000000000001}, {"amount": 4980.8, "department": "Information Technology", "email": "mdiejj@corp.io", "is_active": true, "job_title": "Product Manager", "rate": 0.902}, {"amount": 6226.15, "department": "Customer Support", "email": "hpic@example.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.852}, {"amount": NaN, "department": "Customer Support", "email": "bgcpia@mail.com", "is_active": false, "job_title": "Product Manager", "rate": 0.898}, {"amount": 8835.6, "department": "Information Technology", "email": "olbfkj@test.org", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.052000000000000005}, {"amount": 3967.14, "department": "Information Technology", "email": "mhgefli@mail.com", "is_active": true, "job_title": "Account Executive", "rate": 0.9279999999999999}, {"amount": 6456.26, "department": "Information Technology", "email": "nkoemmm@test.org", "is_active": false, "job_title": "Account Executive", "rate": 0.381}, {"amount": 5806.05, "department": "Operations", "email": "pfnm@corp.io", "is_active": true, "job_title": "Product Manager", "rate": 0.9159999999999999}, {"amount": 4959.38, "department": "Customer Support", "email": "hkcojp@mail.com", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.932}, {"amount": 9999999.0, "department": "Customer Support", "email": "ecmn@mail.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.852}, {"amount": NaN, "department": "Information Technology", "email": "gkme@test.org", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.93}, {"amount": 6487.65, "department": "Customer Support", "email": "njja@example.com", "is_active": true, "job_title": "Product Manager", "rate": 0.9259999999999999}, {"amount": 3423.24, "department": "Marketing", "email": "mnhfb@mail.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.397}, {"amount": 6357.12, "department": "Customer Support", "email": "apkpk@test.org", "is_active": false, "job_title": "Product Manager", "rate": 0.36200000000000004}, {"amount": 1182.9, "department": "Customer Support", "email": "jcaa@mail.com", "is_active": true, "job_title": "Account Executive", "rate": 0.061}, {"amount": 7691.09, "department": "Operations", "email": "hmnbkmd@mail.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.19399999999999998}, {"amount": NaN, "department": "Information Technology", "email": "gnffgg@example.com", "is_active": true, "job_title": "Account Executive", "rate": 0.18}, {"amount": NaN, "department": "Operations", "email": "mndcf@example.com", "is_active": true, "job_title": "Product Manager", "rate": 0.336}, {"amount": 3975.54, "department": "Marketing", "email": "ggbbj@test.org", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.87}, {"amount": 1615.96, "department": "Information Technology", "email": "okoe@example.com", "is_active": true, "job_title": "Account Executive", "rate": 0.9209999999999999}, {"amount": 4403.32, "department": "Operations", "email": "inggkjg@corp.io", "is_active": true, "job_title": "Account Executive", "rate": 0.41600000000000004}, {"amount": 1834.9, "department": "Marketing", "email": "nnejln@example.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.965}, {"amount": 3123.48, "department": "Customer Support", "email": "nhkeh@mail.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.971}, {"amount": 5898.43, "department": "Operations", "email": "dmda@test.org", "is_active": true, "job_title": "Account Executive", "rate": 0.36700000000000005}, {"amount": 8643.47, "department": "Customer Support", "email": "gkbnj@corp.io", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.076}, {"amount": 3347.74, "department": "Operations", "email": "nclok@mail.com", "is_active": true, "job_title": "Chief Executive Officer", "rate": 0.945}, {"amount": NaN, "department": "Information Technology", "email": "bilnp@test.org", "is_active": false, "job_title": "Account Executive", "rate": 0.92}, {"amount": 8592.29, "department": "Customer Support", "email": "opio@example.com", "is_active": false, "job_title": "Product Manager", "rate": 0.36700000000000005}, {"amount": 7184.95, "department": "Customer Support", "email": "pcmiki@mail.com", "is_active": false, "job_title": "Chief Executive Officer", "rate": 0.39799999999999996}], "dirty_cols": ["amount", "department", "email", "is_active", "job_title", "rate", "notes2"], "clean_cols": ["amount", "department", "email", "is_active", "job_title", "rate"], "plan": {"dataset_summary": "39 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cs": "Customer Support", "IT": "Information Technology", "Support": "Customer Support", "operations": "Operations", "Growth": "Marketing", "Ops": "Operations", "it": "Information Technology", "i.tt.": "Information Technology", "support": "Customer Support", "i..": "Information Technology", "information technology": "Information Technology", "CS": "Customer Support", "CUSTOMER SUPPORT": "Customer Support", "i.t.": "Information Technology", "maketing": "Marketing", "SUPPORT": "Customer Support", "Cust Support": "Customer Support", "OPERATIONS": "Operations", "I.T.": "Information Technology", "OPREATIONS": "Operations", "mkt": "Marketing", "Gromth": "Marketing", "ops": "Operations", "cust support": "Customer Support"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"prod mgr": "Product Manager", "ae": "Account Executive", "CEO": "Chief Executive Officer", "PRODUCT MANAGER": "Product Manager", "chief executive offficer": "Chief Executive Officer", "Prod Manager": "Product Manager", "PM": "Product Manager", "PROD MANAGER": "Product Manager", "Acct Exec": "Account Executive", "AE": "Account Executive", "chief executive officer": "Chief Executive Officer", "Chiief Executive Officer": "Chief Executive Officer", "acct exec": "Account Executive", "account executive": "Account Executive", "prod manager": "Product Manager", "C.E.O.": "Chief Executive Officer", "pm": "Product Manager"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"country": "kw", "company": "Stark Industries ", "amount": "#N/A", "city": "Stargard", "signup_date": "2023-04-04", "state": "Iowa", "currency": "Quetzal", "extra": ""}, {"country": "WLF", "company": "Cyberdyne", "amount": "$7,606.44", "city": "Stargard", "signup_date": "2023-07-18", "state": "Id", "currency": "GTQ", "extra": ""}, {"country": "kwt", "company": "Cyberdyne", "amount": "8.170,55", "city": " Mauldin ", "signup_date": "16 Jul 2023", "state": "Idaho", "currency": "Brazilian Real", "extra": ""}, {"country": " RS", "company": "Cyberdyne", "amount": "$237.70", "city": "KENDARI", "signup_date": "44994", "state": "idaho", "currency": "GTQ", "extra": ""}, {"country": "SRB", "company": "Soylent Corp  ", "amount": "6.647,97", "city": "   kendari", "signup_date": "2023-03-25", "state": "ia", "currency": "Quetzal", "extra": ""}, {"country": "STATE OF OUWAIT", "company": " Soylent Corp", "amount": "$2,809.88", "city": "Mauldin", "signup_date": "2023-11-04", "state": "SOUTH CAROLINA  ", "currency": "GTQ", "extra": ""}, {"country": "STATE OF KUWAIT", "company": " Umbrella ", "amount": "4.233,76", "city": "Foothll Farms", "signup_date": "45285", "state": "Idbho", "currency": "Brazilian Real", "extra": ""}, {"country": "state of kuwait", "company": "Wonka Co", "amount": "1.739,34", "city": "irosin", "signup_date": "11/8/2023", "state": "south carolina", "currency": "gtq", "extra": ""}, {"country": "Serbia", "company": "Umbrella", "amount": "599,97", "city": "stargard", "signup_date": "2023-07-14", "state": "ia", "currency": "BRL", "extra": ""}, {"country": "ALB", "company": "Initech", "amount": "$5,045.58", "city": "mauldin", "signup_date": "45017", "state": "IA", "currency": " Quetzal", "extra": ""}, {"country": "kuwait", "company": "   Initech", "amount": "$7,619.78", "city": "  Irosin", "signup_date": "3/15/2023", "state": "ID", "currency": " Quetzal", "extra": ""}, {"country": "Al  ", "company": " Soylent Corp", "amount": "$574.41", "city": "Irosin", "signup_date": "2023-08-10", "state": " idaho", "currency": "BRL   ", "extra": ""}, {"country": "state of kuwait  ", "company": "Soylent Corp", "amount": "$4,513.94", "city": "Stargard", "signup_date": "9 Sep 2023", "state": "South Carolina", "currency": " quetzal ", "extra": ""}, {"country": " Kwt", "company": "Globex", "amount": "8.623,23", "city": "kendari", "signup_date": "11/10/2023", "state": "ia", "currency": "Brazilian Real", "extra": ""}, {"country": "WLF", "company": "   Acme Inc", "amount": "6.437,04", "city": "Sttrgard", "signup_date": "45026", "state": " SC", "currency": " Quetzal ", "extra": ""}, {"country": "kwt  ", "company": "Acme Inc", "amount": "$1,187.25", "city": "kendari", "signup_date": "2023-04-22", "state": " soouth carolina", "currency": "   BRL", "extra": ""}, {"country": "Kwt", "company": "   Stark Industries", "amount": "$5,008.89", "city": "STARGARD", "signup_date": "2023-07-28", "state": "South Carolina", "currency": "BRL", "extra": ""}, {"country": "kuwati", "company": "Vehement", "amount": "$5,867.88", "city": "Mauldin", "signup_date": "9 Jan 2023", "state": "South  Carolina", "currency": "GTQ", "extra": ""}, {"country": "  Kuwait", "company": "  Initech", "amount": "$2,738.27", "city": "iorsin", "signup_date": "44962", "state": "Idaho", "currency": "QUETZAL", "extra": ""}, {"country": " kuwait ", "company": "Umbrella", "amount": "$2,853.50", "city": "Mauldin", "signup_date": "4/26/2023", "state": "Ia   ", "currency": " Brl ", "extra": ""}, {"country": "kwt", "company": " Vehement ", "amount": "$3,463.58", "city": "Starrgard", "signup_date": "9/6/2023", "state": "idaho", "currency": "brl", "extra": ""}, {"country": "WALLIS AND FUTUNA", "company": " Globex ", "amount": "3.278,01", "city": "maludin", "signup_date": "6/11/2023", "state": "SC", "currency": "brl", "extra": ""}, {"country": "WLF", "company": "Globex", "amount": "$7,930.92", "city": "Irosin", "signup_date": "18 Apr 2023", "state": "SC", "currency": "Quetzal", "extra": ""}, {"country": "STATE OF KUWAIT", "company": "Wonka Co", "amount": "$1,752.90", "city": "Irosin", "signup_date": "22 Sep 2023", "state": " ID", "currency": "BRL", "extra": ""}, {"country": "Republic of Serbia", "company": "Acme Inc", "amount": "6.383,89", "city": "FOOTHILL FARMS", "signup_date": "2023-06-18", "state": "Idaho", "currency": "BRL   ", "extra": ""}, {"country": " SRB ", "company": " Hooli ", "amount": "2.629,64", "city": "Iroscn", "signup_date": "14 Jun 2023", "state": "ID", "currency": "QUETZAL", "extra": ""}, {"country": " ALB ", "company": "Vehement ", "amount": "4.024,04", "city": "Starard", "signup_date": "45080", "state": "IA", "currency": " BRL ", "extra": ""}, {"country": "ALB  ", "company": "Acme Inc", "amount": "5.863,47", "city": "Mauldin", "signup_date": "2023-07-03", "state": "ID", "currency": "Brazilian Real", "extra": ""}, {"country": "  Wf", "company": "Stark Industries", "amount": "TBD", "city": "stargard", "signup_date": "1/2/2023", "state": "ia", "currency": "GTQ", "extra": ""}, {"country": "alb  ", "company": "Stark Industries", "amount": "$313.28", "city": " kendari ", "signup_date": "1/22/2023", "state": "ID", "currency": "Brazilian Real", "extra": ""}, {"country": "Stateo f Kuwait", "company": "Wonka Co", "amount": "-", "city": "IROSIN", "signup_date": "44987", "state": "iowa", "currency": "brl", "extra": ""}, {"country": "kuwait", "company": "Umbrella", "amount": "8.587,09", "city": "IROSIN", "signup_date": "1/6/2023", "state": "SC", "currency": " QUETZAL", "extra": ""}, {"country": "kw", "company": "Globex", "amount": "0", "city": "Stargard", "signup_date": "4 Aug 2023", "state": "Iowa", "currency": "Brazilian Real", "extra": ""}, {"country": "Wallis and Futuna", "company": "Acme Inc", "amount": "$5,743.84", "city": "KENDARI", "signup_date": "6/27/2023", "state": "IA", "currency": "GTQ", "extra": ""}, {"country": "Republic of Albania  ", "company": "Hooli", "amount": "1.152,79", "city": "irosin", "signup_date": "2023-09-27", "state": "SC", "currency": "quetzal", "extra": ""}, {"country": " wlf ", "company": "Acme Inc", "amount": "5.726,86", "city": "Stargard", "signup_date": "7/12/2023", "state": "ia", "currency": "Quetzal", "extra": ""}, {"country": "AL", "company": "Soylent Corp", "amount": "$2,481.06", "city": "foothill farms   ", "signup_date": "2023-01-21", "state": " idaho ", "currency": "Quetzal", "extra": ""}, {"country": "Republic Of Albania", "company": " Initech ", "amount": "4.138,07", "city": "Irosin", "signup_date": "6 Nov 2023", "state": "ID", "currency": "Brazilian Real", "extra": ""}, {"country": "Kwt   ", "company": " Umbrella", "amount": "8.988,45", "city": "  Foothill Farms", "signup_date": "18 May 2023", "state": "SC", "currency": " Quetzal", "extra": ""}, {"country": "srb", "company": "Umbrella", "amount": "7.600,93", "city": "KENWARI", "signup_date": "4/8/2023", "state": "IOWA", "currency": "Brazilian Real", "extra": ""}, {"country": "KW", "company": " Initech", "amount": "2.579,70", "city": "Irosin ", "signup_date": "6/19/2023", "state": "ia", "currency": "quetzal", "extra": ""}, {"country": "Wallis and Futuna", "company": "Hooli", "amount": "7.455,15", "city": "irosin", "signup_date": "7/11/2023", "state": " Iowa ", "currency": "Quetzal", "extra": ""}, {"country": "ALB", "company": "Soylent Corp", "amount": "$2,646.51", "city": "Mauldin", "signup_date": "24 Nov 2023", "state": " id", "currency": "BRL", "extra": ""}, {"country": "ALB", "company": "Stark Industries", "amount": "8.206,93", "city": "staragrd", "signup_date": "3/24/2023", "state": "Iwa", "currency": "BRL", "extra": ""}, {"country": "KW", "company": "Hooli", "amount": "3.860,42", "city": "stargard", "signup_date": "19 Dec 2023", "state": "iowa", "currency": "brazilian real", "extra": ""}, {"country": "ALB", "company": "Initech", "amount": "$5,045.58", "city": "mauldin", "signup_date": "45017", "state": "IA", "currency": " Quetzal", "extra": ""}], "clean": [{"country": "Kuwait", "company": "Stark Industries", "amount": NaN, "city": "Stargard", "signup_date": "2023-04-04", "state": "Iowa", "currency": "GTQ"}, {"country": "Wallis and Futuna", "company": "Cyberdyne", "amount": 7606.44, "city": "Stargard", "signup_date": "2023-07-18", "state": "Idaho", "currency": "GTQ"}, {"country": "Kuwait", "company": "Cyberdyne", "amount": 8170.55, "city": "Mauldin", "signup_date": "2023-07-16", "state": "Idaho", "currency": "BRL"}, {"country": "Serbia", "company": "Cyberdyne", "amount": 237.7, "city": "Kendari", "signup_date": "2023-03-09", "state": "Idaho", "currency": "GTQ"}, {"country": "Serbia", "company": "Soylent Corp", "amount": 6647.97, "city": "Kendari", "signup_date": "2023-03-25", "state": "Iowa", "currency": "GTQ"}, {"country": "Kuwait", "company": "Soylent Corp", "amount": 2809.88, "city": "Mauldin", "signup_date": "2023-11-04", "state": "South Carolina", "currency": "GTQ"}, {"country": "Kuwait", "company": "Umbrella", "amount": 4233.76, "city": "Foothill Farms", "signup_date": "2023-12-25", "state": "Idaho", "currency": "BRL"}, {"country": "Kuwait", "company": "Wonka Co", "amount": 1739.34, "city": "Irosin", "signup_date": "2023-11-08", "state": "South Carolina", "currency": "GTQ"}, {"country": "Serbia", "company": "Umbrella", "amount": 599.97, "city": "Stargard", "signup_date": "2023-07-14", "state": "Iowa", "currency": "BRL"}, {"country": "Albania", "company": "Initech", "amount": 5045.58, "city": "Mauldin", "signup_date": "2023-04-01", "state": "Iowa", "currency": "GTQ"}, {"country": "Kuwait", "company": "Initech", "amount": 7619.78, "city": "Irosin", "signup_date": "2023-03-15", "state": "Idaho", "currency": "GTQ"}, {"country": "Albania", "company": "Soylent Corp", "amount": 574.41, "city": "Irosin", "signup_date": "2023-08-10", "state": "Idaho", "currency": "BRL"}, {"country": "Kuwait", "company": "Soylent Corp", "amount": 4513.94, "city": "Stargard", "signup_date": "2023-09-09", "state": "South Carolina", "currency": "GTQ"}, {"country": "Kuwait", "company": "Globex", "amount": 8623.23, "city": "Kendari", "signup_date": "2023-11-10", "state": "Iowa", "currency": "BRL"}, {"country": "Wallis and Futuna", "company": "Acme Inc", "amount": 6437.04, "city": "Stargard", "signup_date": "2023-04-10", "state": "South Carolina", "currency": "GTQ"}, {"country": "Kuwait", "company": "Acme Inc", "amount": 1187.25, "city": "Kendari", "signup_date": "2023-04-22", "state": "South Carolina", "currency": "BRL"}, {"country": "Kuwait", "company": "Stark Industries", "amount": 5008.89, "city": "Stargard", "signup_date": "2023-07-28", "state": "South Carolina", "currency": "BRL"}, {"country": "Kuwait", "company": "Vehement", "amount": 5867.88, "city": "Mauldin", "signup_date": "2023-01-09", "state": "South Carolina", "currency": "GTQ"}, {"country": "Kuwait", "company": "Initech", "amount": 2738.27, "city": "Irosin", "signup_date": "2023-02-05", "state": "Idaho", "currency": "GTQ"}, {"country": "Kuwait", "company": "Umbrella", "amount": 2853.5, "city": "Mauldin", "signup_date": "2023-04-26", "state": "Iowa", "currency": "BRL"}, {"country": "Kuwait", "company": "Vehement", "amount": 3463.58, "city": "Stargard", "signup_date": "2023-09-06", "state": "Idaho", "currency": "BRL"}, {"country": "Wallis and Futuna", "company": "Globex", "amount": 3278.01, "city": "Mauldin", "signup_date": "2023-06-11", "state": "South Carolina", "currency": "BRL"}, {"country": "Wallis and Futuna", "company": "Globex", "amount": 7930.92, "city": "Irosin", "signup_date": "2023-04-18", "state": "South Carolina", "currency": "GTQ"}, {"country": "Kuwait", "company": "Wonka Co", "amount": 1752.9, "city": "Irosin", "signup_date": "2023-09-22", "state": "Idaho", "currency": "BRL"}, {"country": "Serbia", "company": "Acme Inc", "amount": 6383.89, "city": "Foothill Farms", "signup_date": "2023-06-18", "state": "Idaho", "currency": "BRL"}, {"country": "Serbia", "company": "Hooli", "amount": 2629.64, "city": "Irosin", "signup_date": "2023-06-14", "state": "Idaho", "currency": "GTQ"}, {"country": "Albania", "company": "Vehement", "amount": 4024.04, "city": "Stargard", "signup_date": "2023-06-03", "state": "Iowa", "currency": "BRL"}, {"country": "Albania", "company": "Acme Inc", "amount": 5863.47, "city": "Mauldin", "signup_date": "2023-07-03", "state": "Idaho", "currency": "BRL"}, {"country": "Wallis and Futuna", "company": "Stark Industries", "amount": NaN, "city": "Stargard", "signup_date": "2023-01-02", "state": "Iowa", "currency": "GTQ"}, {"country": "Albania", "company": "Stark Industries", "amount": 313.28, "city": "Kendari", "signup_date": "2023-01-22", "state": "Idaho", "currency": "BRL"}, {"country": "Kuwait", "company": "Wonka Co", "amount": NaN, "city": "Irosin", "signup_date": "2023-03-02", "state": "Iowa", "currency": "BRL"}, {"country": "Kuwait", "company": "Umbrella", "amount": 8587.09, "city": "Irosin", "signup_date": "2023-01-06", "state": "South Carolina", "currency": "GTQ"}, {"country": "Kuwait", "company": "Globex", "amount": 0.0, "city": "Stargard", "signup_date": "2023-08-04", "state": "Iowa", "currency": "BRL"}, {"country": "Wallis and Futuna", "company": "Acme Inc", "amount": 5743.84, "city": "Kendari", "signup_date": "2023-06-27", "state": "Iowa", "currency": "GTQ"}, {"country": "Albania", "company": "Hooli", "amount": 1152.79, "city": "Irosin", "signup_date": "2023-09-27", "state": "South Carolina", "currency": "GTQ"}, {"country": "Wallis and Futuna", "company": "Acme Inc", "amount": 5726.86, "city": "Stargard", "signup_date": "2023-07-12", "state": "Iowa", "currency": "GTQ"}, {"country": "Albania", "company": "Soylent Corp", "amount": 2481.06, "city": "Foothill Farms", "signup_date": "2023-01-21", "state": "Idaho", "currency": "GTQ"}, {"country": "Albania", "company": "Initech", "amount": 4138.07, "city": "Irosin", "signup_date": "2023-11-06", "state": "Idaho", "currency": "BRL"}, {"country": "Kuwait", "company": "Umbrella", "amount": 8988.45, "city": "Foothill Farms", "signup_date": "2023-05-18", "state": "South Carolina", "currency": "GTQ"}, {"country": "Serbia", "company": "Umbrella", "amount": 7600.93, "city": "Kendari", "signup_date": "2023-04-08", "state": "Iowa", "currency": "BRL"}, {"country": "Kuwait", "company": "Initech", "amount": 2579.7, "city": "Irosin", "signup_date": "2023-06-19", "state": "Iowa", "currency": "GTQ"}, {"country": "Wallis and Futuna", "company": "Hooli", "amount": 7455.15, "city": "Irosin", "signup_date": "2023-07-11", "state": "Iowa", "currency": "GTQ"}, {"country": "Albania", "company": "Soylent Corp", "amount": 2646.51, "city": "Mauldin", "signup_date": "2023-11-24", "state": "Idaho", "currency": "BRL"}, {"country": "Albania", "company": "Stark Industries", "amount": 8206.93, "city": "Stargard", "signup_date": "2023-03-24", "state": "Iowa", "currency": "BRL"}, {"country": "Kuwait", "company": "Hooli", "amount": 3860.42, "city": "Stargard", "signup_date": "2023-12-19", "state": "Iowa", "currency": "BRL"}], "dirty_cols": ["country", "company", "amount", "city", "signup_date", "state", "currency", "extra"], "clean_cols": ["country", "company", "amount", "city", "signup_date", "state", "currency"], "plan": {"dataset_summary": "46 rows × 8 columns. 7 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kw": "Kuwait", "WLF": "Wallis and Futuna", "kwt": "Kuwait", "RS": "Serbia", "SRB": "Serbia", "STATE OF OUWAIT": "Kuwait", "STATE OF KUWAIT": "Kuwait", "state of kuwait": "Kuwait", "ALB": "Albania", "kuwait": "Kuwait", "Al": "Albania", "Kwt": "Kuwait", "kuwati": "Kuwait", "WALLIS AND FUTUNA": "Wallis and Futuna", "Republic of Serbia": "Serbia", "Wf": "Wallis and Futuna", "alb": "Albania", "Stateo f Kuwait": "Kuwait", "Republic of Albania": "Albania", "wlf": "Wallis and Futuna", "AL": "Albania", "Republic Of Albania": "Albania", "srb": "Serbia", "KW": "Kuwait"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KENDARI": "Kendari", "kendari": "Kendari", "Foothll Farms": "Foothill Farms", "irosin": "Irosin", "stargard": "Stargard", "mauldin": "Mauldin", "Sttrgard": "Stargard", "STARGARD": "Stargard", "iorsin": "Irosin", "Starrgard": "Stargard", "maludin": "Mauldin", "FOOTHILL FARMS": "Foothill Farms", "Iroscn": "Irosin", "Starard": "Stargard", "IROSIN": "Irosin", "foothill farms": "Foothill Farms", "KENWARI": "Kendari", "staragrd": "Stargard"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Id": "Idaho", "idaho": "Idaho", "ia": "Iowa", "SOUTH CAROLINA": "South Carolina", "Idbho": "Idaho", "south carolina": "South Carolina", "IA": "Iowa", "ID": "Idaho", "SC": "South Carolina", "soouth carolina": "South Carolina", "Ia": "Iowa", "iowa": "Iowa", "IOWA": "Iowa", "id": "Idaho", "Iwa": "Iowa"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Quetzal": "GTQ", "Brazilian Real": "BRL", "gtq": "GTQ", "quetzal": "GTQ", "QUETZAL": "GTQ", "Brl": "BRL", "brl": "BRL", "brazilian real": "BRL"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"signup_date": "22 Feb 2023", "phone": "5824021566", "state": "South Dakota", "city": "Mojokerto", "currency": "COU", "unit": " kg ", "name": "Carol Brandt", "notes2": ""}, {"signup_date": "2023-07-02", "phone": "(549)975-5188", "state": "SD", "city": "mojokerto", "currency": "GIP", "unit": " kilogram ", "name": "Bob Khan", "notes2": ""}, {"signup_date": "4/24/2023", "phone": "564-225-5805", "state": "SD", "city": "TENKODOGO", "currency": "cou", "unit": "kilogram", "name": "Omar Brandt", "notes2": ""}, {"signup_date": "45263", "phone": "573-087-2847", "state": "OH", "city": "MOJOKERTO", "currency": "DKK", "unit": "kilograms", "name": "Carol Diaz ", "notes2": ""}, {"signup_date": "2023-09-16", "phone": "(526)836-5557", "state": "sd", "city": "Bloemhof", "currency": "Unidad de Valor Real", "unit": "kilometers", "name": "Mona  Lee", "notes2": ""}, {"signup_date": "10 Mar 2023", "phone": "5001874155", "state": "ohio", "city": "Bloemhof", "currency": "Unidad de Valor Real", "unit": "kilograms", "name": "Karl Moore", "notes2": ""}, {"signup_date": "45163", "phone": "5874545761", "state": "ohio", "city": "Bukit Panjang New Town", "currency": "Unidad de Valor Real", "unit": "Kilogram", "name": "Ivan Diaz", "notes2": ""}, {"signup_date": "4/1/2023", "phone": "5458836810", "state": "SD", "city": "Mojokerto", "currency": "gibraltar pound", "unit": "kg", "name": "Alice Park ", "notes2": ""}, {"signup_date": "45081", "phone": "515.526.9141", "state": "OH", "city": "BUKIT PANJANG NEW TOWN", "currency": "unidad de valor real", "unit": "Km  ", "name": "Carol Johnson", "notes2": ""}, {"signup_date": "1/8/2023", "phone": "5875876110", "state": "Florida", "city": "ZAWIERCIE", "currency": "GIP", "unit": "Kg", "name": "Heidi Fischer", "notes2": ""}, {"signup_date": "45102", "phone": "(524)374-5472", "state": "ohio", "city": "Zawizrcie", "currency": " DKK ", "unit": "kilogram", "name": "  Eve Adams", "notes2": ""}, {"signup_date": "2023-11-01", "phone": "564-729-8017", "state": "Fl", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Priya Novak  ", "notes2": ""}, {"signup_date": "17 Oct 2023", "phone": "(537)326-6607", "state": "South Dakota", "city": "Tenkodogo", "currency": "danish krone", "unit": "kg", "name": "Mona  Lee", "notes2": ""}, {"signup_date": "11/1/2023", "phone": "(506)349-3657", "state": " florida ", "city": "tenkodogo", "currency": "Danish  Krone", "unit": "kilogram  ", "name": "Ivan  Lee", "notes2": ""}, {"signup_date": "11/22/2023", "phone": "596-159-1803", "state": "sd", "city": "Bukit  Panjang New Town", "currency": "GIP", "unit": "kilometer", "name": "Carol Park", "notes2": ""}, {"signup_date": "45008", "phone": "5272535360", "state": "  Ohoi", "city": "zawiercie", "currency": "  DKK", "unit": "KILOGRAMS", "name": "  Eve Fischer", "notes2": ""}, {"signup_date": "44935", "phone": "565.978.3907", "state": " FL", "city": "  BUKIT PANJANG NEW TOWN", "currency": "gip", "unit": "   km", "name": "   Priya Cruz", "notes2": ""}, {"signup_date": "20 Jan 2023", "phone": "542.686.6999", "state": "Ohio   ", "city": "Tenkodogo", "currency": "Gibraltar Pound", "unit": "kg", "name": "Karl Diaz", "notes2": ""}, {"signup_date": "2023-07-11", "phone": "5445295271", "state": "OH", "city": "MOJOKERTO", "currency": "GIP   ", "unit": "kg", "name": "Mona Ali", "notes2": ""}, {"signup_date": "2023-12-07", "phone": "508.449.8443", "state": "fl", "city": "Zawieruie", "currency": "  danish krone", "unit": "  Kg", "name": "Judy Johnson", "notes2": ""}, {"signup_date": "17 Feb 2023", "phone": "(553)389-4169", "state": "florida", "city": "Bloemhof", "currency": "dkk", "unit": "kg ", "name": "Karl Lee ", "notes2": ""}, {"signup_date": "44937", "phone": "5935394252", "state": "OH   ", "city": "bloemhof", "currency": "gip", "unit": "kmlometer", "name": "Omar  Ali", "notes2": ""}, {"signup_date": "2023-12-09", "phone": "543.210.8986", "state": "Sd", "city": "tenkodogo", "currency": "GIBRALTAR POUND", "unit": "km", "name": "Mona Moore", "notes2": ""}, {"signup_date": "12 Jun 2023", "phone": "5701290162", "state": "  OH", "city": "Zawierciie", "currency": "DKK", "unit": "kilometer", "name": "David Wong   ", "notes2": ""}, {"signup_date": "2023-09-07", "phone": "527-423-4271", "state": "OH", "city": "tenkodogo  ", "currency": "COU ", "unit": " kg", "name": "Frank Khan", "notes2": ""}, {"signup_date": "44994", "phone": "(535)315-0100", "state": "oh", "city": "tenkodogo", "currency": "Unidad De Valor Real", "unit": "KILOMETERS", "name": "Karl Cruz", "notes2": ""}, {"signup_date": "1/17/2023", "phone": "(563)610-2753", "state": "oh", "city": "TENKODOGO", "currency": "GIBRALTAR POUND", "unit": "kg", "name": "Lena Fischer  ", "notes2": ""}, {"signup_date": "26 Apr 2023", "phone": "5715950928", "state": "Oh", "city": "Tenkodogo", "currency": "dannish krone  ", "unit": "kilogram", "name": "Carol Fischer   ", "notes2": ""}, {"signup_date": "45285", "phone": "(534)073-0650", "state": "Ohio", "city": "bloemhof", "currency": "Gibraltar Pound", "unit": "kilometers", "name": " Eve Petrov", "notes2": ""}, {"signup_date": "18 Sep 2023", "phone": "5588595669", "state": "OH", "city": "Tenkodogo", "currency": "DKK", "unit": " kilometers ", "name": "Frank Cruz", "notes2": ""}, {"signup_date": "5 Sep 2023", "phone": "5881514953", "state": "FL", "city": "Tenkodogo", "currency": " DKK", "unit": "KM", "name": "Heidi Adams", "notes2": ""}, {"signup_date": "7/28/2023", "phone": "(575)744-9635", "state": "FL", "city": "BUKIT PANJANG NEW TOWN", "currency": "Gibraltar Pound", "unit": "KG", "name": "Bob Diaz", "notes2": ""}, {"signup_date": "45241", "phone": "(556)062-9095", "state": "FL", "city": "Zawiercie", "currency": "   GIBRALTAR POUND", "unit": "killogram", "name": "Ivan Brandt", "notes2": ""}, {"signup_date": "3/24/2023", "phone": "5196358556", "state": "ohio", "city": "tenkodogo", "currency": "dkk", "unit": "km", "name": "  Carol Reyes", "notes2": ""}, {"signup_date": "45262", "phone": "589-618-9670", "state": "Ohio", "city": "tenkoogo", "currency": "DKK", "unit": "kilogram", "name": "  Ivan Adams", "notes2": ""}, {"signup_date": "17 Dec 2023", "phone": "5287086996", "state": "sd ", "city": "   Tenkodogo", "currency": "GIP", "unit": " kilogiams ", "name": "Grace Reyes", "notes2": ""}, {"signup_date": "7/5/2023", "phone": "510.572.1662", "state": "ohio", "city": "  bloemhof", "currency": "  UNIDAD DE VALOR REAL", "unit": "kilogram", "name": "Grace  Moore", "notes2": ""}, {"signup_date": "2023-12-11", "phone": "5895431724", "state": "South Dxkota", "city": "Tenkodogo", "currency": "DKK", "unit": "kilometer", "name": "Karl  Fischer", "notes2": ""}, {"signup_date": "2023-08-13", "phone": "528-114-4708", "state": "FL", "city": "MOJOKERTO", "currency": "cou", "unit": " KM ", "name": "Grace Lee", "notes2": ""}, {"signup_date": "10 Mar 2023", "phone": "(506)312-1780", "state": "ohio", "city": "mojokreto", "currency": "Gibraltar Pound", "unit": "km", "name": "Eve Johnson", "notes2": ""}, {"signup_date": "7/10/2023", "phone": "526.736.8953", "state": "Fl", "city": "tenkodogo", "currency": "Gibraltar Pound", "unit": "km", "name": "  Carol Ali", "notes2": ""}, {"signup_date": "16 Apr 2023", "phone": "565-938-1737", "state": "FL", "city": " Bloemhof", "currency": "Gibraltar Pound", "unit": "kilometers", "name": "Alice Wong", "notes2": ""}, {"signup_date": "45071", "phone": "500.442.9683", "state": "OH", "city": "Mojokerto", "currency": "COU", "unit": "kilometer", "name": " Lena Cruz", "notes2": ""}, {"signup_date": "3/20/2023", "phone": "584.639.7745", "state": "Sotuh Dakota", "city": "Tenkodoo", "currency": "gibraltar pound", "unit": "kilograms", "name": "Carol Diaz", "notes2": ""}, {"signup_date": "7/11/2023", "phone": "588-983-8935", "state": "oh", "city": "Zawiercie", "currency": "Danish Krone", "unit": "KM  ", "name": "Priya Adams", "notes2": ""}, {"signup_date": "44977", "phone": "5841421396", "state": "fllrida", "city": "tenkodogo", "currency": "DKK", "unit": "KM", "name": "Bob Diaz  ", "notes2": ""}, {"signup_date": "45241", "phone": "(525)534-9863", "state": "   SD", "city": "bloemhof", "currency": "unidad de valorr real", "unit": "Kiloweter", "name": "Omar  Smith", "notes2": ""}, {"signup_date": "2023-12-13", "phone": "5819061726", "state": "Ohio   ", "city": "Bukit Panjang New Town", "currency": "DKK", "unit": "kilometens", "name": "Omar Novak ", "notes2": ""}, {"signup_date": "27 Oct 2023", "phone": "502-576-2473", "state": "   South Dakota", "city": "bloemhnf", "currency": "Danish Krone", "unit": "km ", "name": " Eve Reyes", "notes2": ""}, {"signup_date": "28 Apr 2023", "phone": "586-424-5810", "state": "Sd", "city": "Bukit Panjang New Town", "currency": "GIP", "unit": "   KILOMETER", "name": "Carol  Novak", "notes2": ""}, {"signup_date": "2023-10-20", "phone": "(566)118-3817", "state": "Fl", "city": "   bloemhof", "currency": "Unidad de Valor Real", "unit": "KG", "name": "  Judy Petrov", "notes2": ""}, {"signup_date": "16 Mar 2023", "phone": "521-590-5076", "state": "FL", "city": "Mojokerto", "currency": "GIP", "unit": "kilogam", "name": "   Frank Khan", "notes2": ""}, {"signup_date": "2/11/2023", "phone": "502.402.5964", "state": "FL", "city": "Mojokerto", "currency": " DKK ", "unit": "kg", "name": "Karl Smith", "notes2": ""}, {"signup_date": "2023-03-15", "phone": "5234405341", "state": "SD", "city": "Mojokerto   ", "currency": "COU", "unit": "km", "name": "Bob  Petrov", "notes2": ""}, {"signup_date": "2023-01-13", "phone": "(554)016-6933", "state": "sd", "city": "   Tenkodogo", "currency": "DKK", "unit": "Kg", "name": "   Ivan Park", "notes2": ""}, {"signup_date": "12 Jul 2023", "phone": "5921498777", "state": "South  Daota", "city": "Zawiercie", "currency": "Unidad de Valor Real", "unit": "km", "name": "Frank Petrov", "notes2": ""}, {"signup_date": "10 Mar 2023", "phone": "5001874155", "state": "ohio", "city": "Bloemhof", "currency": "Unidad de Valor Real", "unit": "kilograms", "name": "Karl Moore", "notes2": ""}, {"signup_date": "2023-12-09", "phone": "543.210.8986", "state": "Sd", "city": "tenkodogo", "currency": "GIBRALTAR POUND", "unit": "km", "name": "Mona Moore", "notes2": ""}, {"signup_date": "", "phone": "", "state": "", "city": "", "currency": "", "unit": "", "name": "", "notes2": ""}], "clean": [{"signup_date": "2023-02-22", "phone": "(582) 402-1566", "state": "South Dakota", "city": "Mojokerto", "currency": "COU", "unit": "kg", "name": "Carol Brandt"}, {"signup_date": "2023-07-02", "phone": "(549) 975-5188", "state": "South Dakota", "city": "Mojokerto", "currency": "GIP", "unit": "kg", "name": "Bob Khan"}, {"signup_date": "2023-04-24", "phone": "(564) 225-5805", "state": "South Dakota", "city": "Tenkodogo", "currency": "COU", "unit": "kg", "name": "Omar Brandt"}, {"signup_date": "2023-12-03", "phone": "(573) 087-2847", "state": "Ohio", "city": "Mojokerto", "currency": "DKK", "unit": "kg", "name": "Carol Diaz"}, {"signup_date": "2023-09-16", "phone": "(526) 836-5557", "state": "South Dakota", "city": "Bloemhof", "currency": "COU", "unit": "km", "name": "Mona Lee"}, {"signup_date": "2023-03-10", "phone": "(500) 187-4155", "state": "Ohio", "city": "Bloemhof", "currency": "COU", "unit": "kg", "name": "Karl Moore"}, {"signup_date": "2023-08-25", "phone": "(587) 454-5761", "state": "Ohio", "city": "Bukit Panjang New Town", "currency": "COU", "unit": "kg", "name": "Ivan Diaz"}, {"signup_date": "2023-04-01", "phone": "(545) 883-6810", "state": "South Dakota", "city": "Mojokerto", "currency": "GIP", "unit": "kg", "name": "Alice Park"}, {"signup_date": "2023-06-04", "phone": "(515) 526-9141", "state": "Ohio", "city": "Bukit Panjang New Town", "currency": "COU", "unit": "km", "name": "Carol Johnson"}, {"signup_date": "2023-01-08", "phone": "(587) 587-6110", "state": "Florida", "city": "Zawiercie", "currency": "GIP", "unit": "kg", "name": "Heidi Fischer"}, {"signup_date": "2023-06-25", "phone": "(524) 374-5472", "state": "Ohio", "city": "Zawiercie", "currency": "DKK", "unit": "kg", "name": "Eve Adams"}, {"signup_date": "2023-11-01", "phone": "(564) 729-8017", "state": "Florida", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Priya Novak"}, {"signup_date": "2023-10-17", "phone": "(537) 326-6607", "state": "South Dakota", "city": "Tenkodogo", "currency": "DKK", "unit": "kg", "name": "Mona Lee"}, {"signup_date": "2023-11-01", "phone": "(506) 349-3657", "state": "Florida", "city": "Tenkodogo", "currency": "DKK", "unit": "kg", "name": "Ivan Lee"}, {"signup_date": "2023-11-22", "phone": "(596) 159-1803", "state": "South Dakota", "city": "Bukit Panjang New Town", "currency": "GIP", "unit": "km", "name": "Carol Park"}, {"signup_date": "2023-03-23", "phone": "(527) 253-5360", "state": "Ohio", "city": "Zawiercie", "currency": "DKK", "unit": "kg", "name": "Eve Fischer"}, {"signup_date": "2023-01-09", "phone": "(565) 978-3907", "state": "Florida", "city": "Bukit Panjang New Town", "currency": "GIP", "unit": "km", "name": "Priya Cruz"}, {"signup_date": "2023-01-20", "phone": "(542) 686-6999", "state": "Ohio", "city": "Tenkodogo", "currency": "GIP", "unit": "kg", "name": "Karl Diaz"}, {"signup_date": "2023-07-11", "phone": "(544) 529-5271", "state": "Ohio", "city": "Mojokerto", "currency": "GIP", "unit": "kg", "name": "Mona Ali"}, {"signup_date": "2023-12-07", "phone": "(508) 449-8443", "state": "Florida", "city": "Zawiercie", "currency": "DKK", "unit": "kg", "name": "Judy Johnson"}, {"signup_date": "2023-02-17", "phone": "(553) 389-4169", "state": "Florida", "city": "Bloemhof", "currency": "DKK", "unit": "kg", "name": "Karl Lee"}, {"signup_date": "2023-01-11", "phone": "(593) 539-4252", "state": "Ohio", "city": "Bloemhof", "currency": "GIP", "unit": "km", "name": "Omar Ali"}, {"signup_date": "2023-12-09", "phone": "(543) 210-8986", "state": "South Dakota", "city": "Tenkodogo", "currency": "GIP", "unit": "km", "name": "Mona Moore"}, {"signup_date": "2023-06-12", "phone": "(570) 129-0162", "state": "Ohio", "city": "Zawiercie", "currency": "DKK", "unit": "km", "name": "David Wong"}, {"signup_date": "2023-09-07", "phone": "(527) 423-4271", "state": "Ohio", "city": "Tenkodogo", "currency": "COU", "unit": "kg", "name": "Frank Khan"}, {"signup_date": "2023-03-09", "phone": "(535) 315-0100", "state": "Ohio", "city": "Tenkodogo", "currency": "COU", "unit": "km", "name": "Karl Cruz"}, {"signup_date": "2023-01-17", "phone": "(563) 610-2753", "state": "Ohio", "city": "Tenkodogo", "currency": "GIP", "unit": "kg", "name": "Lena Fischer"}, {"signup_date": "2023-04-26", "phone": "(571) 595-0928", "state": "Ohio", "city": "Tenkodogo", "currency": "DKK", "unit": "kg", "name": "Carol Fischer"}, {"signup_date": "2023-12-25", "phone": "(534) 073-0650", "state": "Ohio", "city": "Bloemhof", "currency": "GIP", "unit": "km", "name": "Eve Petrov"}, {"signup_date": "2023-09-18", "phone": "(558) 859-5669", "state": "Ohio", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Frank Cruz"}, {"signup_date": "2023-09-05", "phone": "(588) 151-4953", "state": "Florida", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Heidi Adams"}, {"signup_date": "2023-07-28", "phone": "(575) 744-9635", "state": "Florida", "city": "Bukit Panjang New Town", "currency": "GIP", "unit": "kg", "name": "Bob Diaz"}, {"signup_date": "2023-11-11", "phone": "(556) 062-9095", "state": "Florida", "city": "Zawiercie", "currency": "GIP", "unit": "kg", "name": "Ivan Brandt"}, {"signup_date": "2023-03-24", "phone": "(519) 635-8556", "state": "Ohio", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Carol Reyes"}, {"signup_date": "2023-12-02", "phone": "(589) 618-9670", "state": "Ohio", "city": "Tenkodogo", "currency": "DKK", "unit": "kg", "name": "Ivan Adams"}, {"signup_date": "2023-12-17", "phone": "(528) 708-6996", "state": "South Dakota", "city": "Tenkodogo", "currency": "GIP", "unit": "kg", "name": "Grace Reyes"}, {"signup_date": "2023-07-05", "phone": "(510) 572-1662", "state": "Ohio", "city": "Bloemhof", "currency": "COU", "unit": "kg", "name": "Grace Moore"}, {"signup_date": "2023-12-11", "phone": "(589) 543-1724", "state": "South Dakota", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Karl Fischer"}, {"signup_date": "2023-08-13", "phone": "(528) 114-4708", "state": "Florida", "city": "Mojokerto", "currency": "COU", "unit": "km", "name": "Grace Lee"}, {"signup_date": "2023-03-10", "phone": "(506) 312-1780", "state": "Ohio", "city": "Mojokerto", "currency": "GIP", "unit": "km", "name": "Eve Johnson"}, {"signup_date": "2023-07-10", "phone": "(526) 736-8953", "state": "Florida", "city": "Tenkodogo", "currency": "GIP", "unit": "km", "name": "Carol Ali"}, {"signup_date": "2023-04-16", "phone": "(565) 938-1737", "state": "Florida", "city": "Bloemhof", "currency": "GIP", "unit": "km", "name": "Alice Wong"}, {"signup_date": "2023-05-25", "phone": "(500) 442-9683", "state": "Ohio", "city": "Mojokerto", "currency": "COU", "unit": "km", "name": "Lena Cruz"}, {"signup_date": "2023-03-20", "phone": "(584) 639-7745", "state": "South Dakota", "city": "Tenkodogo", "currency": "GIP", "unit": "kg", "name": "Carol Diaz"}, {"signup_date": "2023-07-11", "phone": "(588) 983-8935", "state": "Ohio", "city": "Zawiercie", "currency": "DKK", "unit": "km", "name": "Priya Adams"}, {"signup_date": "2023-02-20", "phone": "(584) 142-1396", "state": "Florida", "city": "Tenkodogo", "currency": "DKK", "unit": "km", "name": "Bob Diaz"}, {"signup_date": "2023-11-11", "phone": "(525) 534-9863", "state": "South Dakota", "city": "Bloemhof", "currency": "COU", "unit": "km", "name": "Omar Smith"}, {"signup_date": "2023-12-13", "phone": "(581) 906-1726", "state": "Ohio", "city": "Bukit Panjang New Town", "currency": "DKK", "unit": "km", "name": "Omar Novak"}, {"signup_date": "2023-10-27", "phone": "(502) 576-2473", "state": "South Dakota", "city": "Bloemhof", "currency": "DKK", "unit": "km", "name": "Eve Reyes"}, {"signup_date": "2023-04-28", "phone": "(586) 424-5810", "state": "South Dakota", "city": "Bukit Panjang New Town", "currency": "GIP", "unit": "km", "name": "Carol Novak"}, {"signup_date": "2023-10-20", "phone": "(566) 118-3817", "state": "Florida", "city": "Bloemhof", "currency": "COU", "unit": "kg", "name": "Judy Petrov"}, {"signup_date": "2023-03-16", "phone": "(521) 590-5076", "state": "Florida", "city": "Mojokerto", "currency": "GIP", "unit": "kg", "name": "Frank Khan"}, {"signup_date": "2023-02-11", "phone": "(502) 402-5964", "state": "Florida", "city": "Mojokerto", "currency": "DKK", "unit": "kg", "name": "Karl Smith"}, {"signup_date": "2023-03-15", "phone": "(523) 440-5341", "state": "South Dakota", "city": "Mojokerto", "currency": "COU", "unit": "km", "name": "Bob Petrov"}, {"signup_date": "2023-01-13", "phone": "(554) 016-6933", "state": "South Dakota", "city": "Tenkodogo", "currency": "DKK", "unit": "kg", "name": "Ivan Park"}, {"signup_date": "2023-07-12", "phone": "(592) 149-8777", "state": "South Dakota", "city": "Zawiercie", "currency": "COU", "unit": "km", "name": "Frank Petrov"}], "dirty_cols": ["signup_date", "phone", "state", "city", "currency", "unit", "name", "notes2"], "clean_cols": ["signup_date", "phone", "state", "city", "currency", "unit", "name"], "plan": {"dataset_summary": "59 rows × 8 columns. 7 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SD": "South Dakota", "OH": "Ohio", "sd": "South Dakota", "ohio": "Ohio", "Fl": "Florida", "florida": "Florida", "Ohoi": "Ohio", "FL": "Florida", "fl": "Florida", "Sd": "South Dakota", "oh": "Ohio", "Oh": "Ohio", "South Dxkota": "South Dakota", "Sotuh Dakota": "South Dakota", "fllrida": "Florida", "South Daota": "South Dakota"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mojokerto": "Mojokerto", "TENKODOGO": "Tenkodogo", "MOJOKERTO": "Mojokerto", "BUKIT PANJANG NEW TOWN": "Bukit Panjang New Town", "ZAWIERCIE": "Zawiercie", "Zawizrcie": "Zawiercie", "tenkodogo": "Tenkodogo", "zawiercie": "Zawiercie", "Zawieruie": "Zawiercie", "bloemhof": "Bloemhof", "Zawierciie": "Zawiercie", "tenkoogo": "Tenkodogo", "mojokreto": "Mojokerto", "Tenkodoo": "Tenkodogo", "bloemhnf": "Bloemhof"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cou": "COU", "Unidad de Valor Real": "COU", "gibraltar pound": "GIP", "unidad de valor real": "COU", "danish krone": "DKK", "Danish Krone": "DKK", "gip": "GIP", "Gibraltar Pound": "GIP", "dkk": "DKK", "GIBRALTAR POUND": "GIP", "Unidad De Valor Real": "COU", "dannish krone": "DKK", "UNIDAD DE VALOR REAL": "COU", "unidad de valorr real": "COU"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilogram": "kg", "kilograms": "kg", "kilometers": "km", "Kilogram": "kg", "Km": "km", "Kg": "kg", "kilometer": "km", "KILOGRAMS": "kg", "kmlometer": "km", "KILOMETERS": "km", "KM": "km", "KG": "kg", "killogram": "kg", "kilogiams": "kg", "Kiloweter": "km", "kilometens": "km", "KILOMETER": "km", "kilogam": "kg"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"rate": "91.7%", "phone": "532-698-4380", "industry": "finnserv", "unit": "cm", "signup_date": "4 Jul 2023", "name": "Bob Cruz", "company": "Vehement"}, {"rate": "22.7%", "phone": "578.952.2038", "industry": "software ", "unit": "cm", "signup_date": "2023-09-21", "name": " Judy Johnson", "company": " Hooli"}, {"rate": "79.8%", "phone": "5749434899", "industry": "Banking", "unit": "inch", "signup_date": "45246", "name": "Karl Petrov", "company": "Wonka Co"}, {"rate": "78.0%", "phone": "5544169727", "industry": "Software", "unit": "inches", "signup_date": "12/24/2023", "name": "Lena Park", "company": "Soylent Corp  "}, {"rate": "78.0%", "phone": "596-081-5235", "industry": "Financial Services", "unit": "centimeters", "signup_date": "11/14/2023", "name": "Heidi Brandt", "company": "   Initech"}, {"rate": "67.8%", "phone": "545-173-9909", "industry": "Tech", "unit": "  in", "signup_date": "2023-09-10", "name": "Frank  Adams", "company": "Wonka Co"}, {"rate": "91.4%", "phone": "555.983.9443", "industry": "E-COMIERCE  ", "unit": "inches", "signup_date": "2023-06-15", "name": "  Heidi Johnson", "company": "Initech"}, {"rate": "61.8%", "phone": "(580)250-1113", "industry": "fintech", "unit": "centimeters", "signup_date": "25 Feb 2023", "name": "David  Cruz", "company": "   Acme Inc"}, {"rate": "96.4%", "phone": "574.166.3274", "industry": " Technology ", "unit": "  inch", "signup_date": "5/17/2023", "name": "Grace Novak ", "company": "Stark Industries  "}, {"rate": "77.5%", "phone": "5412152401", "industry": "Retail & E-commerce", "unit": "inches", "signup_date": "44982", "name": " Lena Diaz", "company": "  Globex"}, {"rate": "57.8%", "phone": "557-072-1821", "industry": "e-commerce", "unit": " cm ", "signup_date": "2023-10-11", "name": "David  Park", "company": "  Stark Industries"}, {"rate": "92.8%", "phone": "591.769.6580", "industry": "E-COMMERCE", "unit": "\"", "signup_date": "26 Dec 2023", "name": "Carol Lee   ", "company": "Vehement"}, {"rate": "58.1%", "phone": "5680576979", "industry": "FinServ", "unit": "L", "signup_date": "15 Aug 2023", "name": "Lena Park", "company": "Acme  Inc"}, {"rate": "65.9%", "phone": "5019710373", "industry": " Software ", "unit": "in", "signup_date": "45043", "name": "Grace Khan", "company": "Globex"}, {"rate": "18.0%", "phone": "588-559-2015", "industry": " finance ", "unit": "incch", "signup_date": "2023-08-09", "name": "Priya Novak", "company": "Wonka Co  "}, {"rate": "42.9%", "phone": "526.576.9090", "industry": "finance", "unit": "l  ", "signup_date": "45269", "name": " Alice Khan", "company": "Soylent Corp"}, {"rate": "31.2%", "phone": "526.109.6218", "industry": "finserv", "unit": "liter", "signup_date": "2023-02-05", "name": "Lena Wong ", "company": "Stark Industries  "}, {"rate": "28.9%", "phone": "(590)276-5285", "industry": "fintech", "unit": " inches", "signup_date": "2023-11-26", "name": "Heidi Reyes", "company": "Stark Industries"}, {"rate": "28.1%", "phone": "5966648453", "industry": "Financial Services", "unit": "inch", "signup_date": "10/20/2023", "name": "   David Diaz", "company": "Hooli"}, {"rate": "88.5%", "phone": "5661490117", "industry": "Tech", "unit": " litre ", "signup_date": "2023-08-23", "name": "Lena Moore", "company": "Acme Inc "}, {"rate": "84.8%", "phone": "5066542491", "industry": "   Finserv", "unit": "inch", "signup_date": "28 Oct 2023", "name": " Karl Ali", "company": "Umbrella  "}, {"rate": "98.2%", "phone": "532.180.7745", "industry": "Fintech", "unit": " Cm", "signup_date": "2023-02-11", "name": "Alice Ali", "company": "Acme Inc   "}, {"rate": "40.9%", "phone": "(541)820-1394", "industry": "It", "unit": "IN", "signup_date": "26 Mar 2023", "name": "Frank Khan", "company": "Cyberdyne"}, {"rate": "33.8%", "phone": "517.532.3632", "industry": "FINSERV", "unit": "LITERS", "signup_date": "44980", "name": "Mona Cruz", "company": "Initech"}, {"rate": "78.0%", "phone": "592-339-4529", "industry": "Retail", "unit": "CENTIMETER", "signup_date": "12/11/2023", "name": " Mona Brandt", "company": "Umbrella"}, {"rate": "93.9%", "phone": "577.884.4591", "industry": "retail & e-commerce", "unit": "   cm", "signup_date": "2023-01-15", "name": "Karl  Diaz", "company": "Vehement"}, {"rate": "63.9%", "phone": "545-643-0762", "industry": " Banking ", "unit": "lizer", "signup_date": "9/17/2023", "name": "Sara Novak", "company": "Hooli"}, {"rate": "86.0%", "phone": "554.974.3999", "industry": "retail", "unit": "\"", "signup_date": "2023-05-14", "name": "Alice Johnson", "company": "Stark Industries  "}, {"rate": "1.6%", "phone": "5985927665", "industry": "Techonlogy", "unit": "  \"", "signup_date": "45144", "name": "Carol  Moore", "company": "Wonka Co   "}, {"rate": "73.1%", "phone": "524.760.3394", "industry": "Fintech", "unit": "liter", "signup_date": "9/4/2023", "name": "Lena Khan", "company": "Stark Industries"}, {"rate": "77.2%", "phone": "5744354751", "industry": "retail", "unit": "CENTIMETERS", "signup_date": "7 Feb 2023", "name": "Grace Moore", "company": "Globex"}, {"rate": "63.0%", "phone": "564.449.9857", "industry": " E-commerce ", "unit": "l", "signup_date": "2023-07-20", "name": "Karl Moore", "company": "Stark  Industries"}, {"rate": "67.1%", "phone": "540.170.5562", "industry": "e-commerce", "unit": "centimeter", "signup_date": "45236", "name": "Karl Moore", "company": "Globex   "}, {"rate": "89.4%", "phone": "5396568868", "industry": "Retail", "unit": "Inches", "signup_date": "5/22/2023", "name": "Eve Petrov ", "company": "Globex"}, {"rate": "92.7%", "phone": "(589)678-1037", "industry": "FINSERV", "unit": "cm", "signup_date": "16 Mar 2023", "name": "Omar Johnson", "company": "Vehement"}, {"rate": "28.3%", "phone": "578.014.7149", "industry": "financial services", "unit": "cetnimeter", "signup_date": "2023-12-03", "name": "Grace Diaz", "company": " Initech "}, {"rate": "17.0%", "phone": "5671509107", "industry": "technology", "unit": "inches", "signup_date": "26 Jul 2023", "name": "Carol Lee", "company": " Umbrella"}], "clean": [{"rate": 0.917, "phone": "(532) 698-4380", "industry": "Financial Services", "unit": "cm", "signup_date": "2023-07-04", "name": "Bob Cruz", "company": "Vehement"}, {"rate": 0.22699999999999998, "phone": "(578) 952-2038", "industry": "Technology", "unit": "cm", "signup_date": "2023-09-21", "name": "Judy Johnson", "company": "Hooli"}, {"rate": 0.7979999999999999, "phone": "(574) 943-4899", "industry": "Financial Services", "unit": "in", "signup_date": "2023-11-16", "name": "Karl Petrov", "company": "Wonka Co"}, {"rate": 0.78, "phone": "(554) 416-9727", "industry": "Technology", "unit": "in", "signup_date": "2023-12-24", "name": "Lena Park", "company": "Soylent Corp"}, {"rate": 0.78, "phone": "(596) 081-5235", "industry": "Financial Services", "unit": "cm", "signup_date": "2023-11-14", "name": "Heidi Brandt", "company": "Initech"}, {"rate": 0.6779999999999999, "phone": "(545) 173-9909", "industry": "Technology", "unit": "in", "signup_date": "2023-09-10", "name": "Frank Adams", "company": "Wonka Co"}, {"rate": 0.914, "phone": "(555) 983-9443", "industry": "Retail", "unit": "in", "signup_date": "2023-06-15", "name": "Heidi Johnson", "company": "Initech"}, {"rate": 0.618, "phone": "(580) 250-1113", "industry": "Financial Services", "unit": "cm", "signup_date": "2023-02-25", "name": "David Cruz", "company": "Acme Inc"}, {"rate": 0.9640000000000001, "phone": "(574) 166-3274", "industry": "Technology", "unit": "in", "signup_date": "2023-05-17", "name": "Grace Novak", "company": "Stark Industries"}, {"rate": 0.775, "phone": "(541) 215-2401", "industry": "Retail", "unit": "in", "signup_date": "2023-02-25", "name": "Lena Diaz", "company": "Globex"}, {"rate": 0.578, "phone": "(557) 072-1821", "industry": "Retail", "unit": "cm", "signup_date": "2023-10-11", "name": "David Park", "company": "Stark Industries"}, {"rate": 0.9279999999999999, "phone": "(591) 769-6580", "industry": "Retail", "unit": "in", "signup_date": "2023-12-26", "name": "Carol Lee", "company": "Vehement"}, {"rate": 0.581, "phone": "(568) 057-6979", "industry": "Financial Services", "unit": "L", "signup_date": "2023-08-15", "name": "Lena Park", "company": "Acme Inc"}, {"rate": 0.659, "phone": "(501) 971-0373", "industry": "Technology", "unit": "in", "signup_date": "2023-04-27", "name": "Grace Khan", "company": "Globex"}, {"rate": 0.18, "phone": "(588) 559-2015", "industry": "Financial Services", "unit": "in", "signup_date": "2023-08-09", "name": "Priya Novak", "company": "Wonka Co"}, {"rate": 0.429, "phone": "(526) 576-9090", "industry": "Financial Services", "unit": "L", "signup_date": "2023-12-09", "name": "Alice Khan", "company": "Soylent Corp"}, {"rate": 0.312, "phone": "(526) 109-6218", "industry": "Financial Services", "unit": "L", "signup_date": "2023-02-05", "name": "Lena Wong", "company": "Stark Industries"}, {"rate": 0.289, "phone": "(590) 276-5285", "industry": "Financial Services", "unit": "in", "signup_date": "2023-11-26", "name": "Heidi Reyes", "company": "Stark Industries"}, {"rate": 0.281, "phone": "(596) 664-8453", "industry": "Financial Services", "unit": "in", "signup_date": "2023-10-20", "name": "David Diaz", "company": "Hooli"}, {"rate": 0.885, "phone": "(566) 149-0117", "industry": "Technology", "unit": "L", "signup_date": "2023-08-23", "name": "Lena Moore", "company": "Acme Inc"}, {"rate": 0.848, "phone": "(506) 654-2491", "industry": "Financial Services", "unit": "in", "signup_date": "2023-10-28", "name": "Karl Ali", "company": "Umbrella"}, {"rate": 0.982, "phone": "(532) 180-7745", "industry": "Financial Services", "unit": "cm", "signup_date": "2023-02-11", "name": "Alice Ali", "company": "Acme Inc"}, {"rate": 0.409, "phone": "(541) 820-1394", "industry": "Technology", "unit": "in", "signup_date": "2023-03-26", "name": "Frank Khan", "company": "Cyberdyne"}, {"rate": 0.33799999999999997, "phone": "(517) 532-3632", "industry": "Financial Services", "unit": "L", "signup_date": "2023-02-23", "name": "Mona Cruz", "company": "Initech"}, {"rate": 0.78, "phone": "(592) 339-4529", "industry": "Retail", "unit": "cm", "signup_date": "2023-12-11", "name": "Mona Brandt", "company": "Umbrella"}, {"rate": 0.9390000000000001, "phone": "(577) 884-4591", "industry": "Retail", "unit": "cm", "signup_date": "2023-01-15", "name": "Karl Diaz", "company": "Vehement"}, {"rate": 0.639, "phone": "(545) 643-0762", "industry": "Financial Services", "unit": "L", "signup_date": "2023-09-17", "name": "Sara Novak", "company": "Hooli"}, {"rate": 0.86, "phone": "(554) 974-3999", "industry": "Retail", "unit": "in", "signup_date": "2023-05-14", "name": "Alice Johnson", "company": "Stark Industries"}, {"rate": 0.016, "phone": "(598) 592-7665", "industry": "Technology", "unit": "in", "signup_date": "2023-08-06", "name": "Carol Moore", "company": "Wonka Co"}, {"rate": 0.731, "phone": "(524) 760-3394", "industry": "Financial Services", "unit": "L", "signup_date": "2023-09-04", "name": "Lena Khan", "company": "Stark Industries"}, {"rate": 0.772, "phone": "(574) 435-4751", "industry": "Retail", "unit": "cm", "signup_date": "2023-02-07", "name": "Grace Moore", "company": "Globex"}, {"rate": 0.63, "phone": "(564) 449-9857", "industry": "Retail", "unit": "L", "signup_date": "2023-07-20", "name": "Karl Moore", "company": "Stark Industries"}, {"rate": 0.6709999999999999, "phone": "(540) 170-5562", "industry": "Retail", "unit": "cm", "signup_date": "2023-11-06", "name": "Karl Moore", "company": "Globex"}, {"rate": 0.894, "phone": "(539) 656-8868", "industry": "Retail", "unit": "in", "signup_date": "2023-05-22", "name": "Eve Petrov", "company": "Globex"}, {"rate": 0.927, "phone": "(589) 678-1037", "industry": "Financial Services", "unit": "cm", "signup_date": "2023-03-16", "name": "Omar Johnson", "company": "Vehement"}, {"rate": 0.28300000000000003, "phone": "(578) 014-7149", "industry": "Financial Services", "unit": "cm", "signup_date": "2023-12-03", "name": "Grace Diaz", "company": "Initech"}, {"rate": 0.17, "phone": "(567) 150-9107", "industry": "Technology", "unit": "in", "signup_date": "2023-07-26", "name": "Carol Lee", "company": "Umbrella"}], "dirty_cols": ["rate", "phone", "industry", "unit", "signup_date", "name", "company"], "clean_cols": ["rate", "phone", "industry", "unit", "signup_date", "name", "company"], "plan": {"dataset_summary": "37 rows × 7 columns. 7 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"finnserv": "Financial Services", "software": "Technology", "Banking": "Financial Services", "Software": "Technology", "Tech": "Technology", "E-COMIERCE": "Retail", "fintech": "Financial Services", "Retail & E-commerce": "Retail", "e-commerce": "Retail", "E-COMMERCE": "Retail", "FinServ": "Financial Services", "finance": "Financial Services", "finserv": "Financial Services", "Finserv": "Financial Services", "Fintech": "Financial Services", "It": "Technology", "FINSERV": "Financial Services", "retail & e-commerce": "Retail", "retail": "Retail", "Techonlogy": "Technology", "E-commerce": "Retail", "financial services": "Financial Services", "technology": "Technology"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"inch": "in", "inches": "in", "centimeters": "cm", "\"": "in", "incch": "in", "l": "L", "liter": "L", "litre": "L", "Cm": "cm", "IN": "in", "LITERS": "L", "CENTIMETER": "cm", "lizer": "L", "CENTIMETERS": "cm", "centimeter": "cm", "Inches": "in", "cetnimeter": "cm"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"amount": "1,605", "company": "Hooli   ", "state": "Connetcicut", "country": " mx ", "rate": "67.0%", "name": " Sara Park", "unit": "liters", "col_x": ""}, {"amount": "$2,155.92", "company": "Hooli", "state": "Mississippi", "country": "TON ", "rate": "93.1%", "name": "Karl Ali", "unit": "LITES", "col_x": ""}, {"amount": "7.617,15", "company": "Acme Inc", "state": "south carolina", "country": "TO", "rate": "60.9%", "name": "Bob  Brandt", "unit": "KILOMETER", "col_x": ""}, {"amount": "$7,356.38", "company": "Vehement", "state": "   vermont", "country": "iq", "rate": "66.4%", "name": "Eve Wong", "unit": "kilometer", "col_x": ""}, {"amount": "3.957,13", "company": "Initech", "state": "CT", "country": "REPUBLIC OF MOLDOVA", "rate": "75.4%", "name": "David Novak ", "unit": "l", "col_x": ""}, {"amount": "6.033,94", "company": "Vehement", "state": "MS", "country": " TON ", "rate": "8.1%", "name": "Grace  Diaz", "unit": "MLS", "col_x": ""}, {"amount": "na", "company": "Stark Industries", "state": "sc", "country": "toga", "rate": "20.2%", "name": "Priya Johnson  ", "unit": "Km", "col_x": ""}, {"amount": "5.890,45", "company": "Umbrella  ", "state": "Mississippi", "country": "REPUBLIC OF MOLDOVA", "rate": "32.6%", "name": "David Johnson ", "unit": " kilograms", "col_x": ""}, {"amount": "--", "company": "Umbrella", "state": "Connecticut", "country": "KINGDM OF TONGA ", "rate": "52.6%", "name": "   Heidi Johnson", "unit": " milliliter ", "col_x": ""}, {"amount": "7.300,02", "company": "Hooli", "state": "Mississippi", "country": "MEX", "rate": "28.0%", "name": "Heidi Cruz", "unit": "MILLILITER", "col_x": ""}, {"amount": "7.923,39", "company": "Initech", "state": "SC", "country": "republic of moldova", "rate": "47.8%", "name": "Priya Park", "unit": " liters ", "col_x": ""}, {"amount": "1.070,15", "company": "Stark Industries  ", "state": "Oio", "country": "Moldova, Republic of", "rate": "19.4%", "name": "Lena Reyes", "unit": "KG", "col_x": ""}, {"amount": "$7,603.98", "company": "Vehement   ", "state": "south carolina", "country": "MEXICO", "rate": "91.9%", "name": "Ivan Novak", "unit": "ml   ", "col_x": ""}, {"amount": "null", "company": "Globex", "state": "OH", "country": " iraq ", "rate": "58.2%", "name": "Alice Fischer", "unit": "Km", "col_x": ""}, {"amount": "7.596,54", "company": "   Hooli", "state": "Ohio", "country": " tonga ", "rate": "86.0%", "name": "Bob Petrov ", "unit": "ml", "col_x": ""}, {"amount": "4.343,29", "company": "Acme Inc   ", "state": "oh", "country": "united mexican states", "rate": "72.1%", "name": "Mona Diaz", "unit": "KILOMETERS  ", "col_x": ""}, {"amount": "$2,732.81", "company": "Globex", "state": "ohio", "country": "Republic of Iraq", "rate": "89.5%", "name": "Karl  Ali", "unit": "ml", "col_x": ""}, {"amount": "4.409,46", "company": "Umbrella", "state": "MS ", "country": "MD", "rate": "42.5%", "name": " Alice Khan", "unit": "KG", "col_x": ""}, {"amount": "4.387,95", "company": "Cyberdyne   ", "state": " ct ", "country": "iraq", "rate": "66.4%", "name": "Eve Brandt ", "unit": "litre", "col_x": ""}, {"amount": "null", "company": "Umbrella", "state": "MS", "country": "ton", "rate": "89.9%", "name": " Mona Brandt", "unit": "kilometers ", "col_x": ""}, {"amount": "406,26", "company": "Cyberdyne", "state": "ct", "country": "iq", "rate": "25.1%", "name": "Priya Wong", "unit": "Km   ", "col_x": ""}, {"amount": "$1,050.94", "company": " Wonka Co", "state": "VT", "country": "IQ   ", "rate": "53.3%", "name": "Eve  Reyes", "unit": "L", "col_x": ""}, {"amount": "$1,159.23", "company": " Cyberdyne", "state": "VT", "country": "mexico", "rate": "51.2%", "name": "Frank Johnson", "unit": "kilogram", "col_x": ""}, {"amount": "$5,217.27", "company": " Globex ", "state": "South Carolina", "country": "United  Mexican States", "rate": "73.1%", "name": "  Eve Smith", "unit": "kg", "col_x": ""}, {"amount": "7.374,87", "company": "Wonka Co", "state": "Connecticut", "country": "IQ", "rate": "56.6%", "name": "Bob Wong", "unit": " km", "col_x": ""}, {"amount": "$5,933.15", "company": "  Wonka Co", "state": "ohio", "country": "to", "rate": "5.5%", "name": "  Ivan Johnson", "unit": "mL", "col_x": ""}, {"amount": "na", "company": "Hooli", "state": "  Ohio", "country": "TO", "rate": "68.9%", "name": "Omar  Smith", "unit": " KM", "col_x": ""}, {"amount": "$8,482.96", "company": "Stark Industries", "state": "   VT", "country": "ton", "rate": "31.3%", "name": "Frank Lee", "unit": "Km", "col_x": ""}, {"amount": "$2,414.43", "company": "Hooli  ", "state": "South Craolina", "country": "united mexican states", "rate": "84.8%", "name": "Judy Diaz", "unit": "KILOMETERS", "col_x": ""}, {"amount": "None", "company": "Umbrella", "state": "Ct", "country": "Iraq", "rate": "3.2%", "name": "Heidi Johnson", "unit": "l", "col_x": ""}, {"amount": "2.613,50", "company": "Initech", "state": "CT", "country": "Republic of Moldova", "rate": "86.5%", "name": "   Mona Moore", "unit": "km", "col_x": ""}, {"amount": "7.637,91", "company": "Vehement ", "state": "SC", "country": "republic of moldova ", "rate": "9.5%", "name": "Judy Khan", "unit": "mL", "col_x": ""}, {"amount": "$2,367.04", "company": "Hooli", "state": "South Carolina", "country": "   mx", "rate": "33.4%", "name": "  Priya Wong", "unit": "km", "col_x": ""}, {"amount": "582,20", "company": "Soylent  Corp", "state": "MISSISSIPPI", "country": "united mexican states   ", "rate": "11.0%", "name": "Frank Petrov", "unit": "  ml", "col_x": ""}, {"amount": "6.786,21", "company": "Stark Industries", "state": "OHIO", "country": "MEX", "rate": "18.6%", "name": "Mona Moore", "unit": "LITERS", "col_x": ""}, {"amount": "1.951,99", "company": "  Vehement", "state": "CT", "country": "MD", "rate": "14.5%", "name": " Judy Smith", "unit": "liters", "col_x": ""}, {"amount": "5.791,53", "company": "Soylent Corp ", "state": "Connecticut", "country": "iraq", "rate": "36.9%", "name": "Heidi Smith", "unit": "kilometers", "col_x": ""}, {"amount": "6.139,12", "company": "  Wonka Co", "state": "SC", "country": "Iraq", "rate": "58.7%", "name": "Carol Smith", "unit": "Kilograms", "col_x": ""}, {"amount": "3.286,35", "company": " Soylent Corp", "state": "CT   ", "country": "MDA", "rate": "56.4%", "name": "Lena  Ali", "unit": "kilogram", "col_x": ""}, {"amount": "4.883,79", "company": "Umbrella", "state": "  Ohio", "country": "MD ", "rate": "1.7%", "name": "Priya Wong  ", "unit": "Kg", "col_x": ""}, {"amount": "397,55", "company": "Globex", "state": "ms", "country": "Kingdom of Tonga", "rate": "70.3%", "name": "Mona Wong", "unit": "mls", "col_x": ""}, {"amount": "$6,937.28", "company": "Stark Industries", "state": "oh", "country": " MX ", "rate": "9.6%", "name": "Heidi Moore", "unit": "mL", "col_x": ""}, {"amount": "$6,557.67", "company": "Stark  Industries", "state": "SC", "country": "Mexico", "rate": "83.0%", "name": "Eve Diaz", "unit": "Liters", "col_x": ""}, {"amount": "$770.22", "company": "Soylent Corp ", "state": "OH", "country": "Irq", "rate": "55.3%", "name": "   Ivan Adams", "unit": "KG", "col_x": ""}, {"amount": "8.881,85", "company": "Soylent Corp", "state": "Sc", "country": "MEX", "rate": "23.2%", "name": "Sara Fischer", "unit": "kilometer", "col_x": ""}, {"amount": "$3,864.50", "company": " Cyberdyne ", "state": " CT", "country": "irq", "rate": "56.0%", "name": " Karl Wong", "unit": "l", "col_x": ""}, {"amount": "$6,897.10", "company": "Vehement", "state": "ms", "country": "Republic of Iraq", "rate": "26.1%", "name": "Eve Novak", "unit": "ml", "col_x": ""}, {"amount": "2.479,38", "company": "Umbrella", "state": " OH ", "country": "Irq", "rate": "76.4%", "name": "Bob  Cruz", "unit": " milliliter ", "col_x": ""}], "clean": [{"amount": 1605.0, "company": "Hooli", "state": "Connecticut", "country": "Mexico", "rate": 0.67, "name": "Sara Park", "unit": "L"}, {"amount": 2155.92, "company": "Hooli", "state": "Mississippi", "country": "Tonga", "rate": 0.9309999999999999, "name": "Karl Ali", "unit": "L"}, {"amount": 7617.15, "company": "Acme Inc", "state": "South Carolina", "country": "Tonga", "rate": 0.609, "name": "Bob Brandt", "unit": "km"}, {"amount": 7356.38, "company": "Vehement", "state": "Vermont", "country": "Iraq", "rate": 0.664, "name": "Eve Wong", "unit": "km"}, {"amount": 3957.13, "company": "Initech", "state": "Connecticut", "country": "Moldova, Republic of", "rate": 0.754, "name": "David Novak", "unit": "L"}, {"amount": 6033.94, "company": "Vehement", "state": "Mississippi", "country": "Tonga", "rate": 0.081, "name": "Grace Diaz", "unit": "mL"}, {"amount": NaN, "company": "Stark Industries", "state": "South Carolina", "country": "Tonga", "rate": 0.20199999999999999, "name": "Priya Johnson", "unit": "km"}, {"amount": 5890.45, "company": "Umbrella", "state": "Mississippi", "country": "Moldova, Republic of", "rate": 0.326, "name": "David Johnson", "unit": "kg"}, {"amount": NaN, "company": "Umbrella", "state": "Connecticut", "country": "Tonga", "rate": 0.526, "name": "Heidi Johnson", "unit": "mL"}, {"amount": 7300.02, "company": "Hooli", "state": "Mississippi", "country": "Mexico", "rate": 0.28, "name": "Heidi Cruz", "unit": "mL"}, {"amount": 7923.39, "company": "Initech", "state": "South Carolina", "country": "Moldova, Republic of", "rate": 0.478, "name": "Priya Park", "unit": "L"}, {"amount": 1070.15, "company": "Stark Industries", "state": "Ohio", "country": "Moldova, Republic of", "rate": 0.19399999999999998, "name": "Lena Reyes", "unit": "kg"}, {"amount": 7603.98, "company": "Vehement", "state": "South Carolina", "country": "Mexico", "rate": 0.919, "name": "Ivan Novak", "unit": "mL"}, {"amount": NaN, "company": "Globex", "state": "Ohio", "country": "Iraq", "rate": 0.5820000000000001, "name": "Alice Fischer", "unit": "km"}, {"amount": 7596.54, "company": "Hooli", "state": "Ohio", "country": "Tonga", "rate": 0.86, "name": "Bob Petrov", "unit": "mL"}, {"amount": 4343.29, "company": "Acme Inc", "state": "Ohio", "country": "Mexico", "rate": 0.721, "name": "Mona Diaz", "unit": "km"}, {"amount": 2732.81, "company": "Globex", "state": "Ohio", "country": "Iraq", "rate": 0.895, "name": "Karl Ali", "unit": "mL"}, {"amount": 4409.46, "company": "Umbrella", "state": "Mississippi", "country": "Moldova, Republic of", "rate": 0.425, "name": "Alice Khan", "unit": "kg"}, {"amount": 4387.95, "company": "Cyberdyne", "state": "Connecticut", "country": "Iraq", "rate": 0.664, "name": "Eve Brandt", "unit": "L"}, {"amount": NaN, "company": "Umbrella", "state": "Mississippi", "country": "Tonga", "rate": 0.899, "name": "Mona Brandt", "unit": "km"}, {"amount": 406.26, "company": "Cyberdyne", "state": "Connecticut", "country": "Iraq", "rate": 0.251, "name": "Priya Wong", "unit": "km"}, {"amount": 1050.94, "company": "Wonka Co", "state": "Vermont", "country": "Iraq", "rate": 0.5329999999999999, "name": "Eve Reyes", "unit": "L"}, {"amount": 1159.23, "company": "Cyberdyne", "state": "Vermont", "country": "Mexico", "rate": 0.512, "name": "Frank Johnson", "unit": "kg"}, {"amount": 5217.27, "company": "Globex", "state": "South Carolina", "country": "Mexico", "rate": 0.731, "name": "Eve Smith", "unit": "kg"}, {"amount": 7374.87, "company": "Wonka Co", "state": "Connecticut", "country": "Iraq", "rate": 0.5660000000000001, "name": "Bob Wong", "unit": "km"}, {"amount": 5933.15, "company": "Wonka Co", "state": "Ohio", "country": "Tonga", "rate": 0.055, "name": "Ivan Johnson", "unit": "mL"}, {"amount": NaN, "company": "Hooli", "state": "Ohio", "country": "Tonga", "rate": 0.6890000000000001, "name": "Omar Smith", "unit": "km"}, {"amount": 8482.96, "company": "Stark Industries", "state": "Vermont", "country": "Tonga", "rate": 0.313, "name": "Frank Lee", "unit": "km"}, {"amount": 2414.43, "company": "Hooli", "state": "South Carolina", "country": "Mexico", "rate": 0.848, "name": "Judy Diaz", "unit": "km"}, {"amount": NaN, "company": "Umbrella", "state": "Connecticut", "country": "Iraq", "rate": 0.032, "name": "Heidi Johnson", "unit": "L"}, {"amount": 2613.5, "company": "Initech", "state": "Connecticut", "country": "Moldova, Republic of", "rate": 0.865, "name": "Mona Moore", "unit": "km"}, {"amount": 7637.91, "company": "Vehement", "state": "South Carolina", "country": "Moldova, Republic of", "rate": 0.095, "name": "Judy Khan", "unit": "mL"}, {"amount": 2367.04, "company": "Hooli", "state": "South Carolina", "country": "Mexico", "rate": 0.33399999999999996, "name": "Priya Wong", "unit": "km"}, {"amount": 582.2, "company": "Soylent Corp", "state": "Mississippi", "country": "Mexico", "rate": 0.11, "name": "Frank Petrov", "unit": "mL"}, {"amount": 6786.21, "company": "Stark Industries", "state": "Ohio", "country": "Mexico", "rate": 0.18600000000000003, "name": "Mona Moore", "unit": "L"}, {"amount": 1951.99, "company": "Vehement", "state": "Connecticut", "country": "Moldova, Republic of", "rate": 0.145, "name": "Judy Smith", "unit": "L"}, {"amount": 5791.53, "company": "Soylent Corp", "state": "Connecticut", "country": "Iraq", "rate": 0.369, "name": "Heidi Smith", "unit": "km"}, {"amount": 6139.12, "company": "Wonka Co", "state": "South Carolina", "country": "Iraq", "rate": 0.5870000000000001, "name": "Carol Smith", "unit": "kg"}, {"amount": 3286.35, "company": "Soylent Corp", "state": "Connecticut", "country": "Moldova, Republic of", "rate": 0.564, "name": "Lena Ali", "unit": "kg"}, {"amount": 4883.79, "company": "Umbrella", "state": "Ohio", "country": "Moldova, Republic of", "rate": 0.017, "name": "Priya Wong", "unit": "kg"}, {"amount": 397.55, "company": "Globex", "state": "Mississippi", "country": "Tonga", "rate": 0.703, "name": "Mona Wong", "unit": "mL"}, {"amount": 6937.28, "company": "Stark Industries", "state": "Ohio", "country": "Mexico", "rate": 0.096, "name": "Heidi Moore", "unit": "mL"}, {"amount": 6557.67, "company": "Stark Industries", "state": "South Carolina", "country": "Mexico", "rate": 0.83, "name": "Eve Diaz", "unit": "L"}, {"amount": 770.22, "company": "Soylent Corp", "state": "Ohio", "country": "Iraq", "rate": 0.5529999999999999, "name": "Ivan Adams", "unit": "kg"}, {"amount": 8881.85, "company": "Soylent Corp", "state": "South Carolina", "country": "Mexico", "rate": 0.23199999999999998, "name": "Sara Fischer", "unit": "km"}, {"amount": 3864.5, "company": "Cyberdyne", "state": "Connecticut", "country": "Iraq", "rate": 0.56, "name": "Karl Wong", "unit": "L"}, {"amount": 6897.1, "company": "Vehement", "state": "Mississippi", "country": "Iraq", "rate": 0.261, "name": "Eve Novak", "unit": "mL"}, {"amount": 2479.38, "company": "Umbrella", "state": "Ohio", "country": "Iraq", "rate": 0.764, "name": "Bob Cruz", "unit": "mL"}], "dirty_cols": ["amount", "company", "state", "country", "rate", "name", "unit", "col_x"], "clean_cols": ["amount", "company", "state", "country", "rate", "name", "unit"], "plan": {"dataset_summary": "48 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Connetcicut": "Connecticut", "south carolina": "South Carolina", "vermont": "Vermont", "CT": "Connecticut", "MS": "Mississippi", "sc": "South Carolina", "SC": "South Carolina", "Oio": "Ohio", "OH": "Ohio", "oh": "Ohio", "ohio": "Ohio", "ct": "Connecticut", "VT": "Vermont", "South Craolina": "South Carolina", "Ct": "Connecticut", "MISSISSIPPI": "Mississippi", "OHIO": "Ohio", "ms": "Mississippi", "Sc": "South Carolina"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mx": "Mexico", "TON": "Tonga", "TO": "Tonga", "iq": "Iraq", "REPUBLIC OF MOLDOVA": "Moldova, Republic of", "toga": "Tonga", "KINGDM OF TONGA": "Tonga", "MEX": "Mexico", "republic of moldova": "Moldova, Republic of", "MEXICO": "Mexico", "iraq": "Iraq", "tonga": "Tonga", "united mexican states": "Mexico", "Republic of Iraq": "Iraq", "MD": "Moldova, Republic of", "ton": "Tonga", "IQ": "Iraq", "mexico": "Mexico", "United Mexican States": "Mexico", "to": "Tonga", "Republic of Moldova": "Moldova, Republic of", "MDA": "Moldova, Republic of", "Kingdom of Tonga": "Tonga", "MX": "Mexico", "Irq": "Iraq", "irq": "Iraq"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liters": "L", "LITES": "L", "KILOMETER": "km", "kilometer": "km", "l": "L", "MLS": "mL", "Km": "km", "kilograms": "kg", "milliliter": "mL", "MILLILITER": "mL", "KG": "kg", "ml": "mL", "KILOMETERS": "km", "litre": "L", "kilometers": "km", "kilogram": "kg", "KM": "km", "LITERS": "L", "Kilograms": "kg", "Kg": "kg", "mls": "mL", "Liters": "L"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "Chief Executive Officer", "rate": "12.0%", "currency": "Guyana Dollar"}, {"job_title": "VP", "rate": "42.3%", "currency": " Dalasi "}, {"job_title": "Acct Exec", "rate": "45.2%", "currency": "All"}, {"job_title": "VP", "rate": "87.7%", "currency": "GYD"}, {"job_title": "V.P.", "rate": "78.6%", "currency": "Guyana Dollar"}, {"job_title": "ae   ", "rate": "27.4%", "currency": "Dalasi"}, {"job_title": "Vp", "rate": "45.5%", "currency": "GYD"}, {"job_title": "Vice Pres", "rate": "5.7%", "currency": "gmd"}, {"job_title": "VP..", "rate": "88.2%", "currency": "LEK"}, {"job_title": "VP", "rate": "63.1%", "currency": "Dalasi"}, {"job_title": "CEO   ", "rate": "47.0%", "currency": "  DAVASI"}, {"job_title": "CHIEF EXECUTIVE OFFICER", "rate": "29.9%", "currency": "guyana dollar"}, {"job_title": "CEO", "rate": "23.5%", "currency": " GYD "}, {"job_title": " Vice President", "rate": "91.1%", "currency": " GYD "}, {"job_title": "Vice Phes", "rate": "92.6%", "currency": "GYD"}, {"job_title": "c.e.o.", "rate": "67.7%", "currency": "GYD"}, {"job_title": "vice pres", "rate": "76.6%", "currency": " GMD "}, {"job_title": "ae", "rate": "66.4%", "currency": "Gmd"}, {"job_title": "V.P.", "rate": "4.7%", "currency": "Gyd"}, {"job_title": "AE", "rate": "56.6%", "currency": "ALL   "}, {"job_title": "Vice President  ", "rate": "79.1%", "currency": "guyana dollar"}, {"job_title": "V.P.", "rate": "84.3%", "currency": "GYD"}, {"job_title": " VP ", "rate": "61.6%", "currency": "GUYANA  DOLLAR"}, {"job_title": "Chief Executive Officer", "rate": "45.8%", "currency": " Dalasi "}, {"job_title": "Acct Exec", "rate": "45.2%", "currency": "All"}, {"job_title": "V.P.", "rate": "84.3%", "currency": "GYD"}, {"job_title": "", "rate": "", "currency": ""}], "clean": [{"job_title": "Chief Executive Officer", "rate": 0.12, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.423, "currency": "GMD"}, {"job_title": "Account Executive", "rate": 0.452, "currency": "ALL"}, {"job_title": "Vice President", "rate": 0.877, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.7859999999999999, "currency": "GYD"}, {"job_title": "Account Executive", "rate": 0.27399999999999997, "currency": "GMD"}, {"job_title": "Vice President", "rate": 0.455, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.057, "currency": "GMD"}, {"job_title": "Vice President", "rate": 0.882, "currency": "ALL"}, {"job_title": "Vice President", "rate": 0.631, "currency": "GMD"}, {"job_title": "Chief Executive Officer", "rate": 0.47, "currency": "GMD"}, {"job_title": "Chief Executive Officer", "rate": 0.299, "currency": "GYD"}, {"job_title": "Chief Executive Officer", "rate": 0.235, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.9109999999999999, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.9259999999999999, "currency": "GYD"}, {"job_title": "Chief Executive Officer", "rate": 0.677, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.7659999999999999, "currency": "GMD"}, {"job_title": "Account Executive", "rate": 0.664, "currency": "GMD"}, {"job_title": "Vice President", "rate": 0.047, "currency": "GYD"}, {"job_title": "Account Executive", "rate": 0.5660000000000001, "currency": "ALL"}, {"job_title": "Vice President", "rate": 0.7909999999999999, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.843, "currency": "GYD"}, {"job_title": "Vice President", "rate": 0.616, "currency": "GYD"}, {"job_title": "Chief Executive Officer", "rate": 0.45799999999999996, "currency": "GMD"}], "dirty_cols": ["job_title", "rate", "currency"], "clean_cols": ["job_title", "rate", "currency"], "plan": {"dataset_summary": "27 rows × 3 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"VP": "Vice President", "Acct Exec": "Account Executive", "V.P.": "Vice President", "ae": "Account Executive", "Vp": "Vice President", "Vice Pres": "Vice President", "VP..": "Vice President", "CEO": "Chief Executive Officer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "Vice Phes": "Vice President", "c.e.o.": "Chief Executive Officer", "vice pres": "Vice President", "AE": "Account Executive"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Guyana Dollar": "GYD", "Dalasi": "GMD", "All": "ALL", "gmd": "GMD", "LEK": "ALL", "DAVASI": "GMD", "guyana dollar": "GYD", "Gmd": "GMD", "Gyd": "GYD", "GUYANA DOLLAR": "GYD"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "63.9%", "city": "Changyi", "name": "Omar Wong"}, {"rate": "67.5%", "city": "   Fair Oaks", "name": "Omar Cruz   "}, {"rate": "86.0%", "city": "Changyi", "name": "David Ali "}, {"rate": "42.4%", "city": "Fair Oaks", "name": " Heidi Moore"}, {"rate": "79.2%", "city": "changyi", "name": "Priya  Petrov"}, {"rate": "5.1%", "city": "Fair Oaks ", "name": "Heidi Diaz "}, {"rate": "89.9%", "city": "Peicheng", "name": " Sara Brandt"}, {"rate": "98.1%", "city": "  changyi", "name": "Carol Fischer"}, {"rate": "89.5%", "city": "Fir Oaks", "name": "Grace Fischer"}, {"rate": "14.1%", "city": "Peicheng", "name": "Karl Brandt"}, {"rate": "75.8%", "city": "changyi", "name": "  Omar Park"}, {"rate": "0.7%", "city": "Fair Oaks", "name": "Frank Park"}, {"rate": "90.5%", "city": "Peicheng", "name": "Sara Wong"}, {"rate": "75.8%", "city": "Fair Oaks", "name": "Carol Moore"}, {"rate": "43.4%", "city": "changyi", "name": "Grace Reyes"}, {"rate": "3.7%", "city": "changyi", "name": " Judy Diaz"}, {"rate": "67.3%", "city": " changyi ", "name": "Grace Novak "}, {"rate": "98.1%", "city": "  changyi", "name": "Carol Fischer"}, {"rate": "", "city": "", "name": ""}], "clean": [{"rate": 0.639, "city": "Changyi", "name": "Omar Wong"}, {"rate": 0.675, "city": "Fair Oaks", "name": "Omar Cruz"}, {"rate": 0.86, "city": "Changyi", "name": "David Ali"}, {"rate": 0.424, "city": "Fair Oaks", "name": "Heidi Moore"}, {"rate": 0.792, "city": "Changyi", "name": "Priya Petrov"}, {"rate": 0.051, "city": "Fair Oaks", "name": "Heidi Diaz"}, {"rate": 0.899, "city": "Peicheng", "name": "Sara Brandt"}, {"rate": 0.981, "city": "Changyi", "name": "Carol Fischer"}, {"rate": 0.895, "city": "Fair Oaks", "name": "Grace Fischer"}, {"rate": 0.141, "city": "Peicheng", "name": "Karl Brandt"}, {"rate": 0.758, "city": "Changyi", "name": "Omar Park"}, {"rate": 0.006999999999999999, "city": "Fair Oaks", "name": "Frank Park"}, {"rate": 0.905, "city": "Peicheng", "name": "Sara Wong"}, {"rate": 0.758, "city": "Fair Oaks", "name": "Carol Moore"}, {"rate": 0.434, "city": "Changyi", "name": "Grace Reyes"}, {"rate": 0.037000000000000005, "city": "Changyi", "name": "Judy Diaz"}, {"rate": 0.6729999999999999, "city": "Changyi", "name": "Grace Novak"}], "dirty_cols": ["rate", "city", "name"], "clean_cols": ["rate", "city", "name"], "plan": {"dataset_summary": "19 rows × 3 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"changyi": "Changyi", "Fir Oaks": "Fair Oaks"}, "rationale": "Unified 2 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"name": "Frank Petrov  ", "state": "tennesese", "city": " denver ", "phone": "5162850682", "job_title": "Eng Mgr", "signup_date": "5/14/2023", "col_x": ""}, {"name": "Eve Fischer   ", "state": "ariznoa   ", "city": "Vladimir", "phone": "530.767.4379", "job_title": "Engineering  Mgr", "signup_date": "21 Aug 2023", "col_x": ""}, {"name": "Bob Adams", "state": "ARIIZONA", "city": "Vladimir", "phone": "(541)640-2485", "job_title": "Chief Technology Officer  ", "signup_date": "2023-04-06", "col_x": ""}, {"name": "Heidi Novak", "state": "Arizona", "city": "Vladimir", "phone": "505-182-0594", "job_title": " dcr.", "signup_date": "9/1/2023", "col_x": ""}, {"name": "  Heidi Diaz", "state": "tn", "city": "Vladimir", "phone": "503-303-1174", "job_title": "admin  assistant", "signup_date": "2023-01-16", "col_x": ""}, {"name": "Bob Johnson", "state": "AZ", "city": "Denver", "phone": "5311015314", "job_title": "DIIR.", "signup_date": "45071", "col_x": ""}, {"name": "Carol Moore", "state": "AZ", "city": "Toronto", "phone": "5854109072", "job_title": "cto", "signup_date": "15 Jan 2023", "col_x": ""}, {"name": "Priya Petrov   ", "state": "Arizona", "city": "  Waldkirch", "phone": "5338479853", "job_title": " CTO", "signup_date": "9/11/2023", "col_x": ""}, {"name": "Carol Petrov   ", "state": "tennessee", "city": "Toronto", "phone": "566-851-7360", "job_title": "Eng Mgr", "signup_date": "6/20/2023", "col_x": ""}, {"name": "Eve Reyes", "state": "   az", "city": "Denver", "phone": "562-555-3437", "job_title": "CTO", "signup_date": "10/14/2023", "col_x": ""}, {"name": "Alice Park  ", "state": " az ", "city": " Vladmir ", "phone": "(592)300-9221", "job_title": "Director", "signup_date": "12/2/2023", "col_x": ""}, {"name": "Judy Moore", "state": "tennessee", "city": "Vladimir", "phone": "515.460.3035", "job_title": "Admin", "signup_date": "18 Nov 2023", "col_x": ""}, {"name": "Grace Moore", "state": "tenessee", "city": "Waldkirch", "phone": "(572)642-2385", "job_title": "ADMIN ASSISTANT", "signup_date": "2023-07-17", "col_x": ""}, {"name": "Alice Novak  ", "state": " arizona ", "city": "WALDKIRCH", "phone": "594.222.8273", "job_title": "Administrative Assistant", "signup_date": "2023-06-02", "col_x": ""}, {"name": "  Priya Brandt", "state": "Az", "city": "toronto", "phone": "(597)364-0603", "job_title": "   Dir", "signup_date": "4/11/2023", "col_x": ""}, {"name": "Eve Moore", "state": "tennessee", "city": "Toronto", "phone": "5438910310", "job_title": "administrative assistant", "signup_date": "27 Aug 2023", "col_x": ""}, {"name": "Ivan Khan", "state": "Arizona", "city": "Toronto", "phone": "500.775.5951", "job_title": "CTO", "signup_date": "11/12/2023", "col_x": ""}, {"name": "Priya Fischer", "state": "Tennessee", "city": "Denver", "phone": "(591)971-3917", "job_title": "Dir.", "signup_date": "44987", "col_x": ""}, {"name": "Heidi Brandt", "state": "tn", "city": "DENVER", "phone": "(593)298-6169", "job_title": "C.T.O.", "signup_date": "20 May 2023", "col_x": ""}, {"name": "Bob Cruz", "state": "  az", "city": "WALDKIRCH", "phone": "5236917274", "job_title": "admin assistant", "signup_date": "3 Aug 2023", "col_x": ""}, {"name": "", "state": "", "city": "", "phone": "", "job_title": "", "signup_date": "", "col_x": ""}], "clean": [{"name": "Frank Petrov", "state": "Tennessee", "city": "Denver", "phone": "(516) 285-0682", "job_title": "Engineering Manager", "signup_date": "2023-05-14"}, {"name": "Eve Fischer", "state": "Arizona", "city": "Vladimir", "phone": "(530) 767-4379", "job_title": "Engineering Manager", "signup_date": "2023-08-21"}, {"name": "Bob Adams", "state": "Arizona", "city": "Vladimir", "phone": "(541) 640-2485", "job_title": "Chief Technology Officer", "signup_date": "2023-04-06"}, {"name": "Heidi Novak", "state": "Arizona", "city": "Vladimir", "phone": "(505) 182-0594", "job_title": "Director", "signup_date": "2023-09-01"}, {"name": "Heidi Diaz", "state": "Tennessee", "city": "Vladimir", "phone": "(503) 303-1174", "job_title": "Administrative Assistant", "signup_date": "2023-01-16"}, {"name": "Bob Johnson", "state": "Arizona", "city": "Denver", "phone": "(531) 101-5314", "job_title": "Director", "signup_date": "2023-05-25"}, {"name": "Carol Moore", "state": "Arizona", "city": "Toronto", "phone": "(585) 410-9072", "job_title": "Chief Technology Officer", "signup_date": "2023-01-15"}, {"name": "Priya Petrov", "state": "Arizona", "city": "Waldkirch", "phone": "(533) 847-9853", "job_title": "Chief Technology Officer", "signup_date": "2023-09-11"}, {"name": "Carol Petrov", "state": "Tennessee", "city": "Toronto", "phone": "(566) 851-7360", "job_title": "Engineering Manager", "signup_date": "2023-06-20"}, {"name": "Eve Reyes", "state": "Arizona", "city": "Denver", "phone": "(562) 555-3437", "job_title": "Chief Technology Officer", "signup_date": "2023-10-14"}, {"name": "Alice Park", "state": "Arizona", "city": "Vladimir", "phone": "(592) 300-9221", "job_title": "Director", "signup_date": "2023-12-02"}, {"name": "Judy Moore", "state": "Tennessee", "city": "Vladimir", "phone": "(515) 460-3035", "job_title": "Administrative Assistant", "signup_date": "2023-11-18"}, {"name": "Grace Moore", "state": "Tennessee", "city": "Waldkirch", "phone": "(572) 642-2385", "job_title": "Administrative Assistant", "signup_date": "2023-07-17"}, {"name": "Alice Novak", "state": "Arizona", "city": "Waldkirch", "phone": "(594) 222-8273", "job_title": "Administrative Assistant", "signup_date": "2023-06-02"}, {"name": "Priya Brandt", "state": "Arizona", "city": "Toronto", "phone": "(597) 364-0603", "job_title": "Director", "signup_date": "2023-04-11"}, {"name": "Eve Moore", "state": "Tennessee", "city": "Toronto", "phone": "(543) 891-0310", "job_title": "Administrative Assistant", "signup_date": "2023-08-27"}, {"name": "Ivan Khan", "state": "Arizona", "city": "Toronto", "phone": "(500) 775-5951", "job_title": "Chief Technology Officer", "signup_date": "2023-11-12"}, {"name": "Priya Fischer", "state": "Tennessee", "city": "Denver", "phone": "(591) 971-3917", "job_title": "Director", "signup_date": "2023-03-02"}, {"name": "Heidi Brandt", "state": "Tennessee", "city": "Denver", "phone": "(593) 298-6169", "job_title": "Chief Technology Officer", "signup_date": "2023-05-20"}, {"name": "Bob Cruz", "state": "Arizona", "city": "Waldkirch", "phone": "(523) 691-7274", "job_title": "Administrative Assistant", "signup_date": "2023-08-03"}], "dirty_cols": ["name", "state", "city", "phone", "job_title", "signup_date", "col_x"], "clean_cols": ["name", "state", "city", "phone", "job_title", "signup_date"], "plan": {"dataset_summary": "21 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"tennesese": "Tennessee", "ariznoa": "Arizona", "ARIIZONA": "Arizona", "tn": "Tennessee", "AZ": "Arizona", "tennessee": "Tennessee", "az": "Arizona", "tenessee": "Tennessee", "arizona": "Arizona", "Az": "Arizona"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"denver": "Denver", "Vladmir": "Vladimir", "WALDKIRCH": "Waldkirch", "toronto": "Toronto", "DENVER": "Denver"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Eng Mgr": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "dcr.": "Director", "admin assistant": "Administrative Assistant", "DIIR.": "Director", "cto": "Chief Technology Officer", "CTO": "Chief Technology Officer", "Admin": "Administrative Assistant", "ADMIN ASSISTANT": "Administrative Assistant", "Dir": "Director", "administrative assistant": "Administrative Assistant", "Dir.": "Director", "C.T.O.": "Chief Technology Officer"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"department": "Sales", "name": "Omar Park  ", "country": "Syrian Arnb Republic", "status": " WIP ", "industry": "E-Commerce", "is_active": "false"}, {"department": "Information Technology", "name": "Eve Khan", "country": "FO", "status": "Open", "industry": "Medical", "is_active": "TRUE"}, {"department": "sales", "name": "Eve Ali", "country": "   FO", "status": "new", "industry": "Retail & E-Commerce", "is_active": "F"}, {"department": "  pm", "name": "Sara Adams", "country": "SYRIAN ARAB  REPUBLIC", "status": "  open", "industry": "Healthcare", "is_active": "Yes"}, {"department": "Information Technology", "name": "Omar Fischer", "country": "Syrian Arab Republic", "status": "Open", "industry": " E-commerce ", "is_active": "TRUE"}, {"department": "I.T.", "name": "Bob Reyes", "country": "sy", "status": "ongoing   ", "industry": "   health", "is_active": "1"}, {"department": "It", "name": "Mona Fischer", "country": "Syrian Arab Republic", "status": "ongoing", "industry": "MEDICAL", "is_active": "FALSE"}, {"department": "   Biz Dev", "name": "David Petrov", "country": "SYR", "status": "CLOSED LOST", "industry": "Health", "is_active": "FALSE"}, {"department": "Fin", "name": "Omar Park", "country": "SY", "status": "open", "industry": "Healthcare", "is_active": "Y"}, {"department": "it", "name": "Priya Lee", "country": "SY", "status": "Lost", "industry": "retail", "is_active": "T"}, {"department": "I.T.", "name": "Ivan Reyes", "country": "Syrian Arab Republic", "status": " Won", "industry": " Healthcare ", "is_active": "No"}, {"department": "Biz Dev   ", "name": "Ivan Park", "country": "Faroe Islands", "status": "closed  lost", "industry": " E-commerce ", "is_active": "1"}, {"department": "Finance", "name": "   Frank Moore", "country": "FO", "status": "Lost", "industry": "HEALTH", "is_active": "Y"}, {"department": "sales", "name": "   Sara Novak", "country": "SY", "status": "closed-won", "industry": "RETAI L& E-COMMERCE", "is_active": "1"}, {"department": "FIN   ", "name": "Priya  Adams", "country": "   faroe islands", "status": "CLOSED WON", "industry": "RETAIL", "is_active": "Yes"}, {"department": "IT", "name": "David Smith  ", "country": " FO", "status": " lost ", "industry": "retail & e-commerce", "is_active": "true"}, {"department": "product", "name": "Bob  Reyes", "country": "syr", "status": "  Closed Lost", "industry": "  Reteil & E-commerce", "is_active": "Yes"}], "clean": [{"department": "Sales", "name": "Omar Park", "country": "Syrian Arab Republic", "status": "In Progress", "industry": "Retail", "is_active": false}, {"department": "Information Technology", "name": "Eve Khan", "country": "Faroe Islands", "status": "Open", "industry": "Healthcare", "is_active": true}, {"department": "Sales", "name": "Eve Ali", "country": "Faroe Islands", "status": "Open", "industry": "Retail", "is_active": false}, {"department": "Product", "name": "Sara Adams", "country": "Syrian Arab Republic", "status": "Open", "industry": "Healthcare", "is_active": true}, {"department": "Information Technology", "name": "Omar Fischer", "country": "Syrian Arab Republic", "status": "Open", "industry": "Retail", "is_active": true}, {"department": "Information Technology", "name": "Bob Reyes", "country": "Syrian Arab Republic", "status": "In Progress", "industry": "Healthcare", "is_active": true}, {"department": "Information Technology", "name": "Mona Fischer", "country": "Syrian Arab Republic", "status": "In Progress", "industry": "Healthcare", "is_active": false}, {"department": "Sales", "name": "David Petrov", "country": "Syrian Arab Republic", "status": "Lost", "industry": "Healthcare", "is_active": false}, {"department": "Finance", "name": "Omar Park", "country": "Syrian Arab Republic", "status": "Open", "industry": "Healthcare", "is_active": true}, {"department": "Information Technology", "name": "Priya Lee", "country": "Syrian Arab Republic", "status": "Lost", "industry": "Retail", "is_active": true}, {"department": "Information Technology", "name": "Ivan Reyes", "country": "Syrian Arab Republic", "status": "Won", "industry": "Healthcare", "is_active": false}, {"department": "Sales", "name": "Ivan Park", "country": "Faroe Islands", "status": "Lost", "industry": "Retail", "is_active": true}, {"department": "Finance", "name": "Frank Moore", "country": "Faroe Islands", "status": "Lost", "industry": "Healthcare", "is_active": true}, {"department": "Sales", "name": "Sara Novak", "country": "Syrian Arab Republic", "status": "Won", "industry": "Retail", "is_active": true}, {"department": "Finance", "name": "Priya Adams", "country": "Faroe Islands", "status": "Won", "industry": "Retail", "is_active": true}, {"department": "Information Technology", "name": "David Smith", "country": "Faroe Islands", "status": "Lost", "industry": "Retail", "is_active": true}, {"department": "Product", "name": "Bob Reyes", "country": "Syrian Arab Republic", "status": "Lost", "industry": "Retail", "is_active": true}], "dirty_cols": ["department", "name", "country", "status", "industry", "is_active"], "clean_cols": ["department", "name", "country", "status", "industry", "is_active"], "plan": {"dataset_summary": "17 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sales": "Sales", "pm": "Product", "I.T.": "Information Technology", "It": "Information Technology", "Biz Dev": "Sales", "Fin": "Finance", "it": "Information Technology", "FIN": "Finance", "IT": "Information Technology", "product": "Product"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Syrian Arnb Republic": "Syrian Arab Republic", "FO": "Faroe Islands", "SYRIAN ARAB REPUBLIC": "Syrian Arab Republic", "sy": "Syrian Arab Republic", "SYR": "Syrian Arab Republic", "SY": "Syrian Arab Republic", "faroe islands": "Faroe Islands", "syr": "Syrian Arab Republic"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WIP": "In Progress", "new": "Open", "open": "Open", "ongoing": "In Progress", "CLOSED LOST": "Lost", "closed lost": "Lost", "closed-won": "Won", "CLOSED WON": "Won", "lost": "Lost", "Closed Lost": "Lost"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"E-Commerce": "Retail", "Medical": "Healthcare", "Retail & E-Commerce": "Retail", "E-commerce": "Retail", "health": "Healthcare", "MEDICAL": "Healthcare", "Health": "Healthcare", "retail": "Retail", "HEALTH": "Healthcare", "RETAI L& E-COMMERCE": "Retail", "RETAIL": "Retail", "retail & e-commerce": "Retail", "Reteil & E-commerce": "Retail"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"name": "Mona Moore", "status": "Medium", "company": " Vehement ", "is_active": "Yes", "extra": ""}, {"name": "Carol Smith", "status": "P3", "company": "Acme Inc", "is_active": "F", "extra": ""}, {"name": "Ivan  Cruz", "status": "high", "company": "Umbrella  ", "is_active": "Yes", "extra": ""}, {"name": " Judy Diaz", "status": "p1  ", "company": "Stark  Industries", "is_active": "No", "extra": ""}, {"name": "Frank Petrov", "status": "Medium", "company": "Cyberdyne", "is_active": "TRUE", "extra": ""}, {"name": "Judy Reyes", "status": "HGIH", "company": " Umbrella", "is_active": "TRUE", "extra": ""}, {"name": "Carol Smith", "status": "medium", "company": "  Cyberdyne", "is_active": "N", "extra": ""}, {"name": "Priya Wong", "status": "   P1", "company": "Cyberdyne   ", "is_active": "FALSE", "extra": ""}, {"name": "Mona Ali  ", "status": "H", "company": "Stark Industries ", "is_active": "1", "extra": ""}, {"name": "Mona Cruz ", "status": "  med", "company": "Acme Inc  ", "is_active": "F", "extra": ""}, {"name": "Bob  Novak", "status": "HIGH", "company": "Acme Inc", "is_active": "FALSE", "extra": ""}, {"name": "David Moore", "status": "P2", "company": " Umbrella ", "is_active": "false", "extra": ""}, {"name": "Eve Lee", "status": " H ", "company": "Hooli", "is_active": "T", "extra": ""}, {"name": "David  Park", "status": "HUGH", "company": " Wonka Co", "is_active": "N", "extra": ""}], "clean": [{"name": "Mona Moore", "status": "Medium", "company": "Vehement", "is_active": true}, {"name": "Carol Smith", "status": "Low", "company": "Acme Inc", "is_active": false}, {"name": "Ivan Cruz", "status": "High", "company": "Umbrella", "is_active": true}, {"name": "Judy Diaz", "status": "High", "company": "Stark Industries", "is_active": false}, {"name": "Frank Petrov", "status": "Medium", "company": "Cyberdyne", "is_active": true}, {"name": "Judy Reyes", "status": "High", "company": "Umbrella", "is_active": true}, {"name": "Carol Smith", "status": "Medium", "company": "Cyberdyne", "is_active": false}, {"name": "Priya Wong", "status": "High", "company": "Cyberdyne", "is_active": false}, {"name": "Mona Ali", "status": "High", "company": "Stark Industries", "is_active": true}, {"name": "Mona Cruz", "status": "Medium", "company": "Acme Inc", "is_active": false}, {"name": "Bob Novak", "status": "High", "company": "Acme Inc", "is_active": false}, {"name": "David Moore", "status": "Medium", "company": "Umbrella", "is_active": false}, {"name": "Eve Lee", "status": "High", "company": "Hooli", "is_active": true}, {"name": "David Park", "status": "High", "company": "Wonka Co", "is_active": false}], "dirty_cols": ["name", "status", "company", "is_active", "extra"], "clean_cols": ["name", "status", "company", "is_active"], "plan": {"dataset_summary": "14 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P3": "Low", "high": "High", "p1": "High", "HGIH": "High", "medium": "Medium", "P1": "High", "H": "High", "med": "Medium", "HIGH": "High", "P2": "Medium", "HUGH": "High"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"unit": " In", "country": "people's republic of bangladesh", "phone": "543-021-4566", "signup_date": "45143", "job_title": "Adminestrative Assistant   ", "email": "nmfn@test.org"}, {"unit": "Kilometers", "country": "bgd", "phone": "516-781-0928", "signup_date": "2023-07-17", "job_title": "CHIEF TECHNOLOGY OFFICER", "email": "kcmcd@corp.io"}, {"unit": "mL", "country": "bd", "phone": "504-195-9632", "signup_date": "45193", "job_title": "chief technology officer", "email": "icgo@mail.com   "}, {"unit": "inches", "country": "BD", "phone": "500.761.3739", "signup_date": "26 Sep 2023", "job_title": "engineering mgr", "email": "MFHFBEG@CORP.IO"}, {"unit": "Km", "country": "PEOPLES REPUBLIC OF BANGLADESH", "phone": "536-084-8536", "signup_date": "45092", "job_title": "Engineering Manager  ", "email": "pjbh@corp.io"}, {"unit": "l", "country": "FIN", "phone": "5179689564", "signup_date": "10/16/2023", "job_title": "ADMIN ASST", "email": " PAOGPHF@TEST.ORG "}, {"unit": "mls", "country": "PEOPLES REPUBLIC OF BANGLADESH", "phone": "(557)620-4428", "signup_date": "12/19/2023", "job_title": "Eng Manager", "email": "hdnki@corp.io  "}, {"unit": "kplometer", "country": "finland", "phone": "5177329251", "signup_date": "6/10/2023", "job_title": " ADMIN ", "email": "MKMAMB@EXAMPLE.COM"}, {"unit": "  l", "country": "fin   ", "phone": "(537)778-7166", "signup_date": "6/17/2023", "job_title": "engineering manager", "email": "ecpka@example.com"}, {"unit": "l", "country": "Republic of Finland", "phone": "(598)643-1373", "signup_date": "44996", "job_title": "Admin Assistant", "email": " HOOH@CORP.IO "}, {"unit": "Inch", "country": "BGD   ", "phone": "544-455-3245", "signup_date": "45177", "job_title": "Eng Manager", "email": "aiool@corp.io"}, {"unit": "  in", "country": " fin ", "phone": "(597)492-6713", "signup_date": "45093", "job_title": "administrative assistant", "email": "NLMHDA@TEST.ORG"}, {"unit": "inchjs", "country": "Bangladesh", "phone": "549.465.1226", "signup_date": "6/16/2023", "job_title": "AdminA sst", "email": "cpoaioo@test.org"}, {"unit": "l", "country": "Fin", "phone": "562-956-1158", "signup_date": "26 Sep 2023", "job_title": "Engineering Mgr", "email": "FHFOHG@TEST.ORG "}, {"unit": "inch  ", "country": "Republic of Finland", "phone": "528.804.0496", "signup_date": "44980", "job_title": "cto", "email": "EPCM@MAIL.COM"}, {"unit": "incches", "country": "Republic ofFinland", "phone": "(580)456-5009", "signup_date": "2 Jan 2023", "job_title": "cto", "email": "PNEPG@EXAMPLE.COM  "}, {"unit": "inches", "country": " fin ", "phone": "562.757.5022", "signup_date": "45003", "job_title": "ADMIN", "email": "LJIDBM@TEST.ORG"}, {"unit": "inches", "country": "finland", "phone": "(581)938-8658", "signup_date": "45219", "job_title": "CHIEF TECHNOLOGY OFFICER   ", "email": "HELEB@TEST.ORG"}, {"unit": "inches", "country": "People'S Republic Of Bangladesh", "phone": "(533)364-6860", "signup_date": "44999", "job_title": " ADMINISTRATIVE ASSISTANT", "email": "FOMB@MAIL.COM"}, {"unit": "INCH", "country": " bd ", "phone": "(546)285-0553", "signup_date": "1/14/2023", "job_title": "eng manager   ", "email": "jdjbhk@test.org"}, {"unit": "inches", "country": "FI", "phone": "537.022.9225", "signup_date": "10/9/2023", "job_title": "Eng Mgr", "email": "   lcaocg@test.org"}, {"unit": "litre", "country": "BGD", "phone": "(560)831-0786", "signup_date": "17 Dec 2023", "job_title": "chief technology officer", "email": "AJNNHK@MAIL.COM"}, {"unit": "\"", "country": " fi ", "phone": "5978920190", "signup_date": "10/20/2023", "job_title": "Eng Mgr", "email": "NPKDFFD@MAIL.COM"}, {"unit": "MLS", "country": "bd", "phone": "581-809-1740", "signup_date": "9/8/2023", "job_title": "c.t.o.", "email": "   dnmfba@example.com"}, {"unit": "Inches", "country": "Bangladesh", "phone": "5059442776", "signup_date": "5 Nov 2023", "job_title": "C.T.O.", "email": "DAJKCKE@EXAMPLE.COM "}, {"unit": "km", "country": "Bgd", "phone": "5129291686", "signup_date": "44981", "job_title": "eng manager", "email": "LBHIA@MAIL.COM"}, {"unit": " liteers", "country": "FI", "phone": "555.154.9145", "signup_date": "45280", "job_title": "chief technology officer  ", "email": "iefpgmc@test.org"}, {"unit": "IN", "country": "People's Republic of Bngladesh", "phone": "589-675-9521", "signup_date": "10 Nov 2023", "job_title": "CTO", "email": " ihnpe@test.org "}, {"unit": "indh ", "country": "FIN", "phone": "(529)538-9158", "signup_date": "2023-01-10", "job_title": "  chief technology officer", "email": "iimifkn@corp.io"}, {"unit": "In ", "country": "bangladesh  ", "phone": "555-054-6879", "signup_date": "44928", "job_title": "C.T.O.", "email": "CLGMP@EXAMPLE.COM"}, {"unit": "Liters", "country": "Finland", "phone": "510.506.6285", "signup_date": "11/10/2023", "job_title": "Admin Assistant", "email": "MFML@TEST.ORG "}, {"unit": "Km", "country": "FI", "phone": "549.806.4647", "signup_date": "2023-01-16", "job_title": " C.T.O. ", "email": "djlehle@test.org"}, {"unit": "KILOMETER   ", "country": "fin", "phone": "5128392329", "signup_date": "45121", "job_title": "engineering manager ", "email": "LPIFKAB@TEST.ORG"}, {"unit": "  inches", "country": "FIN", "phone": "(572)969-7254", "signup_date": "18 Aug 2023", "job_title": "CHIEF TECHNOLOGY OFFICER", "email": " hckcnk@corp.io "}, {"unit": "Inches", "country": "Bangladesh", "phone": "5059442776", "signup_date": "5 Nov 2023", "job_title": "C.T.O.", "email": "DAJKCKE@EXAMPLE.COM "}, {"unit": "  l", "country": "fin   ", "phone": "(537)778-7166", "signup_date": "6/17/2023", "job_title": "engineering manager", "email": "ecpka@example.com"}], "clean": [{"unit": "in", "country": "Bangladesh", "phone": "(543) 021-4566", "signup_date": "2023-08-05", "job_title": "Administrative Assistant", "email": "nmfn@test.org"}, {"unit": "km", "country": "Bangladesh", "phone": "(516) 781-0928", "signup_date": "2023-07-17", "job_title": "Chief Technology Officer", "email": "kcmcd@corp.io"}, {"unit": "mL", "country": "Bangladesh", "phone": "(504) 195-9632", "signup_date": "2023-09-24", "job_title": "Chief Technology Officer", "email": "icgo@mail.com"}, {"unit": "in", "country": "Bangladesh", "phone": "(500) 761-3739", "signup_date": "2023-09-26", "job_title": "Engineering Manager", "email": "mfhfbeg@corp.io"}, {"unit": "km", "country": "Bangladesh", "phone": "(536) 084-8536", "signup_date": "2023-06-15", "job_title": "Engineering Manager", "email": "pjbh@corp.io"}, {"unit": "L", "country": "Finland", "phone": "(517) 968-9564", "signup_date": "2023-10-16", "job_title": "Administrative Assistant", "email": "paogphf@test.org"}, {"unit": "mL", "country": "Bangladesh", "phone": "(557) 620-4428", "signup_date": "2023-12-19", "job_title": "Engineering Manager", "email": "hdnki@corp.io"}, {"unit": "km", "country": "Finland", "phone": "(517) 732-9251", "signup_date": "2023-06-10", "job_title": "Administrative Assistant", "email": "mkmamb@example.com"}, {"unit": "L", "country": "Finland", "phone": "(537) 778-7166", "signup_date": "2023-06-17", "job_title": "Engineering Manager", "email": "ecpka@example.com"}, {"unit": "L", "country": "Finland", "phone": "(598) 643-1373", "signup_date": "2023-03-11", "job_title": "Administrative Assistant", "email": "hooh@corp.io"}, {"unit": "in", "country": "Bangladesh", "phone": "(544) 455-3245", "signup_date": "2023-09-08", "job_title": "Engineering Manager", "email": "aiool@corp.io"}, {"unit": "in", "country": "Finland", "phone": "(597) 492-6713", "signup_date": "2023-06-16", "job_title": "Administrative Assistant", "email": "nlmhda@test.org"}, {"unit": "in", "country": "Bangladesh", "phone": "(549) 465-1226", "signup_date": "2023-06-16", "job_title": "Administrative Assistant", "email": "cpoaioo@test.org"}, {"unit": "L", "country": "Finland", "phone": "(562) 956-1158", "signup_date": "2023-09-26", "job_title": "Engineering Manager", "email": "fhfohg@test.org"}, {"unit": "in", "country": "Finland", "phone": "(528) 804-0496", "signup_date": "2023-02-23", "job_title": "Chief Technology Officer", "email": "epcm@mail.com"}, {"unit": "in", "country": "Finland", "phone": "(580) 456-5009", "signup_date": "2023-01-02", "job_title": "Chief Technology Officer", "email": "pnepg@example.com"}, {"unit": "in", "country": "Finland", "phone": "(562) 757-5022", "signup_date": "2023-03-18", "job_title": "Administrative Assistant", "email": "ljidbm@test.org"}, {"unit": "in", "country": "Finland", "phone": "(581) 938-8658", "signup_date": "2023-10-20", "job_title": "Chief Technology Officer", "email": "heleb@test.org"}, {"unit": "in", "country": "Bangladesh", "phone": "(533) 364-6860", "signup_date": "2023-03-14", "job_title": "Administrative Assistant", "email": "fomb@mail.com"}, {"unit": "in", "country": "Bangladesh", "phone": "(546) 285-0553", "signup_date": "2023-01-14", "job_title": "Engineering Manager", "email": "jdjbhk@test.org"}, {"unit": "in", "country": "Finland", "phone": "(537) 022-9225", "signup_date": "2023-10-09", "job_title": "Engineering Manager", "email": "lcaocg@test.org"}, {"unit": "L", "country": "Bangladesh", "phone": "(560) 831-0786", "signup_date": "2023-12-17", "job_title": "Chief Technology Officer", "email": "ajnnhk@mail.com"}, {"unit": "in", "country": "Finland", "phone": "(597) 892-0190", "signup_date": "2023-10-20", "job_title": "Engineering Manager", "email": "npkdffd@mail.com"}, {"unit": "mL", "country": "Bangladesh", "phone": "(581) 809-1740", "signup_date": "2023-09-08", "job_title": "Chief Technology Officer", "email": "dnmfba@example.com"}, {"unit": "in", "country": "Bangladesh", "phone": "(505) 944-2776", "signup_date": "2023-11-05", "job_title": "Chief Technology Officer", "email": "dajkcke@example.com"}, {"unit": "km", "country": "Bangladesh", "phone": "(512) 929-1686", "signup_date": "2023-02-24", "job_title": "Engineering Manager", "email": "lbhia@mail.com"}, {"unit": "L", "country": "Finland", "phone": "(555) 154-9145", "signup_date": "2023-12-20", "job_title": "Chief Technology Officer", "email": "iefpgmc@test.org"}, {"unit": "in", "country": "Bangladesh", "phone": "(589) 675-9521", "signup_date": "2023-11-10", "job_title": "Chief Technology Officer", "email": "ihnpe@test.org"}, {"unit": "in", "country": "Finland", "phone": "(529) 538-9158", "signup_date": "2023-01-10", "job_title": "Chief Technology Officer", "email": "iimifkn@corp.io"}, {"unit": "in", "country": "Bangladesh", "phone": "(555) 054-6879", "signup_date": "2023-01-02", "job_title": "Chief Technology Officer", "email": "clgmp@example.com"}, {"unit": "L", "country": "Finland", "phone": "(510) 506-6285", "signup_date": "2023-11-10", "job_title": "Administrative Assistant", "email": "mfml@test.org"}, {"unit": "km", "country": "Finland", "phone": "(549) 806-4647", "signup_date": "2023-01-16", "job_title": "Chief Technology Officer", "email": "djlehle@test.org"}, {"unit": "km", "country": "Finland", "phone": "(512) 839-2329", "signup_date": "2023-07-14", "job_title": "Engineering Manager", "email": "lpifkab@test.org"}, {"unit": "in", "country": "Finland", "phone": "(572) 969-7254", "signup_date": "2023-08-18", "job_title": "Chief Technology Officer", "email": "hckcnk@corp.io"}], "dirty_cols": ["unit", "country", "phone", "signup_date", "job_title", "email"], "clean_cols": ["unit", "country", "phone", "signup_date", "job_title", "email"], "plan": {"dataset_summary": "36 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"In": "in", "Kilometers": "km", "inches": "in", "Km": "km", "l": "L", "mls": "mL", "kplometer": "km", "Inch": "in", "inchjs": "in", "inch": "in", "incches": "in", "INCH": "in", "litre": "L", "\"": "in", "MLS": "mL", "Inches": "in", "liteers": "L", "IN": "in", "indh": "in", "Liters": "L", "KILOMETER": "km"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"people's republic of bangladesh": "Bangladesh", "bgd": "Bangladesh", "bd": "Bangladesh", "BD": "Bangladesh", "PEOPLES REPUBLIC OF BANGLADESH": "Bangladesh", "FIN": "Finland", "finland": "Finland", "fin": "Finland", "Republic of Finland": "Finland", "BGD": "Bangladesh", "Fin": "Finland", "Republic ofFinland": "Finland", "People'S Republic Of Bangladesh": "Bangladesh", "FI": "Finland", "fi": "Finland", "Bgd": "Bangladesh", "People's Republic of Bngladesh": "Bangladesh", "bangladesh": "Bangladesh"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Adminestrative Assistant": "Administrative Assistant", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "engineering mgr": "Engineering Manager", "ADMIN ASST": "Administrative Assistant", "Eng Manager": "Engineering Manager", "ADMIN": "Administrative Assistant", "engineering manager": "Engineering Manager", "Admin Assistant": "Administrative Assistant", "administrative assistant": "Administrative Assistant", "AdminA sst": "Administrative Assistant", "Engineering Mgr": "Engineering Manager", "cto": "Chief Technology Officer", "ADMINISTRATIVE ASSISTANT": "Administrative Assistant", "eng manager": "Engineering Manager", "Eng Mgr": "Engineering Manager", "c.t.o.": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "CTO": "Chief Technology Officer"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"status": "paid", "amount": "$2,204.83", "is_active": "Y", "unit": "\" ", "rate": "73.8%", "name": " David Reyes"}, {"status": "refund", "amount": "3.731,15", "is_active": "0", "unit": "mile", "rate": "11.2%", "name": "Frank Cruz"}, {"status": " refunded ", "amount": "7.208,66", "is_active": "N", "unit": "Kg", "rate": "0.4%", "name": "  Priya Ali"}, {"status": "late", "amount": "8.474,18", "is_active": "true", "unit": "mile", "rate": "12.5%", "name": "Karl Adams"}, {"status": "REFUNDED", "amount": "3.907,60", "is_active": "TRUE", "unit": "inch", "rate": "71.4%", "name": "Karl  Petrov"}, {"status": "Refunded", "amount": "$1,280.85", "is_active": "1", "unit": "Kg", "rate": "86.3%", "name": "   Heidi Diaz"}, {"status": "  Overwue", "amount": "2.464,10", "is_active": "1", "unit": "Kilogrms", "rate": "37.9%", "name": "  Alice Wong"}, {"status": "refund", "amount": "1.108,50", "is_active": "TRUE", "unit": "kilograms", "rate": "25.4%", "name": "David Cruz"}, {"status": " paid ", "amount": "$5,697.25", "is_active": "Y", "unit": "inch", "rate": "44.4%", "name": "Heidi Adams"}, {"status": " refunded ", "amount": "4.225,05", "is_active": "No", "unit": "Kg", "rate": "65.3%", "name": "Frank  Moore"}, {"status": "paid", "amount": "$2,204.83", "is_active": "Y", "unit": "\" ", "rate": "73.8%", "name": " David Reyes"}], "clean": [{"status": "Paid", "amount": 2204.83, "is_active": true, "unit": "in", "rate": 0.738, "name": "David Reyes"}, {"status": "Refunded", "amount": 3731.15, "is_active": false, "unit": "mi", "rate": 0.11199999999999999, "name": "Frank Cruz"}, {"status": "Refunded", "amount": 7208.66, "is_active": false, "unit": "kg", "rate": 0.004, "name": "Priya Ali"}, {"status": "Overdue", "amount": 8474.18, "is_active": true, "unit": "mi", "rate": 0.125, "name": "Karl Adams"}, {"status": "Refunded", "amount": 3907.6, "is_active": true, "unit": "in", "rate": 0.7140000000000001, "name": "Karl Petrov"}, {"status": "Refunded", "amount": 1280.85, "is_active": true, "unit": "kg", "rate": 0.863, "name": "Heidi Diaz"}, {"status": "Overdue", "amount": 2464.1, "is_active": true, "unit": "kg", "rate": 0.379, "name": "Alice Wong"}, {"status": "Refunded", "amount": 1108.5, "is_active": true, "unit": "kg", "rate": 0.254, "name": "David Cruz"}, {"status": "Paid", "amount": 5697.25, "is_active": true, "unit": "in", "rate": 0.444, "name": "Heidi Adams"}, {"status": "Refunded", "amount": 4225.05, "is_active": false, "unit": "kg", "rate": 0.653, "name": "Frank Moore"}], "dirty_cols": ["status", "amount", "is_active", "unit", "rate", "name"], "clean_cols": ["status", "amount", "is_active", "unit", "rate", "name"], "plan": {"dataset_summary": "11 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paid": "Paid", "refund": "Refunded", "refunded": "Refunded", "late": "Overdue", "REFUNDED": "Refunded", "Overwue": "Overdue"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"\"": "in", "mile": "mi", "Kg": "kg", "inch": "in", "Kilogrms": "kg", "kilograms": "kg"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"company": "Globex", "rate": "50.6%", "email": "jhljaf@example.com", "city": "   LONDON", "country": "Gum"}, {"company": "Globex", "rate": "12.7%", "email": "abkbpp@mail.com", "city": "ldn", "country": "Republic of Cyprus"}, {"company": "Stark Industries ", "rate": "86.8%", "email": "NOKLI@CORP.IO", "city": "toronto", "country": "GU"}, {"company": "   Vehement", "rate": "53.9%", "email": "   FFAG@EXAMPLE.COM", "city": "Toronto", "country": "   republic of cyprus"}, {"company": "Acme Inc", "rate": "27.5%", "email": "GJOINLB@TEST.ORG", "city": "LDN", "country": " GU "}, {"company": "Cyberdyne", "rate": "50.5%", "email": " ppcee@mail.com ", "city": "Toronto", "country": "Cy"}, {"company": "Acme Inc", "rate": "21.9%", "email": "  ACPIH@TEST.ORG", "city": "TRONTO", "country": "CYPRRUS"}, {"company": "Stark  Industries", "rate": "40.1%", "email": " ojpce@mail.com ", "city": "toronto", "country": "cyp"}, {"company": "Stark Industries   ", "rate": "58.4%", "email": "GMEA@MAIL.COM", "city": "London", "country": " CYP"}, {"company": "Soylent  Corp", "rate": "11.4%", "email": "imadd@mail.com", "city": "Toronto", "country": "REPBLIC OF CYPRUS"}, {"company": "Vehement", "rate": "51.1%", "email": "HMON@MAIL.COM   ", "city": "LDN", "country": "GU"}, {"company": "", "rate": "", "email": "", "city": "", "country": ""}], "clean": [{"company": "Globex", "rate": 0.506, "email": "jhljaf@example.com", "city": "London", "country": "Guam"}, {"company": "Globex", "rate": 0.127, "email": "abkbpp@mail.com", "city": "London", "country": "Cyprus"}, {"company": "Stark Industries", "rate": 0.868, "email": "nokli@corp.io", "city": "Toronto", "country": "Guam"}, {"company": "Vehement", "rate": 0.539, "email": "ffag@example.com", "city": "Toronto", "country": "Cyprus"}, {"company": "Acme Inc", "rate": 0.275, "email": "gjoinlb@test.org", "city": "London", "country": "Guam"}, {"company": "Cyberdyne", "rate": 0.505, "email": "ppcee@mail.com", "city": "Toronto", "country": "Cyprus"}, {"company": "Acme Inc", "rate": 0.21899999999999997, "email": "acpih@test.org", "city": "Toronto", "country": "Cyprus"}, {"company": "Stark Industries", "rate": 0.401, "email": "ojpce@mail.com", "city": "Toronto", "country": "Cyprus"}, {"company": "Stark Industries", "rate": 0.584, "email": "gmea@mail.com", "city": "London", "country": "Cyprus"}, {"company": "Soylent Corp", "rate": 0.114, "email": "imadd@mail.com", "city": "Toronto", "country": "Cyprus"}, {"company": "Vehement", "rate": 0.511, "email": "hmon@mail.com", "city": "London", "country": "Guam"}], "dirty_cols": ["company", "rate", "email", "city", "country"], "clean_cols": ["company", "rate", "email", "city", "country"], "plan": {"dataset_summary": "12 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LONDON": "London", "ldn": "London", "toronto": "Toronto", "LDN": "London", "TRONTO": "Toronto"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Gum": "Guam", "Republic of Cyprus": "Cyprus", "GU": "Guam", "republic of cyprus": "Cyprus", "Cy": "Cyprus", "CYPRRUS": "Cyprus", "cyp": "Cyprus", "CYP": "Cyprus", "REPBLIC OF CYPRUS": "Cyprus"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "522.043.3591", "name": "Karl Novak", "status": "pending", "signup_date": "2023-12-26", "email": "gimdb@test.org", "currency": "bmd", "company": "   Acme Inc", "is_active": "TRUE"}, {"phone": "5406658844", "name": "Frank  Park", "status": "Refunded", "signup_date": "9 Mar 2023", "email": "PBCJKC@CORP.IO", "currency": "Bahamian Dollar", "company": "Initech", "is_active": "true"}, {"phone": "(518)425-3338", "name": "Heidi Adams", "status": "refunded", "signup_date": "11/26/2023", "email": "abmegkf@example.com", "currency": "BSD", "company": "Acme Inc", "is_active": "TRUE"}, {"phone": "(520)075-9351", "name": "Ivan Smith   ", "status": "OVERDUE", "signup_date": "2023-12-15", "email": "ONJBBN@TEST.ORG", "currency": "Bahamian Dollar", "company": "Hooli", "is_active": "Y"}, {"phone": "559.673.7995", "name": "Ivan Cruz", "status": "late", "signup_date": "11/8/2023", "email": "edkb@mail.com", "currency": "Dalasi", "company": "Umbrella", "is_active": "FALSE"}, {"phone": "561.880.8256", "name": "Carol Petrov", "status": "Refunded", "signup_date": "2023-05-18", "email": "mcafapc@corp.io", "currency": "BMD", "company": "Acme Inc  ", "is_active": "No"}, {"phone": "5194565785", "name": "Carol  Novak", "status": " refunded ", "signup_date": "1/13/2023", "email": "hnmfh@mail.com", "currency": "Bahamian Dollar", "company": " Globex ", "is_active": "false"}, {"phone": "503-610-0343", "name": "Grace Diaz   ", "status": " refunded", "signup_date": "19 Dec 2023", "email": "ompdmdm@test.org", "currency": "GMD", "company": "Globex   ", "is_active": "0"}, {"phone": "(595)567-3054", "name": "Carol  Smith", "status": "Refund", "signup_date": "22 May 2023", "email": "BEDGLI@CORP.IO", "currency": "BMD", "company": "Acme Inc ", "is_active": "N"}, {"phone": "581-031-4342", "name": "Heidi Lee", "status": " Overdue", "signup_date": "2023-08-16", "email": "bfibnf@mail.com", "currency": "   Bsd", "company": "Stark Industries", "is_active": "Yes"}, {"phone": "530-641-3008", "name": "Sara Novak", "status": "ovedue", "signup_date": "5 Nov 2023", "email": "  eabfai@mail.com", "currency": "bmd   ", "company": "Hooli", "is_active": "Yes"}, {"phone": "581.012.7719", "name": "Ivan Reyes ", "status": "Unyaid", "signup_date": "2023-02-24", "email": " HKGADFF@EXAMPLE.COM ", "currency": "Dalasi", "company": "Globex", "is_active": "No"}, {"phone": "(534)655-5418", "name": "Alice Petrov", "status": "laet", "signup_date": "2023-01-16", "email": "CEGC@TEST.ORG", "currency": "Dalasi", "company": "Umbrella", "is_active": "T"}, {"phone": "585-125-5236", "name": "  Frank Brandt", "status": "Pending", "signup_date": "45081", "email": "   BCJFMLH@EXAMPLE.COM", "currency": "bahamina dollar", "company": "Acme Inc ", "is_active": "0"}, {"phone": "538-726-3884", "name": "Alice  Park", "status": "late  ", "signup_date": "45264", "email": "CDOAM@MAIL.COM", "currency": "Bahamian Dollar", "company": "Globex  ", "is_active": "Yes"}, {"phone": "(565)030-2074", "name": "Ivan Khan", "status": "DUE", "signup_date": "45086", "email": " GKCHN@CORP.IO ", "currency": "GMD", "company": "  Vehement", "is_active": "true"}, {"phone": "5163109090", "name": "Grace  Brandt", "status": "refunded", "signup_date": "27 Sep 2023", "email": "   AINM@MAIL.COM", "currency": "BMD", "company": "Globex", "is_active": "T"}, {"phone": "(552)582-8986", "name": "Lena Reyes", "status": "unpaid ", "signup_date": "5/13/2023", "email": "  CCDE@TEST.ORG", "currency": "  bmd", "company": "Soylent Corp", "is_active": "N"}, {"phone": "507.797.3243", "name": "Lena Wong", "status": "pending", "signup_date": "11/12/2023", "email": "BLLEEN@MAIL.COM", "currency": " BSD", "company": "Umbrella  ", "is_active": "T"}, {"phone": "535.997.8877", "name": "Eve Cruz", "status": "refunded", "signup_date": "16 Jun 2023", "email": "hcgk@corp.io", "currency": " bahamian dollar", "company": "Vehement", "is_active": "FALSE"}, {"phone": "575.204.7916", "name": "Eve Cruz", "status": "refunded", "signup_date": "45280", "email": " DNIDHJL@MAIL.COM ", "currency": "Bahamian Dollar", "company": "Cyberdyne", "is_active": "true"}, {"phone": "570-940-1646", "name": "Heidi  Petrov", "status": "rfund   ", "signup_date": "7 May 2023", "email": "LAMEHD@EXAMPLE.COM", "currency": "BSD", "company": "Initech ", "is_active": "F"}, {"phone": "5101999012", "name": "Heidi Ali", "status": "refund   ", "signup_date": "2023-05-01", "email": " ljekegf@test.org ", "currency": "Bmd", "company": "Globex", "is_active": "true"}, {"phone": "591-834-3544", "name": "Priya Lee  ", "status": "overdue", "signup_date": "3 Apr 2023", "email": "jheag@example.com", "currency": "GMD", "company": "Wonka Co", "is_active": "0"}, {"phone": "(543)844-7587", "name": "Bob Khan", "status": "Pending", "signup_date": "22 Dec 2023", "email": "bjcj@corp.io ", "currency": " GMD", "company": "  Wonka Co", "is_active": "TRUE"}, {"phone": "", "name": "", "status": "", "signup_date": "", "email": "", "currency": "", "company": "", "is_active": ""}], "clean": [{"phone": "(522) 043-3591", "name": "Karl Novak", "status": "Pending", "signup_date": "2023-12-26", "email": "gimdb@test.org", "currency": "BMD", "company": "Acme Inc", "is_active": true}, {"phone": "(540) 665-8844", "name": "Frank Park", "status": "Refunded", "signup_date": "2023-03-09", "email": "pbcjkc@corp.io", "currency": "BSD", "company": "Initech", "is_active": true}, {"phone": "(518) 425-3338", "name": "Heidi Adams", "status": "Refunded", "signup_date": "2023-11-26", "email": "abmegkf@example.com", "currency": "BSD", "company": "Acme Inc", "is_active": true}, {"phone": "(520) 075-9351", "name": "Ivan Smith", "status": "Overdue", "signup_date": "2023-12-15", "email": "onjbbn@test.org", "currency": "BSD", "company": "Hooli", "is_active": true}, {"phone": "(559) 673-7995", "name": "Ivan Cruz", "status": "Overdue", "signup_date": "2023-11-08", "email": "edkb@mail.com", "currency": "GMD", "company": "Umbrella", "is_active": false}, {"phone": "(561) 880-8256", "name": "Carol Petrov", "status": "Refunded", "signup_date": "2023-05-18", "email": "mcafapc@corp.io", "currency": "BMD", "company": "Acme Inc", "is_active": false}, {"phone": "(519) 456-5785", "name": "Carol Novak", "status": "Refunded", "signup_date": "2023-01-13", "email": "hnmfh@mail.com", "currency": "BSD", "company": "Globex", "is_active": false}, {"phone": "(503) 610-0343", "name": "Grace Diaz", "status": "Refunded", "signup_date": "2023-12-19", "email": "ompdmdm@test.org", "currency": "GMD", "company": "Globex", "is_active": false}, {"phone": "(595) 567-3054", "name": "Carol Smith", "status": "Refunded", "signup_date": "2023-05-22", "email": "bedgli@corp.io", "currency": "BMD", "company": "Acme Inc", "is_active": false}, {"phone": "(581) 031-4342", "name": "Heidi Lee", "status": "Overdue", "signup_date": "2023-08-16", "email": "bfibnf@mail.com", "currency": "BSD", "company": "Stark Industries", "is_active": true}, {"phone": "(530) 641-3008", "name": "Sara Novak", "status": "Overdue", "signup_date": "2023-11-05", "email": "eabfai@mail.com", "currency": "BMD", "company": "Hooli", "is_active": true}, {"phone": "(581) 012-7719", "name": "Ivan Reyes", "status": "Pending", "signup_date": "2023-02-24", "email": "hkgadff@example.com", "currency": "GMD", "company": "Globex", "is_active": false}, {"phone": "(534) 655-5418", "name": "Alice Petrov", "status": "Overdue", "signup_date": "2023-01-16", "email": "cegc@test.org", "currency": "GMD", "company": "Umbrella", "is_active": true}, {"phone": "(585) 125-5236", "name": "Frank Brandt", "status": "Pending", "signup_date": "2023-06-04", "email": "bcjfmlh@example.com", "currency": "BSD", "company": "Acme Inc", "is_active": false}, {"phone": "(538) 726-3884", "name": "Alice Park", "status": "Overdue", "signup_date": "2023-12-04", "email": "cdoam@mail.com", "currency": "BSD", "company": "Globex", "is_active": true}, {"phone": "(565) 030-2074", "name": "Ivan Khan", "status": "Pending", "signup_date": "2023-06-09", "email": "gkchn@corp.io", "currency": "GMD", "company": "Vehement", "is_active": true}, {"phone": "(516) 310-9090", "name": "Grace Brandt", "status": "Refunded", "signup_date": "2023-09-27", "email": "ainm@mail.com", "currency": "BMD", "company": "Globex", "is_active": true}, {"phone": "(552) 582-8986", "name": "Lena Reyes", "status": "Pending", "signup_date": "2023-05-13", "email": "ccde@test.org", "currency": "BMD", "company": "Soylent Corp", "is_active": false}, {"phone": "(507) 797-3243", "name": "Lena Wong", "status": "Pending", "signup_date": "2023-11-12", "email": "blleen@mail.com", "currency": "BSD", "company": "Umbrella", "is_active": true}, {"phone": "(535) 997-8877", "name": "Eve Cruz", "status": "Refunded", "signup_date": "2023-06-16", "email": "hcgk@corp.io", "currency": "BSD", "company": "Vehement", "is_active": false}, {"phone": "(575) 204-7916", "name": "Eve Cruz", "status": "Refunded", "signup_date": "2023-12-20", "email": "dnidhjl@mail.com", "currency": "BSD", "company": "Cyberdyne", "is_active": true}, {"phone": "(570) 940-1646", "name": "Heidi Petrov", "status": "Refunded", "signup_date": "2023-05-07", "email": "lamehd@example.com", "currency": "BSD", "company": "Initech", "is_active": false}, {"phone": "(510) 199-9012", "name": "Heidi Ali", "status": "Refunded", "signup_date": "2023-05-01", "email": "ljekegf@test.org", "currency": "BMD", "company": "Globex", "is_active": true}, {"phone": "(591) 834-3544", "name": "Priya Lee", "status": "Overdue", "signup_date": "2023-04-03", "email": "jheag@example.com", "currency": "GMD", "company": "Wonka Co", "is_active": false}, {"phone": "(543) 844-7587", "name": "Bob Khan", "status": "Pending", "signup_date": "2023-12-22", "email": "bjcj@corp.io", "currency": "GMD", "company": "Wonka Co", "is_active": true}], "dirty_cols": ["phone", "name", "status", "signup_date", "email", "currency", "company", "is_active"], "clean_cols": ["phone", "name", "status", "signup_date", "email", "currency", "company", "is_active"], "plan": {"dataset_summary": "26 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pending": "Pending", "refunded": "Refunded", "OVERDUE": "Overdue", "late": "Overdue", "Refund": "Refunded", "ovedue": "Overdue", "Unyaid": "Pending", "laet": "Overdue", "DUE": "Pending", "unpaid": "Pending", "rfund": "Refunded", "refund": "Refunded", "overdue": "Overdue"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bmd": "BMD", "Bahamian Dollar": "BSD", "Dalasi": "GMD", "Bsd": "BSD", "bahamina dollar": "BSD", "bahamian dollar": "BSD", "Bmd": "BMD"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"email": "   LHKOCK@TEST.ORG", "company": " Cyberdyne", "job_title": "C.E.O.", "state": "Puuerto Rico", "city": "yautepec", "signup_date": "45031"}, {"email": "apafm@example.com", "company": "Soylent  Corp", "job_title": "CEO", "state": " ILLINOIS", "city": "NOLA", "signup_date": "2023-07-11"}, {"email": "mlbhad@mail.com ", "company": "Wonka Co", "job_title": "engineering mgr", "state": " illinois ", "city": "Yautepec", "signup_date": "22 Jun 2023"}, {"email": " eojlffl@example.com ", "company": "Umbrella", "job_title": "Egn Mgr", "state": "Illinois", "city": "Kabompo", "signup_date": "6/9/2023"}, {"email": "   DHGMCC@CORP.IO", "company": "Vehement", "job_title": "Sales Rep", "state": "il", "city": "RUDOW", "signup_date": "6 Nov 2023"}, {"email": "EHPDKH@MAIL.COM", "company": "Soylent Corp", "job_title": "ENGMGR", "state": "Illinois", "city": "NEW ORLEANS", "signup_date": "2023-08-23"}, {"email": "CHILOFM@TEST.ORG", "company": " Umbrella ", "job_title": "Sales Representative", "state": " pr ", "city": "Rudow", "signup_date": "45091"}, {"email": " CLFH@TEST.ORG ", "company": " Umbrella", "job_title": "Eng Manager", "state": "IL ", "city": "Palos Hills", "signup_date": "4/2/2023"}, {"email": "ehfkpdk@example.com", "company": "Acme Inc", "job_title": "engineering mgr ", "state": "California", "city": "Palos Hills", "signup_date": "10/28/2023"}, {"email": " bgfgmae@corp.io ", "company": " Globex ", "job_title": "CEO", "state": "New Hampshire", "city": "Yautepec", "signup_date": "45074"}, {"email": " hiblp@test.org ", "company": "Wonka Co", "job_title": "c.e.o.", "state": "NH", "city": "kabompo", "signup_date": "4/18/2023"}, {"email": "   mhbdmgi@mail.com", "company": "Initech", "job_title": "   ENGM GR", "state": " nh ", "city": "KABOMPO", "signup_date": "14 Nov 2023"}, {"email": "nnjo@mail.com", "company": " Hooli", "job_title": "Rep", "state": "nm", "city": " PALOS HILLS", "signup_date": "3/9/2023"}, {"email": "jbanjj@test.org  ", "company": "Wonka Co  ", "job_title": "ceo", "state": "IL", "city": " KABOMPO ", "signup_date": "4 Jan 2023"}, {"email": "CBPEFKL@EXAMPLE.COM   ", "company": "Vehement", "job_title": "REP", "state": "IL", "city": "  palos hills", "signup_date": "2023-01-02"}, {"email": "dijncno@corp.io   ", "company": "Acme Inc ", "job_title": "SALES REP", "state": "NH", "city": "New Orleans", "signup_date": "11/17/2023"}, {"email": "pfhnk@example.com", "company": "Umbrella ", "job_title": "Sales Rep", "state": "camifornia", "city": "Yautepec   ", "signup_date": "45231"}, {"email": "EFAFHCM@MAIL.COM", "company": " Hooli ", "job_title": "eng manager", "state": "New Hampshire", "city": "New Orleans  ", "signup_date": "20 Apr 2023"}, {"email": "bjmcdk@test.org", "company": "  Globex", "job_title": "Sales  Rep", "state": "nm", "city": "Palos Hills", "signup_date": "2023-11-20"}, {"email": "aahoool@example.com   ", "company": "Hooli", "job_title": "Sales Rep", "state": "Perto Rico", "city": "rudow", "signup_date": "3/3/2023"}, {"email": "FLMMIF@CORP.IO ", "company": "Initech", "job_title": "C.E.O.", "state": "New Mexico", "city": "RUDOW", "signup_date": "7 Nov 2023"}, {"email": "ififfj@example.com", "company": " Hooli", "job_title": "C.E.O.", "state": "CA", "city": "rudow", "signup_date": "45070"}, {"email": " EBLB@CORP.IO ", "company": "Initech   ", "job_title": " CEO ", "state": "IL", "city": "Kabompo", "signup_date": "2023-07-01"}, {"email": " HGOF@TEST.ORG", "company": "Soylent Corp", "job_title": "sales rep", "state": "IL", "city": "nolla", "signup_date": "2023-11-28"}, {"email": "   fkaj@corp.io", "company": "Initech  ", "job_title": " ceo ", "state": "NH", "city": "yautepec", "signup_date": "2023-11-10"}, {"email": " edaa@mail.com", "company": "Cyberdyne", "job_title": "Sales Rep", "state": "il   ", "city": "Kabompo", "signup_date": "45243"}, {"email": "aodmkp@corp.io", "company": "Vehement", "job_title": "sales representative", "state": "il", "city": "Palos Hills", "signup_date": "2023-03-11"}, {"email": " klfcdk@corp.io ", "company": "   Globex", "job_title": "Rep", "state": "new hampshire   ", "city": "New Orleans", "signup_date": "22 Jun 2023"}, {"email": "KJIPFC@MAIL.COM", "company": "Wonka Co   ", "job_title": "Engineering Mgr", "state": "New Mexico", "city": "RUDOW", "signup_date": "24 Mar 2023"}, {"email": "cmol@mail.com ", "company": "   Initech", "job_title": "en gmgr", "state": "new hampsshire", "city": "NLA", "signup_date": "17 Jul 2023"}, {"email": "LOCLC@EXAMPLE.COM", "company": " Umbrella", "job_title": "Saales Representative", "state": "Pr", "city": "PALOS HILLS", "signup_date": "45070"}, {"email": "hiiica@mail.com   ", "company": "   Acme Inc", "job_title": "sales  rep", "state": "Il", "city": "  Kabompo", "signup_date": "2023-04-23"}, {"email": "LPGIHKM@CORP.IO  ", "company": "Acme Inc", "job_title": "chief executive officer", "state": "ILLINOIS", "city": "NLLA", "signup_date": "45135"}, {"email": "nmobl@corp.io", "company": " Initech ", "job_title": "Engineering Mgr   ", "state": "CA", "city": "Yautepec", "signup_date": "3/4/2023"}, {"email": " GOECHL@MAIL.COM", "company": "   Umbrella", "job_title": "Rep", "state": "pr", "city": "Kabompo", "signup_date": "6/26/2023"}, {"email": "HMJD@TEST.ORG", "company": "Wonka  Co", "job_title": "CEO", "state": " california ", "city": "yautepec", "signup_date": "12/17/2023"}, {"email": " akikbme@corp.io ", "company": "Wonka Co  ", "job_title": "CEO", "state": "Ca", "city": "   Rudow", "signup_date": "45073"}, {"email": "FFLBD@EXAMPLE.COM", "company": "Stark Industries", "job_title": "Eng Manager", "state": "  NM", "city": "nola", "signup_date": "5/28/2023"}, {"email": "dajc@test.org", "company": "Umbrella   ", "job_title": "  Chief Executive Officer", "state": "New  Mecico", "city": " RUDOW ", "signup_date": "2023-02-28"}, {"email": " MGMBD@TEST.ORG ", "company": "Globex", "job_title": "   C.E.OO.", "state": "IL", "city": "Yauetpec", "signup_date": "2023-02-27"}, {"email": " iepeec@example.com ", "company": " Globex", "job_title": "Rep", "state": "New Mexxico", "city": "NOLA", "signup_date": "2023-04-10"}, {"email": "GPJBKL@MAIL.COM", "company": " Umbrella ", "job_title": "CEO", "state": "Illinois", "city": "Nola  ", "signup_date": "3/13/2023"}, {"email": "FHLD@MAIL.COM", "company": "Globex", "job_title": "REP", "state": "CALIFORNIA", "city": "yautepec", "signup_date": "28 Oct 2023"}, {"email": "ipde@mail.com", "company": "Acme Inc", "job_title": "Rep", "state": "   NM", "city": "Yautepec   ", "signup_date": "2023-02-10"}, {"email": "jpakgee@test.org", "company": "   Wonka Co", "job_title": "eng manager", "state": "NM", "city": "Yautepec", "signup_date": "6/25/2023"}, {"email": "NGNFG@CORP.IO", "company": "  Hooli", "job_title": "engineering mgr   ", "state": "PR", "city": "New Orleans ", "signup_date": "15 Jul 2023"}, {"email": "IBBBAM@TEST.ORG", "company": "   Umbrella", "job_title": "Engineering Manager", "state": "NM", "city": "rudow", "signup_date": "2023-03-04"}, {"email": "DDMDAAO@MAIL.COM ", "company": "Globex", "job_title": "CEO", "state": "nm", "city": "YAUTEPEC   ", "signup_date": "11/3/2023"}, {"email": "hocool@example.com", "company": "Umbrella", "job_title": "rep", "state": "NH", "city": "Rudow ", "signup_date": "9/27/2023"}, {"email": "MFHNM@CORP.IO", "company": " Initech ", "job_title": " C.E.O. ", "state": "PR", "city": "Yautepec", "signup_date": "9/12/2023"}, {"email": "dajc@test.org", "company": "Umbrella   ", "job_title": "  Chief Executive Officer", "state": "New  Mecico", "city": " RUDOW ", "signup_date": "2023-02-28"}, {"email": "", "company": "", "job_title": "", "state": "", "city": "", "signup_date": ""}], "clean": [{"email": "lhkock@test.org", "company": "Cyberdyne", "job_title": "Chief Executive Officer", "state": "Puerto Rico", "city": "Yautepec", "signup_date": "2023-04-15"}, {"email": "apafm@example.com", "company": "Soylent Corp", "job_title": "Chief Executive Officer", "state": "Illinois", "city": "New Orleans", "signup_date": "2023-07-11"}, {"email": "mlbhad@mail.com", "company": "Wonka Co", "job_title": "Engineering Manager", "state": "Illinois", "city": "Yautepec", "signup_date": "2023-06-22"}, {"email": "eojlffl@example.com", "company": "Umbrella", "job_title": "Engineering Manager", "state": "Illinois", "city": "Kabompo", "signup_date": "2023-06-09"}, {"email": "dhgmcc@corp.io", "company": "Vehement", "job_title": "Sales Representative", "state": "Illinois", "city": "Rudow", "signup_date": "2023-11-06"}, {"email": "ehpdkh@mail.com", "company": "Soylent Corp", "job_title": "Engineering Manager", "state": "Illinois", "city": "New Orleans", "signup_date": "2023-08-23"}, {"email": "chilofm@test.org", "company": "Umbrella", "job_title": "Sales Representative", "state": "Puerto Rico", "city": "Rudow", "signup_date": "2023-06-14"}, {"email": "clfh@test.org", "company": "Umbrella", "job_title": "Engineering Manager", "state": "Illinois", "city": "Palos Hills", "signup_date": "2023-04-02"}, {"email": "ehfkpdk@example.com", "company": "Acme Inc", "job_title": "Engineering Manager", "state": "California", "city": "Palos Hills", "signup_date": "2023-10-28"}, {"email": "bgfgmae@corp.io", "company": "Globex", "job_title": "Chief Executive Officer", "state": "New Hampshire", "city": "Yautepec", "signup_date": "2023-05-28"}, {"email": "hiblp@test.org", "company": "Wonka Co", "job_title": "Chief Executive Officer", "state": "New Hampshire", "city": "Kabompo", "signup_date": "2023-04-18"}, {"email": "mhbdmgi@mail.com", "company": "Initech", "job_title": "Engineering Manager", "state": "New Hampshire", "city": "Kabompo", "signup_date": "2023-11-14"}, {"email": "nnjo@mail.com", "company": "Hooli", "job_title": "Sales Representative", "state": "New Mexico", "city": "Palos Hills", "signup_date": "2023-03-09"}, {"email": "jbanjj@test.org", "company": "Wonka Co", "job_title": "Chief Executive Officer", "state": "Illinois", "city": "Kabompo", "signup_date": "2023-01-04"}, {"email": "cbpefkl@example.com", "company": "Vehement", "job_title": "Sales Representative", "state": "Illinois", "city": "Palos Hills", "signup_date": "2023-01-02"}, {"email": "dijncno@corp.io", "company": "Acme Inc", "job_title": "Sales Representative", "state": "New Hampshire", "city": "New Orleans", "signup_date": "2023-11-17"}, {"email": "pfhnk@example.com", "company": "Umbrella", "job_title": "Sales Representative", "state": "California", "city": "Yautepec", "signup_date": "2023-11-01"}, {"email": "efafhcm@mail.com", "company": "Hooli", "job_title": "Engineering Manager", "state": "New Hampshire", "city": "New Orleans", "signup_date": "2023-04-20"}, {"email": "bjmcdk@test.org", "company": "Globex", "job_title": "Sales Representative", "state": "New Mexico", "city": "Palos Hills", "signup_date": "2023-11-20"}, {"email": "aahoool@example.com", "company": "Hooli", "job_title": "Sales Representative", "state": "Puerto Rico", "city": "Rudow", "signup_date": "2023-03-03"}, {"email": "flmmif@corp.io", "company": "Initech", "job_title": "Chief Executive Officer", "state": "New Mexico", "city": "Rudow", "signup_date": "2023-11-07"}, {"email": "ififfj@example.com", "company": "Hooli", "job_title": "Chief Executive Officer", "state": "California", "city": "Rudow", "signup_date": "2023-05-24"}, {"email": "eblb@corp.io", "company": "Initech", "job_title": "Chief Executive Officer", "state": "Illinois", "city": "Kabompo", "signup_date": "2023-07-01"}, {"email": "hgof@test.org", "company": "Soylent Corp", "job_title": "Sales Representative", "state": "Illinois", "city": "New Orleans", "signup_date": "2023-11-28"}, {"email": "fkaj@corp.io", "company": "Initech", "job_title": "Chief Executive Officer", "state": "New Hampshire", "city": "Yautepec", "signup_date": "2023-11-10"}, {"email": "edaa@mail.com", "company": "Cyberdyne", "job_title": "Sales Representative", "state": "Illinois", "city": "Kabompo", "signup_date": "2023-11-13"}, {"email": "aodmkp@corp.io", "company": "Vehement", "job_title": "Sales Representative", "state": "Illinois", "city": "Palos Hills", "signup_date": "2023-03-11"}, {"email": "klfcdk@corp.io", "company": "Globex", "job_title": "Sales Representative", "state": "New Hampshire", "city": "New Orleans", "signup_date": "2023-06-22"}, {"email": "kjipfc@mail.com", "company": "Wonka Co", "job_title": "Engineering Manager", "state": "New Mexico", "city": "Rudow", "signup_date": "2023-03-24"}, {"email": "cmol@mail.com", "company": "Initech", "job_title": "Engineering Manager", "state": "New Hampshire", "city": "New Orleans", "signup_date": "2023-07-17"}, {"email": "loclc@example.com", "company": "Umbrella", "job_title": "Sales Representative", "state": "Puerto Rico", "city": "Palos Hills", "signup_date": "2023-05-24"}, {"email": "hiiica@mail.com", "company": "Acme Inc", "job_title": "Sales Representative", "state": "Illinois", "city": "Kabompo", "signup_date": "2023-04-23"}, {"email": "lpgihkm@corp.io", "company": "Acme Inc", "job_title": "Chief Executive Officer", "state": "Illinois", "city": "New Orleans", "signup_date": "2023-07-28"}, {"email": "nmobl@corp.io", "company": "Initech", "job_title": "Engineering Manager", "state": "California", "city": "Yautepec", "signup_date": "2023-03-04"}, {"email": "goechl@mail.com", "company": "Umbrella", "job_title": "Sales Representative", "state": "Puerto Rico", "city": "Kabompo", "signup_date": "2023-06-26"}, {"email": "hmjd@test.org", "company": "Wonka Co", "job_title": "Chief Executive Officer", "state": "California", "city": "Yautepec", "signup_date": "2023-12-17"}, {"email": "akikbme@corp.io", "company": "Wonka Co", "job_title": "Chief Executive Officer", "state": "California", "city": "Rudow", "signup_date": "2023-05-27"}, {"email": "fflbd@example.com", "company": "Stark Industries", "job_title": "Engineering Manager", "state": "New Mexico", "city": "New Orleans", "signup_date": "2023-05-28"}, {"email": "dajc@test.org", "company": "Umbrella", "job_title": "Chief Executive Officer", "state": "New Mexico", "city": "Rudow", "signup_date": "2023-02-28"}, {"email": "mgmbd@test.org", "company": "Globex", "job_title": "Chief Executive Officer", "state": "Illinois", "city": "Yautepec", "signup_date": "2023-02-27"}, {"email": "iepeec@example.com", "company": "Globex", "job_title": "Sales Representative", "state": "New Mexico", "city": "New Orleans", "signup_date": "2023-04-10"}, {"email": "gpjbkl@mail.com", "company": "Umbrella", "job_title": "Chief Executive Officer", "state": "Illinois", "city": "New Orleans", "signup_date": "2023-03-13"}, {"email": "fhld@mail.com", "company": "Globex", "job_title": "Sales Representative", "state": "California", "city": "Yautepec", "signup_date": "2023-10-28"}, {"email": "ipde@mail.com", "company": "Acme Inc", "job_title": "Sales Representative", "state": "New Mexico", "city": "Yautepec", "signup_date": "2023-02-10"}, {"email": "jpakgee@test.org", "company": "Wonka Co", "job_title": "Engineering Manager", "state": "New Mexico", "city": "Yautepec", "signup_date": "2023-06-25"}, {"email": "ngnfg@corp.io", "company": "Hooli", "job_title": "Engineering Manager", "state": "Puerto Rico", "city": "New Orleans", "signup_date": "2023-07-15"}, {"email": "ibbbam@test.org", "company": "Umbrella", "job_title": "Engineering Manager", "state": "New Mexico", "city": "Rudow", "signup_date": "2023-03-04"}, {"email": "ddmdaao@mail.com", "company": "Globex", "job_title": "Chief Executive Officer", "state": "New Mexico", "city": "Yautepec", "signup_date": "2023-11-03"}, {"email": "hocool@example.com", "company": "Umbrella", "job_title": "Sales Representative", "state": "New Hampshire", "city": "Rudow", "signup_date": "2023-09-27"}, {"email": "mfhnm@corp.io", "company": "Initech", "job_title": "Chief Executive Officer", "state": "Puerto Rico", "city": "Yautepec", "signup_date": "2023-09-12"}], "dirty_cols": ["email", "company", "job_title", "state", "city", "signup_date"], "clean_cols": ["email", "company", "job_title", "state", "city", "signup_date"], "plan": {"dataset_summary": "52 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"C.E.O.": "Chief Executive Officer", "CEO": "Chief Executive Officer", "engineering mgr": "Engineering Manager", "Egn Mgr": "Engineering Manager", "Sales Rep": "Sales Representative", "ENGMGR": "Engineering Manager", "Eng Manager": "Engineering Manager", "c.e.o.": "Chief Executive Officer", "ENGM GR": "Engineering Manager", "Rep": "Sales Representative", "ceo": "Chief Executive Officer", "REP": "Sales Representative", "SALES REP": "Sales Representative", "eng manager": "Engineering Manager", "sales rep": "Sales Representative", "sales representative": "Sales Representative", "Engineering Mgr": "Engineering Manager", "en gmgr": "Engineering Manager", "Saales Representative": "Sales Representative", "chief executive officer": "Chief Executive Officer", "C.E.OO.": "Chief Executive Officer", "rep": "Sales Representative"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Puuerto Rico": "Puerto Rico", "ILLINOIS": "Illinois", "illinois": "Illinois", "il": "Illinois", "pr": "Puerto Rico", "IL": "Illinois", "NH": "New Hampshire", "nh": "New Hampshire", "nm": "New Mexico", "camifornia": "California", "Perto Rico": "Puerto Rico", "CA": "California", "new hampshire": "New Hampshire", "new hampsshire": "New Hampshire", "Pr": "Puerto Rico", "Il": "Illinois", "california": "California", "Ca": "California", "NM": "New Mexico", "New Mecico": "New Mexico", "New Mexxico": "New Mexico", "CALIFORNIA": "California", "PR": "Puerto Rico"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"yautepec": "Yautepec", "NOLA": "New Orleans", "RUDOW": "Rudow", "NEW ORLEANS": "New Orleans", "kabompo": "Kabompo", "KABOMPO": "Kabompo", "PALOS HILLS": "Palos Hills", "palos hills": "Palos Hills", "rudow": "Rudow", "nolla": "New Orleans", "NLA": "New Orleans", "NLLA": "New Orleans", "nola": "New Orleans", "Yauetpec": "Yautepec", "Nola": "New Orleans", "YAUTEPEC": "Yautepec"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"name": " Bob Lee", "rate": "63.4%", "status": "active", "department": "People", "country": "   Bes", "phone": "596.074.2859"}, {"name": "Karl Adams   ", "rate": "93.3%", "status": "Active", "department": "ACCOUNTING ", "country": "Bonaire Sint Eustatius and  Saba", "phone": "555-138-2308"}, {"name": "Karl Ali", "rate": "63.5%", "status": "paused", "department": "SALES", "country": "Bq", "phone": "(506)641-6587"}, {"name": "Alice Moore", "rate": "29.6%", "status": "on hold  ", "department": "Biz Dev", "country": "Fin", "phone": "501.403.4768"}, {"name": "Lena Park", "rate": "59.0%", "status": "Paused", "department": "People  ", "country": "   Republic of Finland", "phone": "(580)047-0752"}, {"name": "Alice Reyes", "rate": "7.2%", "status": "active", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(597)712-1906"}, {"name": "   Grace Smith", "rate": "96.1%", "status": "   active", "department": " Pevple ", "country": "FIN", "phone": "(591)646-3613"}, {"name": "Judy  Adams", "rate": "26.0%", "status": "active", "department": " sales ", "country": "BES", "phone": "576-391-5695"}, {"name": "   Mona Lee", "rate": "79.1%", "status": "ACTIVE", "department": "Human Resources", "country": "TUNISIA", "phone": "(562)925-9477"}, {"name": "   Alice Johnson", "rate": "40.8%", "status": "PAUSED", "department": "People", "country": "NORTH MACEDONIA", "phone": "501-379-0569"}, {"name": "Carol Moore", "rate": "7.1%", "status": "pauned", "department": " HR", "country": "TUN", "phone": "583-024-1290"}, {"name": "   Mona Novak", "rate": "40.5%", "status": "active", "department": "Finance", "country": "bq", "phone": "547.136.1032"}, {"name": "Omar Novak", "rate": "13.0%", "status": "ACTIVE   ", "department": "Finance", "country": "Finland  ", "phone": "(572)907-7924"}, {"name": "Grace Novak", "rate": "5.3%", "status": "active", "department": "hr", "country": "BONAIRE, SINT EUSTATIUS AND SABA", "phone": "580.353.3340"}, {"name": "Frank Fischer", "rate": "57.2%", "status": "paused", "department": "HR", "country": "North Macedonia", "phone": "(559)491-1247"}, {"name": "Karl Brandt   ", "rate": "58.4%", "status": "paused", "department": " Accounting ", "country": "FIN", "phone": "549.013.5684"}, {"name": "  Priya Brandt", "rate": "63.1%", "status": "paused", "department": "HR", "country": "Bonaire, Sint Eustatius and Sabba", "phone": "5952722790"}, {"name": "Karl Lee", "rate": "84.0%", "status": " paused ", "department": "  Accounting", "country": "BES", "phone": "530.753.2687"}, {"name": "Eve Petrov", "rate": "36.9%", "status": "on  hold", "department": "Sales", "country": " Finlannd ", "phone": "5663467562"}, {"name": "Karl Ali", "rate": "39.2%", "status": "ON HOLD", "department": "finance", "country": "bes", "phone": "565.004.5422"}, {"name": "Omar  Novak", "rate": "6.3%", "status": "paused", "department": "ACCOUNTIGN", "country": "  fi", "phone": "530-398-5905"}, {"name": "Omar Park", "rate": "57.0%", "status": "Active", "department": " finance ", "country": "tunisia  ", "phone": "(581)880-5755"}, {"name": "Sara Reyes", "rate": "88.3%", "status": "paused", "department": "BizlDev", "country": "  Finland", "phone": "528.911.2665"}, {"name": " Carol Moore", "rate": "19.2%", "status": "paused ", "department": "People Ops  ", "country": "republic of north macedonia", "phone": "555-262-9706"}, {"name": "Judy Lee ", "rate": "56.2%", "status": "active   ", "department": "Fin", "country": "FI ", "phone": "(561)824-4765"}, {"name": "Karl Wong", "rate": "85.8%", "status": "active", "department": "biz dev", "country": "north macedonia", "phone": "(588)167-9398"}, {"name": "Ivan  Johnson", "rate": "52.2%", "status": "Active", "department": "BIZ DEV", "country": "REPUBLIC FO FINLAND", "phone": "5163345741"}, {"name": "   Alice Adams", "rate": "1.8%", "status": "ACTIVE", "department": "Finnance", "country": "MK", "phone": "519-949-0521"}, {"name": "David Moore  ", "rate": "64.5%", "status": " active ", "department": "accounting", "country": "MK", "phone": "538-064-4665"}, {"name": "Eve Petrov", "rate": "42.0%", "status": "  active", "department": "  Accounting", "country": "FI", "phone": "559.754.5552"}, {"name": "Ivan Lee ", "rate": "66.0%", "status": "active   ", "department": "Finance", "country": "bes   ", "phone": "5489912762"}, {"name": "   Carol Adams", "rate": "87.4%", "status": "ACTIVE", "department": "People", "country": "TN", "phone": "(561)563-2929"}, {"name": "Alice Ali", "rate": "60.6%", "status": "PAUUSED   ", "department": "hr", "country": " bq ", "phone": "588.972.7125"}, {"name": "Omar  Cruz", "rate": "96.5%", "status": " paused ", "department": "HR", "country": "REPUBLIC OF TUNISIA", "phone": "528.430.4455"}, {"name": "Heidi Smith", "rate": "61.8%", "status": "active", "department": "Accounting", "country": "MK", "phone": "568-136-4143"}, {"name": "Mona Smith", "rate": "82.0%", "status": "paused", "department": "  Biz Dev", "country": "fi   ", "phone": "522.785.5792"}, {"name": "Alice Johnson", "rate": "2.7%", "status": "actize   ", "department": "   Saels", "country": " Finlland ", "phone": "504-578-8297"}, {"name": "Omar Moore  ", "rate": "40.3%", "status": "Paused", "department": "Human Resources  ", "country": "North Macedonia", "phone": "(537)648-2258"}, {"name": "Ivan Park", "rate": "53.2%", "status": "Active", "department": "Biz Dev", "country": "FIN", "phone": "585-173-9292"}, {"name": "Frank Khan", "rate": "85.5%", "status": "on hold ", "department": "  Hmuan Resources", "country": "Tunisa", "phone": "509-274-8636"}, {"name": "Heidi Diaz", "rate": "11.2%", "status": "pused", "department": "accounting", "country": "MKD", "phone": "588-401-0552"}, {"name": "Carol Ali  ", "rate": "79.5%", "status": "  paused", "department": "fin", "country": "  fin", "phone": "5169865001"}, {"name": "", "rate": "", "status": "", "department": "", "country": "", "phone": ""}], "clean": [{"name": "Bob Lee", "rate": 0.634, "status": "Active", "department": "Human Resources", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(596) 074-2859"}, {"name": "Karl Adams", "rate": 0.9329999999999999, "status": "Active", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(555) 138-2308"}, {"name": "Karl Ali", "rate": 0.635, "status": "Paused", "department": "Sales", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(506) 641-6587"}, {"name": "Alice Moore", "rate": 0.29600000000000004, "status": "Paused", "department": "Sales", "country": "Finland", "phone": "(501) 403-4768"}, {"name": "Lena Park", "rate": 0.59, "status": "Paused", "department": "Human Resources", "country": "Finland", "phone": "(580) 047-0752"}, {"name": "Alice Reyes", "rate": 0.07200000000000001, "status": "Active", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(597) 712-1906"}, {"name": "Grace Smith", "rate": 0.961, "status": "Active", "department": "Human Resources", "country": "Finland", "phone": "(591) 646-3613"}, {"name": "Judy Adams", "rate": 0.26, "status": "Active", "department": "Sales", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(576) 391-5695"}, {"name": "Mona Lee", "rate": 0.7909999999999999, "status": "Active", "department": "Human Resources", "country": "Tunisia", "phone": "(562) 925-9477"}, {"name": "Alice Johnson", "rate": 0.408, "status": "Paused", "department": "Human Resources", "country": "North Macedonia", "phone": "(501) 379-0569"}, {"name": "Carol Moore", "rate": 0.071, "status": "Paused", "department": "Human Resources", "country": "Tunisia", "phone": "(583) 024-1290"}, {"name": "Mona Novak", "rate": 0.405, "status": "Active", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(547) 136-1032"}, {"name": "Omar Novak", "rate": 0.13, "status": "Active", "department": "Finance", "country": "Finland", "phone": "(572) 907-7924"}, {"name": "Grace Novak", "rate": 0.053, "status": "Active", "department": "Human Resources", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(580) 353-3340"}, {"name": "Frank Fischer", "rate": 0.5720000000000001, "status": "Paused", "department": "Human Resources", "country": "North Macedonia", "phone": "(559) 491-1247"}, {"name": "Karl Brandt", "rate": 0.584, "status": "Paused", "department": "Finance", "country": "Finland", "phone": "(549) 013-5684"}, {"name": "Priya Brandt", "rate": 0.631, "status": "Paused", "department": "Human Resources", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(595) 272-2790"}, {"name": "Karl Lee", "rate": 0.84, "status": "Paused", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(530) 753-2687"}, {"name": "Eve Petrov", "rate": 0.369, "status": "Paused", "department": "Sales", "country": "Finland", "phone": "(566) 346-7562"}, {"name": "Karl Ali", "rate": 0.392, "status": "Paused", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(565) 004-5422"}, {"name": "Omar Novak", "rate": 0.063, "status": "Paused", "department": "Finance", "country": "Finland", "phone": "(530) 398-5905"}, {"name": "Omar Park", "rate": 0.57, "status": "Active", "department": "Finance", "country": "Tunisia", "phone": "(581) 880-5755"}, {"name": "Sara Reyes", "rate": 0.883, "status": "Paused", "department": "Sales", "country": "Finland", "phone": "(528) 911-2665"}, {"name": "Carol Moore", "rate": 0.192, "status": "Paused", "department": "Human Resources", "country": "North Macedonia", "phone": "(555) 262-9706"}, {"name": "Judy Lee", "rate": 0.562, "status": "Active", "department": "Finance", "country": "Finland", "phone": "(561) 824-4765"}, {"name": "Karl Wong", "rate": 0.858, "status": "Active", "department": "Sales", "country": "North Macedonia", "phone": "(588) 167-9398"}, {"name": "Ivan Johnson", "rate": 0.522, "status": "Active", "department": "Sales", "country": "Finland", "phone": "(516) 334-5741"}, {"name": "Alice Adams", "rate": 0.018000000000000002, "status": "Active", "department": "Finance", "country": "North Macedonia", "phone": "(519) 949-0521"}, {"name": "David Moore", "rate": 0.645, "status": "Active", "department": "Finance", "country": "North Macedonia", "phone": "(538) 064-4665"}, {"name": "Eve Petrov", "rate": 0.42, "status": "Active", "department": "Finance", "country": "Finland", "phone": "(559) 754-5552"}, {"name": "Ivan Lee", "rate": 0.66, "status": "Active", "department": "Finance", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(548) 991-2762"}, {"name": "Carol Adams", "rate": 0.8740000000000001, "status": "Active", "department": "Human Resources", "country": "Tunisia", "phone": "(561) 563-2929"}, {"name": "Alice Ali", "rate": 0.606, "status": "Paused", "department": "Human Resources", "country": "Bonaire, Sint Eustatius and Saba", "phone": "(588) 972-7125"}, {"name": "Omar Cruz", "rate": 0.965, "status": "Paused", "department": "Human Resources", "country": "Tunisia", "phone": "(528) 430-4455"}, {"name": "Heidi Smith", "rate": 0.618, "status": "Active", "department": "Finance", "country": "North Macedonia", "phone": "(568) 136-4143"}, {"name": "Mona Smith", "rate": 0.82, "status": "Paused", "department": "Sales", "country": "Finland", "phone": "(522) 785-5792"}, {"name": "Alice Johnson", "rate": 0.027000000000000003, "status": "Active", "department": "Sales", "country": "Finland", "phone": "(504) 578-8297"}, {"name": "Omar Moore", "rate": 0.40299999999999997, "status": "Paused", "department": "Human Resources", "country": "North Macedonia", "phone": "(537) 648-2258"}, {"name": "Ivan Park", "rate": 0.532, "status": "Active", "department": "Sales", "country": "Finland", "phone": "(585) 173-9292"}, {"name": "Frank Khan", "rate": 0.855, "status": "Paused", "department": "Human Resources", "country": "Tunisia", "phone": "(509) 274-8636"}, {"name": "Heidi Diaz", "rate": 0.11199999999999999, "status": "Paused", "department": "Finance", "country": "North Macedonia", "phone": "(588) 401-0552"}, {"name": "Carol Ali", "rate": 0.795, "status": "Paused", "department": "Finance", "country": "Finland", "phone": "(516) 986-5001"}], "dirty_cols": ["name", "rate", "status", "department", "country", "phone"], "clean_cols": ["name", "rate", "status", "department", "country", "phone"], "plan": {"dataset_summary": "43 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"active": "Active", "paused": "Paused", "on hold": "Paused", "ACTIVE": "Active", "PAUSED": "Paused", "pauned": "Paused", "ON HOLD": "Paused", "PAUUSED": "Paused", "actize": "Active", "pused": "Paused"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"People": "Human Resources", "ACCOUNTING": "Finance", "SALES": "Sales", "Biz Dev": "Sales", "Pevple": "Human Resources", "sales": "Sales", "HR": "Human Resources", "hr": "Human Resources", "Accounting": "Finance", "finance": "Finance", "ACCOUNTIGN": "Finance", "BizlDev": "Sales", "People Ops": "Human Resources", "Fin": "Finance", "biz dev": "Sales", "BIZ DEV": "Sales", "Finnance": "Finance", "accounting": "Finance", "Saels": "Sales", "Hmuan Resources": "Human Resources", "fin": "Finance"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Bes": "Bonaire, Sint Eustatius and Saba", "Bonaire Sint Eustatius and Saba": "Bonaire, Sint Eustatius and Saba", "Bq": "Bonaire, Sint Eustatius and Saba", "Fin": "Finland", "Republic of Finland": "Finland", "FIN": "Finland", "BES": "Bonaire, Sint Eustatius and Saba", "TUNISIA": "Tunisia", "NORTH MACEDONIA": "North Macedonia", "TUN": "Tunisia", "bq": "Bonaire, Sint Eustatius and Saba", "BONAIRE, SINT EUSTATIUS AND SABA": "Bonaire, Sint Eustatius and Saba", "Bonaire, Sint Eustatius and Sabba": "Bonaire, Sint Eustatius and Saba", "Finlannd": "Finland", "bes": "Bonaire, Sint Eustatius and Saba", "fi": "Finland", "tunisia": "Tunisia", "republic of north macedonia": "North Macedonia", "FI": "Finland", "north macedonia": "North Macedonia", "REPUBLIC FO FINLAND": "Finland", "MK": "North Macedonia", "TN": "Tunisia", "REPUBLIC OF TUNISIA": "Tunisia", "Finlland": "Finland", "Tunisa": "Tunisia", "MKD": "North Macedonia", "fin": "Finland"}, "rationale": "Unified 28 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"city": "SAMAL", "company": "Cyberdyne", "name": "Karl Smith", "department": "Biz  Dev", "email": "JPNFBHM@EXAMPLE.COM", "signup_date": "7 Mar 2023"}, {"city": " Nezlobnaya", "company": "   Hooli", "name": "Alice Adams", "department": "Ops", "email": "mpdf@corp.io", "signup_date": "45080"}, {"city": "NEZLOBNAYA   ", "company": "Umbrella", "name": "Carol Adams  ", "department": "Biz Dev", "email": "egjlamc@test.org   ", "signup_date": "2023-03-23"}, {"city": "Samal", "company": " Umbrella ", "name": " Heidi Johnson", "department": "People", "email": "   fikijee@corp.io", "signup_date": "2023-11-25"}, {"city": "Samal", "company": "Hooli", "name": "Omar  Adams", "department": " HR ", "email": "occgi@mail.com", "signup_date": "2023-11-21"}, {"city": "saal", "company": "Hooli  ", "name": "Heidi  Cruz", "department": "Ops", "email": "imip@corp.io", "signup_date": "1 Jan 2023"}, {"city": " Samal", "company": "Acme Inc", "name": "Eve Diaz  ", "department": "PEOPLE OPS", "email": "afdmp@example.com", "signup_date": "2023-04-07"}, {"city": "Samal", "company": "   Cyberdyne", "name": "  Omar Lee", "department": " ops ", "email": "EJPCIM@TEST.ORG ", "signup_date": "6/1/2023"}, {"city": "   Samal", "company": "Globex", "name": "Eve Moore", "department": "biz dev", "email": "ghnehl@corp.io", "signup_date": "8/16/2023"}, {"city": "SAMAL", "company": "  Soylent Corp", "name": "Alice Brandt   ", "department": "HR", "email": " FJCIMI@TEST.ORG ", "signup_date": "2023-05-26"}, {"city": "nezlobnaya", "company": "Vehement", "name": "Alice Adams", "department": "People Ops", "email": "jfjonf@corp.io", "signup_date": "6/9/2023"}, {"city": "Nezlobnaya", "company": "  Stark Industries", "name": "  Lena Ali", "department": "biz dgv", "email": "lmfa@test.org", "signup_date": "10 Aug 2023"}, {"city": "saamal", "company": "Globex", "name": "Priya  Park", "department": "human resources", "email": "bmji@corp.io", "signup_date": "18 Jun 2023"}, {"city": "Nezlobnaya", "company": "Stark Industries", "name": "Ivan  Diaz", "department": "Opertions", "email": "MLIK@EXAMPLE.COM", "signup_date": "2023-10-26"}, {"city": "samal", "company": "   Initech", "name": "Bob Petrov", "department": "Sales", "email": "NBPP@TEST.ORG", "signup_date": "2023-01-21"}, {"city": " Nezlobnaya ", "company": "Wonka  Co", "name": "Lena Cruz", "department": "Human Resources", "email": " POLOBN@MAIL.COM ", "signup_date": "45287"}, {"city": " Nezlobnaja", "company": "   Initech", "name": "Ivan Smith", "department": "Sales", "email": " DJPH@MAIL.COM ", "signup_date": "2023-08-18"}, {"city": "nezlobnaya", "company": "Wonka Co", "name": "Lena Adams", "department": "HR", "email": "bfaika@corp.io", "signup_date": "45149"}, {"city": "nezlobnaya", "company": " Vehement ", "name": "Bob Johnson", "department": "HR", "email": "cimmfip@example.com", "signup_date": "16 Jan 2023"}, {"city": " Nezlobnaya ", "company": "  Stark Industries", "name": "Lena  Novak", "department": "Ops", "email": "  hmcgoh@test.org", "signup_date": "19 Sep 2023"}, {"city": "Nezlobanya", "company": "Stark Industries", "name": "Karl Diaz", "department": "ops", "email": "MHFBE@EXAMPLE.COM", "signup_date": "8/5/2023"}, {"city": "SAMAL", "company": "   Umbrella", "name": "Judy Johnson", "department": " Operations ", "email": "   gfmcg@mail.com", "signup_date": "44987"}, {"city": "NEZLOBNAYA", "company": "Vehement", "name": "Omar Adams ", "department": "people ops", "email": "   KKCLL@TEST.ORG", "signup_date": "14 Apr 2023"}, {"city": "   samaal", "company": "Umbrella ", "name": "Karl Petrov   ", "department": "PEOPLE", "email": "DAAB@TEST.ORG", "signup_date": "12 Aug 2023"}, {"city": "Samal", "company": " Initech ", "name": "Priya Lee", "department": "People Ops", "email": " eoana@example.com", "signup_date": "11 Nov 2023"}, {"city": "nezlobnaya", "company": "Initech", "name": "Judy  Park", "department": "HR", "email": "LDNIIKM@MAIL.COM", "signup_date": "2023-07-13"}, {"city": "nezlobnaya", "company": "Wonka Co", "name": "Mona Adams  ", "department": "biz dev", "email": " MILA@MAIL.COM", "signup_date": "3 Jul 2023"}, {"city": "SAMAL", "company": "Vehement  ", "name": "Frank Wong", "department": "OPS", "email": "BEHBDD@CORP.IO", "signup_date": "2023-01-11"}, {"city": "samal", "company": "  Globex", "name": "Lena Wong", "department": "Operations", "email": "ekmb@mail.com", "signup_date": "45284"}, {"city": "Nezlobnaya", "company": "Wonka Co", "name": "Alice  Adams", "department": " OPS ", "email": "khgo@example.com", "signup_date": "2023-08-21"}, {"city": "Samrl", "company": "Initech", "name": " Bob Lee", "department": "Biz Dev", "email": "  LIDNB@EXAMPLE.COM", "signup_date": "45115"}, {"city": "Smal", "company": "  Wonka Co", "name": "  David Johnson", "department": "Operations   ", "email": " GAHNIG@MAIL.COM ", "signup_date": "2023-03-09"}, {"city": "Nezlobnaya", "company": "Acme Inc", "name": "Heidi Ali  ", "department": " HR ", "email": "  ahem@test.org", "signup_date": "5/8/2023"}, {"city": "nezlobnaya", "company": "Globex", "name": " Bob Reyes", "department": "Biz Dev", "email": "olmco@example.com ", "signup_date": "2023-01-05"}, {"city": "samal", "company": "Wonka Co", "name": "Sara  Moore", "department": "Human Resources", "email": "hlmkh@mail.com", "signup_date": "2023-05-01"}], "clean": [{"city": "Samal", "company": "Cyberdyne", "name": "Karl Smith", "department": "Sales", "email": "jpnfbhm@example.com", "signup_date": "2023-03-07"}, {"city": "Nezlobnaya", "company": "Hooli", "name": "Alice Adams", "department": "Operations", "email": "mpdf@corp.io", "signup_date": "2023-06-03"}, {"city": "Nezlobnaya", "company": "Umbrella", "name": "Carol Adams", "department": "Sales", "email": "egjlamc@test.org", "signup_date": "2023-03-23"}, {"city": "Samal", "company": "Umbrella", "name": "Heidi Johnson", "department": "Human Resources", "email": "fikijee@corp.io", "signup_date": "2023-11-25"}, {"city": "Samal", "company": "Hooli", "name": "Omar Adams", "department": "Human Resources", "email": "occgi@mail.com", "signup_date": "2023-11-21"}, {"city": "Samal", "company": "Hooli", "name": "Heidi Cruz", "department": "Operations", "email": "imip@corp.io", "signup_date": "2023-01-01"}, {"city": "Samal", "company": "Acme Inc", "name": "Eve Diaz", "department": "Human Resources", "email": "afdmp@example.com", "signup_date": "2023-04-07"}, {"city": "Samal", "company": "Cyberdyne", "name": "Omar Lee", "department": "Operations", "email": "ejpcim@test.org", "signup_date": "2023-06-01"}, {"city": "Samal", "company": "Globex", "name": "Eve Moore", "department": "Sales", "email": "ghnehl@corp.io", "signup_date": "2023-08-16"}, {"city": "Samal", "company": "Soylent Corp", "name": "Alice Brandt", "department": "Human Resources", "email": "fjcimi@test.org", "signup_date": "2023-05-26"}, {"city": "Nezlobnaya", "company": "Vehement", "name": "Alice Adams", "department": "Human Resources", "email": "jfjonf@corp.io", "signup_date": "2023-06-09"}, {"city": "Nezlobnaya", "company": "Stark Industries", "name": "Lena Ali", "department": "Sales", "email": "lmfa@test.org", "signup_date": "2023-08-10"}, {"city": "Samal", "company": "Globex", "name": "Priya Park", "department": "Human Resources", "email": "bmji@corp.io", "signup_date": "2023-06-18"}, {"city": "Nezlobnaya", "company": "Stark Industries", "name": "Ivan Diaz", "department": "Operations", "email": "mlik@example.com", "signup_date": "2023-10-26"}, {"city": "Samal", "company": "Initech", "name": "Bob Petrov", "department": "Sales", "email": "nbpp@test.org", "signup_date": "2023-01-21"}, {"city": "Nezlobnaya", "company": "Wonka Co", "name": "Lena Cruz", "department": "Human Resources", "email": "polobn@mail.com", "signup_date": "2023-12-27"}, {"city": "Nezlobnaya", "company": "Initech", "name": "Ivan Smith", "department": "Sales", "email": "djph@mail.com", "signup_date": "2023-08-18"}, {"city": "Nezlobnaya", "company": "Wonka Co", "name": "Lena Adams", "department": "Human Resources", "email": "bfaika@corp.io", "signup_date": "2023-08-11"}, {"city": "Nezlobnaya", "company": "Vehement", "name": "Bob Johnson", "department": "Human Resources", "email": "cimmfip@example.com", "signup_date": "2023-01-16"}, {"city": "Nezlobnaya", "company": "Stark Industries", "name": "Lena Novak", "department": "Operations", "email": "hmcgoh@test.org", "signup_date": "2023-09-19"}, {"city": "Nezlobnaya", "company": "Stark Industries", "name": "Karl Diaz", "department": "Operations", "email": "mhfbe@example.com", "signup_date": "2023-08-05"}, {"city": "Samal", "company": "Umbrella", "name": "Judy Johnson", "department": "Operations", "email": "gfmcg@mail.com", "signup_date": "2023-03-02"}, {"city": "Nezlobnaya", "company": "Vehement", "name": "Omar Adams", "department": "Human Resources", "email": "kkcll@test.org", "signup_date": "2023-04-14"}, {"city": "Samal", "company": "Umbrella", "name": "Karl Petrov", "department": "Human Resources", "email": "daab@test.org", "signup_date": "2023-08-12"}, {"city": "Samal", "company": "Initech", "name": "Priya Lee", "department": "Human Resources", "email": "eoana@example.com", "signup_date": "2023-11-11"}, {"city": "Nezlobnaya", "company": "Initech", "name": "Judy Park", "department": "Human Resources", "email": "ldniikm@mail.com", "signup_date": "2023-07-13"}, {"city": "Nezlobnaya", "company": "Wonka Co", "name": "Mona Adams", "department": "Sales", "email": "mila@mail.com", "signup_date": "2023-07-03"}, {"city": "Samal", "company": "Vehement", "name": "Frank Wong", "department": "Operations", "email": "behbdd@corp.io", "signup_date": "2023-01-11"}, {"city": "Samal", "company": "Globex", "name": "Lena Wong", "department": "Operations", "email": "ekmb@mail.com", "signup_date": "2023-12-24"}, {"city": "Nezlobnaya", "company": "Wonka Co", "name": "Alice Adams", "department": "Operations", "email": "khgo@example.com", "signup_date": "2023-08-21"}, {"city": "Samal", "company": "Initech", "name": "Bob Lee", "department": "Sales", "email": "lidnb@example.com", "signup_date": "2023-07-08"}, {"city": "Samal", "company": "Wonka Co", "name": "David Johnson", "department": "Operations", "email": "gahnig@mail.com", "signup_date": "2023-03-09"}, {"city": "Nezlobnaya", "company": "Acme Inc", "name": "Heidi Ali", "department": "Human Resources", "email": "ahem@test.org", "signup_date": "2023-05-08"}, {"city": "Nezlobnaya", "company": "Globex", "name": "Bob Reyes", "department": "Sales", "email": "olmco@example.com", "signup_date": "2023-01-05"}, {"city": "Samal", "company": "Wonka Co", "name": "Sara Moore", "department": "Human Resources", "email": "hlmkh@mail.com", "signup_date": "2023-05-01"}], "dirty_cols": ["city", "company", "name", "department", "email", "signup_date"], "clean_cols": ["city", "company", "name", "department", "email", "signup_date"], "plan": {"dataset_summary": "35 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SAMAL": "Samal", "NEZLOBNAYA": "Nezlobnaya", "saal": "Samal", "nezlobnaya": "Nezlobnaya", "saamal": "Samal", "samal": "Samal", "Nezlobnaja": "Nezlobnaya", "Nezlobanya": "Nezlobnaya", "samaal": "Samal", "Samrl": "Samal", "Smal": "Samal"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Biz Dev": "Sales", "Ops": "Operations", "People": "Human Resources", "HR": "Human Resources", "PEOPLE OPS": "Human Resources", "ops": "Operations", "biz dev": "Sales", "People Ops": "Human Resources", "biz dgv": "Sales", "human resources": "Human Resources", "Opertions": "Operations", "people ops": "Human Resources", "PEOPLE": "Human Resources", "OPS": "Operations"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"currency": "Afghani", "status": "   canceled", "email": " GGBHN@EXAMPLE.COM", "city": "Sungai Lalang", "company": "Stark Industries", "job_title": "VICE PRES"}, {"currency": "guinean franc", "status": "churned", "email": "  nhlmmhf@corp.io", "city": "Tolga  ", "company": " Acme Inc", "job_title": "vice presideent"}, {"currency": "GNF", "status": " ON HOLD", "email": "FLGF@CORP.IO", "city": "TOLGA", "company": "Wonka Co", "job_title": "CEO"}, {"currency": " AFN ", "status": "active", "email": " fjijgk@test.org ", "city": "Sungai Lalang", "company": "Initech", "job_title": "Admin Asst"}, {"currency": "afghani ", "status": "cancelled", "email": "BKIEPNL@EXAMPLE.COM", "city": "tolga", "company": "Hooli", "job_title": "sales representative"}, {"currency": "gnf", "status": "   Paused", "email": " jngflen@mail.com", "city": "Sungai Lalang", "company": " Initech ", "job_title": "Vice Pres"}, {"currency": "afghain", "status": "active", "email": "mdpnpob@example.com ", "city": "SUNGAI LALANG", "company": "Initech", "job_title": "Admin"}, {"currency": "Afghani", "status": "Active", "email": "OLKD@TEST.ORG", "city": " Sungai Lalang", "company": "Initech", "job_title": "c.e.o."}, {"currency": "AFN", "status": "Paused", "email": "helhglh@corp.io   ", "city": " Hagen", "company": "Vehement", "job_title": "Rep"}, {"currency": " afn ", "status": "ON HOLD", "email": "FADDEA@TEST.ORG", "city": "tolga", "company": "  Soylent Corp", "job_title": " Sales Rep"}, {"currency": "   guinean franc", "status": "Churned", "email": " NNJEKFG@EXAMPLE.COM ", "city": "Semporna", "company": " Initech ", "job_title": "Chief Executive Officer"}, {"currency": "Guinean Franc", "status": " paused ", "email": "   LNAJP@EXAMPLE.COM", "city": "Tolga", "company": "Initech", "job_title": "CEO"}, {"currency": "afn", "status": "active", "email": "jjmk@corp.io", "city": "Semporna", "company": "Hooli", "job_title": "vp"}, {"currency": "afn", "status": "churned", "email": "   MAALOB@TEST.ORG", "city": "tolga", "company": "Acme Inc", "job_title": " Sales Representative"}, {"currency": "Guinean Franc", "status": "active ", "email": "fmnaa@corp.io", "city": "Seporna", "company": "Stark Industries", "job_title": "Admin"}, {"currency": "  AFGHANI", "status": "ACTIVE", "email": " BOKBH@CORP.IO ", "city": " hagen ", "company": "Hooli  ", "job_title": "Sales Representative"}, {"currency": "   afn", "status": "Churned", "email": " DNBELKP@CORP.IO", "city": "Semporna", "company": "Acme Inc", "job_title": "Chief Executive Officer"}, {"currency": "GNF", "status": "Paused", "email": "LAPBNFL@MAIL.COM", "city": "sungai lalang", "company": "Initech", "job_title": "  ADMINISTRATIVE ASSISTANT"}, {"currency": "Afghani", "status": "canceled", "email": "EMGGD@EXAMPLE.COM", "city": "Sungai Lalang", "company": "Stark Industries", "job_title": "v.p."}, {"currency": "   AFN", "status": "ACTIVE", "email": "gmml@mail.com", "city": "hagen", "company": "Initech", "job_title": "C.E.O."}, {"currency": "guinean franc", "status": " canceled ", "email": "DGNFPIH@TEST.ORG", "city": " Semporna ", "company": "Globex", "job_title": "CEO "}, {"currency": " gnf ", "status": " ACTIVE ", "email": "MGIHN@EXAMPLE.COM", "city": "sunga lalang  ", "company": "Stark  Industries", "job_title": "Sales Representative"}, {"currency": "", "status": "", "email": "", "city": "", "company": "", "job_title": ""}], "clean": [{"currency": "AFN", "status": "Churned", "email": "ggbhn@example.com", "city": "Sungai Lalang", "company": "Stark Industries", "job_title": "Vice President"}, {"currency": "GNF", "status": "Churned", "email": "nhlmmhf@corp.io", "city": "Tolga", "company": "Acme Inc", "job_title": "Vice President"}, {"currency": "GNF", "status": "Paused", "email": "flgf@corp.io", "city": "Tolga", "company": "Wonka Co", "job_title": "Chief Executive Officer"}, {"currency": "AFN", "status": "Active", "email": "fjijgk@test.org", "city": "Sungai Lalang", "company": "Initech", "job_title": "Administrative Assistant"}, {"currency": "AFN", "status": "Churned", "email": "bkiepnl@example.com", "city": "Tolga", "company": "Hooli", "job_title": "Sales Representative"}, {"currency": "GNF", "status": "Paused", "email": "jngflen@mail.com", "city": "Sungai Lalang", "company": "Initech", "job_title": "Vice President"}, {"currency": "AFN", "status": "Active", "email": "mdpnpob@example.com", "city": "Sungai Lalang", "company": "Initech", "job_title": "Administrative Assistant"}, {"currency": "AFN", "status": "Active", "email": "olkd@test.org", "city": "Sungai Lalang", "company": "Initech", "job_title": "Chief Executive Officer"}, {"currency": "AFN", "status": "Paused", "email": "helhglh@corp.io", "city": "Hagen", "company": "Vehement", "job_title": "Sales Representative"}, {"currency": "AFN", "status": "Paused", "email": "faddea@test.org", "city": "Tolga", "company": "Soylent Corp", "job_title": "Sales Representative"}, {"currency": "GNF", "status": "Churned", "email": "nnjekfg@example.com", "city": "Semporna", "company": "Initech", "job_title": "Chief Executive Officer"}, {"currency": "GNF", "status": "Paused", "email": "lnajp@example.com", "city": "Tolga", "company": "Initech", "job_title": "Chief Executive Officer"}, {"currency": "AFN", "status": "Active", "email": "jjmk@corp.io", "city": "Semporna", "company": "Hooli", "job_title": "Vice President"}, {"currency": "AFN", "status": "Churned", "email": "maalob@test.org", "city": "Tolga", "company": "Acme Inc", "job_title": "Sales Representative"}, {"currency": "GNF", "status": "Active", "email": "fmnaa@corp.io", "city": "Semporna", "company": "Stark Industries", "job_title": "Administrative Assistant"}, {"currency": "AFN", "status": "Active", "email": "bokbh@corp.io", "city": "Hagen", "company": "Hooli", "job_title": "Sales Representative"}, {"currency": "AFN", "status": "Churned", "email": "dnbelkp@corp.io", "city": "Semporna", "company": "Acme Inc", "job_title": "Chief Executive Officer"}, {"currency": "GNF", "status": "Paused", "email": "lapbnfl@mail.com", "city": "Sungai Lalang", "company": "Initech", "job_title": "Administrative Assistant"}, {"currency": "AFN", "status": "Churned", "email": "emggd@example.com", "city": "Sungai Lalang", "company": "Stark Industries", "job_title": "Vice President"}, {"currency": "AFN", "status": "Active", "email": "gmml@mail.com", "city": "Hagen", "company": "Initech", "job_title": "Chief Executive Officer"}, {"currency": "GNF", "status": "Churned", "email": "dgnfpih@test.org", "city": "Semporna", "company": "Globex", "job_title": "Chief Executive Officer"}, {"currency": "GNF", "status": "Active", "email": "mgihn@example.com", "city": "Sungai Lalang", "company": "Stark Industries", "job_title": "Sales Representative"}], "dirty_cols": ["currency", "status", "email", "city", "company", "job_title"], "clean_cols": ["currency", "status", "email", "city", "company", "job_title"], "plan": {"dataset_summary": "23 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Afghani": "AFN", "guinean franc": "GNF", "afghani": "AFN", "gnf": "GNF", "afghain": "AFN", "afn": "AFN", "Guinean Franc": "GNF", "AFGHANI": "AFN"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"canceled": "Churned", "churned": "Churned", "ON HOLD": "Paused", "active": "Active", "cancelled": "Churned", "paused": "Paused", "ACTIVE": "Active"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TOLGA": "Tolga", "tolga": "Tolga", "SUNGAI LALANG": "Sungai Lalang", "Seporna": "Semporna", "hagen": "Hagen", "sungai lalang": "Sungai Lalang", "sunga lalang": "Sungai Lalang"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"VICE PRES": "Vice President", "vice presideent": "Vice President", "CEO": "Chief Executive Officer", "Admin Asst": "Administrative Assistant", "sales representative": "Sales Representative", "Vice Pres": "Vice President", "Admin": "Administrative Assistant", "c.e.o.": "Chief Executive Officer", "Rep": "Sales Representative", "Sales Rep": "Sales Representative", "vp": "Vice President", "ADMINISTRATIVE ASSISTANT": "Administrative Assistant", "v.p.": "Vice President", "C.E.O.": "Chief Executive Officer"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "Carol Ali", "unit": "inhc", "phone": "565-863-1941", "department": "mkt", "state": "Wyoming", "email": "pdbg@example.com", "extra": ""}, {"name": "Judy Moore", "unit": "LB ", "phone": "5074061762", "department": "MKTG", "state": "Minnesota", "email": "edknof@mail.com", "extra": ""}, {"name": "Heidi  Ali", "unit": "LITER", "phone": "5582494922", "department": "accounting  ", "state": "   wyoming", "email": "amcjbip@corp.io  ", "extra": ""}, {"name": "Grace Adams", "unit": " litre ", "phone": "(593)949-3652", "department": "growth", "state": "wv", "email": "  fogklib@corp.io", "extra": ""}, {"name": "Frank Johnson   ", "unit": "lb", "phone": "509-715-3559", "department": "Accounting", "state": "Wyoming", "email": "MOKN@TEST.ORG", "extra": ""}, {"name": "Frank Petrov", "unit": "liter", "phone": "501-238-1226", "department": " GROWTH ", "state": "Minnesota", "email": "MHEMCM@MAIL.COM ", "extra": ""}, {"name": "   Lena Diaz", "unit": "INCHES", "phone": "532.676.5239", "department": "Growth", "state": "   Virgin Islands, U.S.", "email": "NGDDHCH@CORP.IO", "extra": ""}, {"name": "Eve Cruz  ", "unit": "lbs", "phone": "(558)772-1499", "department": "Fin", "state": "   Wyoming", "email": "EILOHGJ@CORP.IO", "extra": ""}, {"name": "Omar Smith", "unit": "\"", "phone": "596.799.9816", "department": "Marketing", "state": "West Virginia", "email": "GDII@TEST.ORG   ", "extra": ""}, {"name": "Bob Park", "unit": "L", "phone": "5086038011", "department": "Marketing  ", "state": "MN", "email": "pjhdc@test.org", "extra": ""}, {"name": "Frank Reyes", "unit": "L", "phone": "556.093.6740", "department": "   Mkt", "state": "Wyoming", "email": " pcboipc@example.com ", "extra": ""}, {"name": "Eve Wong", "unit": "lb", "phone": "(505)715-3031", "department": "FINANCE", "state": "Ne   ", "email": "KELGIA@MAIL.COM", "extra": ""}, {"name": "Lena Adams  ", "unit": "Liters", "phone": "533.665.1914", "department": "MKTG", "state": "west virginia", "email": "oajaga@test.org   ", "extra": ""}, {"name": "Frank Ali", "unit": " LITRE", "phone": "5134775482", "department": "Marketing", "state": "WY", "email": "gapjbah@example.com", "extra": ""}, {"name": "Carol Fischer", "unit": "l", "phone": "(536)624-0706", "department": "Marketing", "state": "NE", "email": " CHGP@CORP.IO ", "extra": ""}, {"name": "Frank Moore", "unit": "INCHES ", "phone": "5017428815", "department": "Marketing", "state": " WY ", "email": "FNKBI@MAIL.COM", "extra": ""}, {"name": "", "unit": "", "phone": "", "department": "", "state": "", "email": "", "extra": ""}], "clean": [{"name": "Carol Ali", "unit": "in", "phone": "(565) 863-1941", "department": "Marketing", "state": "Wyoming", "email": "pdbg@example.com"}, {"name": "Judy Moore", "unit": "lb", "phone": "(507) 406-1762", "department": "Marketing", "state": "Minnesota", "email": "edknof@mail.com"}, {"name": "Heidi Ali", "unit": "L", "phone": "(558) 249-4922", "department": "Finance", "state": "Wyoming", "email": "amcjbip@corp.io"}, {"name": "Grace Adams", "unit": "L", "phone": "(593) 949-3652", "department": "Marketing", "state": "West Virginia", "email": "fogklib@corp.io"}, {"name": "Frank Johnson", "unit": "lb", "phone": "(509) 715-3559", "department": "Finance", "state": "Wyoming", "email": "mokn@test.org"}, {"name": "Frank Petrov", "unit": "L", "phone": "(501) 238-1226", "department": "Marketing", "state": "Minnesota", "email": "mhemcm@mail.com"}, {"name": "Lena Diaz", "unit": "in", "phone": "(532) 676-5239", "department": "Marketing", "state": "Virgin Islands, U.S.", "email": "ngddhch@corp.io"}, {"name": "Eve Cruz", "unit": "lb", "phone": "(558) 772-1499", "department": "Finance", "state": "Wyoming", "email": "eilohgj@corp.io"}, {"name": "Omar Smith", "unit": "in", "phone": "(596) 799-9816", "department": "Marketing", "state": "West Virginia", "email": "gdii@test.org"}, {"name": "Bob Park", "unit": "L", "phone": "(508) 603-8011", "department": "Marketing", "state": "Minnesota", "email": "pjhdc@test.org"}, {"name": "Frank Reyes", "unit": "L", "phone": "(556) 093-6740", "department": "Marketing", "state": "Wyoming", "email": "pcboipc@example.com"}, {"name": "Eve Wong", "unit": "lb", "phone": "(505) 715-3031", "department": "Finance", "state": "Nebraska", "email": "kelgia@mail.com"}, {"name": "Lena Adams", "unit": "L", "phone": "(533) 665-1914", "department": "Marketing", "state": "West Virginia", "email": "oajaga@test.org"}, {"name": "Frank Ali", "unit": "L", "phone": "(513) 477-5482", "department": "Marketing", "state": "Wyoming", "email": "gapjbah@example.com"}, {"name": "Carol Fischer", "unit": "L", "phone": "(536) 624-0706", "department": "Marketing", "state": "Nebraska", "email": "chgp@corp.io"}, {"name": "Frank Moore", "unit": "in", "phone": "(501) 742-8815", "department": "Marketing", "state": "Wyoming", "email": "fnkbi@mail.com"}], "dirty_cols": ["name", "unit", "phone", "department", "state", "email", "extra"], "clean_cols": ["name", "unit", "phone", "department", "state", "email"], "plan": {"dataset_summary": "17 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"inhc": "in", "LB": "lb", "LITER": "L", "litre": "L", "liter": "L", "INCHES": "in", "lbs": "lb", "\"": "in", "Liters": "L", "LITRE": "L", "l": "L"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mkt": "Marketing", "MKTG": "Marketing", "accounting": "Finance", "growth": "Marketing", "Accounting": "Finance", "GROWTH": "Marketing", "Growth": "Marketing", "Fin": "Finance", "Mkt": "Marketing", "FINANCE": "Finance"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"wyoming": "Wyoming", "wv": "West Virginia", "MN": "Minnesota", "Ne": "Nebraska", "west virginia": "West Virginia", "WY": "Wyoming", "NE": "Nebraska"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"company": "Vehement   ", "phone": "516-920-8001", "unit": "milliliter ", "signup_date": "10/28/2023", "job_title": "Admni Assistant"}, {"company": " Initech ", "phone": "5002191121", "unit": " milliliter", "signup_date": "2023-10-17", "job_title": " cvt.o. "}, {"company": " Hooli", "phone": "5079002028", "unit": "litre", "signup_date": "2023-01-16", "job_title": "Admin Asst"}, {"company": "Vehement   ", "phone": "5835961534", "unit": "milliliter", "signup_date": "2023-11-27", "job_title": "Chief Technology Officer"}, {"company": "Cyberdyne", "phone": "(562)997-5495", "unit": "litre", "signup_date": "2023-03-06", "job_title": "Admin"}, {"company": "Hooli", "phone": "553.360.3909", "unit": "mi", "signup_date": "2023-04-22", "job_title": "Product Manager"}, {"company": "Acme Inc", "phone": "5934404420", "unit": "mls", "signup_date": "11 Apr 2023", "job_title": "Chief Technology Officer"}, {"company": "Vehement", "phone": "(522)747-7768", "unit": " ltre ", "signup_date": "44990", "job_title": "Admin Asst"}, {"company": "Initech ", "phone": "596-182-4092", "unit": "   miile", "signup_date": "2023-01-10", "job_title": "C.T.O."}, {"company": "Acme Inc", "phone": "520.291.7711", "unit": "pounds", "signup_date": "20 Jan 2023", "job_title": "chief technology officer"}, {"company": "   Initech", "phone": "590-207-1125", "unit": "mipe", "signup_date": "20 May 2023", "job_title": "Product Manager"}, {"company": "Stark  Industries", "phone": "524.271.9153", "unit": "MILE", "signup_date": "5/28/2023", "job_title": "  Chief Technology Officer"}, {"company": "Stark Industries", "phone": "5050876773", "unit": "mls  ", "signup_date": "2023-06-06", "job_title": "admin asst   "}, {"company": "Stark Industries", "phone": "570-578-0097", "unit": "lb", "signup_date": "2023-06-20", "job_title": "Administrative Assistant"}, {"company": "Soylent Corp", "phone": "(516)322-2287", "unit": "LITERS", "signup_date": "7/7/2023", "job_title": "Admin"}, {"company": "Vehement", "phone": "577.263.2260", "unit": "L", "signup_date": "45063", "job_title": "Admin Assistant"}, {"company": "Acme Inc", "phone": "5600544027", "unit": "   L", "signup_date": "8/7/2023", "job_title": "ADMIN ASST"}, {"company": "   Soylent Corp", "phone": "545-120-8342", "unit": "ml", "signup_date": "15 Oct 2023", "job_title": "ADMIN ASSISTANT"}, {"company": "   Acme Inc", "phone": "(552)048-7785", "unit": "mls   ", "signup_date": "2023-09-01", "job_title": "cto"}, {"company": "Umbrella", "phone": "504-941-1414", "unit": "Litre", "signup_date": "2023-03-05", "job_title": "CHIEF TECHNOLOGY OFFICER"}, {"company": "   Umbrella", "phone": "(555)996-7184", "unit": "l", "signup_date": "2023-04-21", "job_title": "CTO"}, {"company": "Stark  Industries", "phone": "522.277.7501", "unit": "Lb", "signup_date": "1/24/2023", "job_title": "cto"}, {"company": "Stark Industries   ", "phone": "5870244024", "unit": "mls", "signup_date": "45160", "job_title": "Chief Technology Officer"}, {"company": " Globex ", "phone": "503-866-9425", "unit": "Lb", "signup_date": "6/11/2023", "job_title": "Admin Asst"}, {"company": "Umbrella", "phone": "595-668-6705", "unit": "lb", "signup_date": "1/1/2023", "job_title": "administrative assistant"}, {"company": " Wonka Co", "phone": "5474024741", "unit": " pound", "signup_date": "2023-10-09", "job_title": "Admin"}, {"company": " Hooli ", "phone": "5232626765", "unit": " Mile", "signup_date": "2 May 2023", "job_title": " PROD MANAGER"}, {"company": "Wonka Co", "phone": "552-131-3878", "unit": " lbs", "signup_date": "3/10/2023", "job_title": "Adinistrative  Assistant"}, {"company": " Wonka Co", "phone": "5406518073", "unit": "L", "signup_date": "11 Apr 2023", "job_title": " PROD MGR"}, {"company": "Wonka Co", "phone": "583-403-0199", "unit": "   LB", "signup_date": "2023-09-22", "job_title": "Product Manager"}, {"company": "Vehement", "phone": "5619046933", "unit": "milliliter", "signup_date": "7 Feb 2023", "job_title": "Administrative Assistant"}, {"company": "Vehement", "phone": "508-636-8853", "unit": "L", "signup_date": "44998", "job_title": "ADMNI"}, {"company": "Vehement", "phone": "522-093-0729", "unit": " Lb ", "signup_date": "45280", "job_title": "PRODUCT MANAGER"}, {"company": "Vehement   ", "phone": "516-920-8001", "unit": "milliliter ", "signup_date": "10/28/2023", "job_title": "Admni Assistant"}], "clean": [{"company": "Vehement", "phone": "(516) 920-8001", "unit": "mL", "signup_date": "2023-10-28", "job_title": "Administrative Assistant"}, {"company": "Initech", "phone": "(500) 219-1121", "unit": "mL", "signup_date": "2023-10-17", "job_title": "Chief Technology Officer"}, {"company": "Hooli", "phone": "(507) 900-2028", "unit": "L", "signup_date": "2023-01-16", "job_title": "Administrative Assistant"}, {"company": "Vehement", "phone": "(583) 596-1534", "unit": "mL", "signup_date": "2023-11-27", "job_title": "Chief Technology Officer"}, {"company": "Cyberdyne", "phone": "(562) 997-5495", "unit": "L", "signup_date": "2023-03-06", "job_title": "Administrative Assistant"}, {"company": "Hooli", "phone": "(553) 360-3909", "unit": "mi", "signup_date": "2023-04-22", "job_title": "Product Manager"}, {"company": "Acme Inc", "phone": "(593) 440-4420", "unit": "mL", "signup_date": "2023-04-11", "job_title": "Chief Technology Officer"}, {"company": "Vehement", "phone": "(522) 747-7768", "unit": "L", "signup_date": "2023-03-05", "job_title": "Administrative Assistant"}, {"company": "Initech", "phone": "(596) 182-4092", "unit": "mi", "signup_date": "2023-01-10", "job_title": "Chief Technology Officer"}, {"company": "Acme Inc", "phone": "(520) 291-7711", "unit": "lb", "signup_date": "2023-01-20", "job_title": "Chief Technology Officer"}, {"company": "Initech", "phone": "(590) 207-1125", "unit": "mi", "signup_date": "2023-05-20", "job_title": "Product Manager"}, {"company": "Stark Industries", "phone": "(524) 271-9153", "unit": "mi", "signup_date": "2023-05-28", "job_title": "Chief Technology Officer"}, {"company": "Stark Industries", "phone": "(505) 087-6773", "unit": "mL", "signup_date": "2023-06-06", "job_title": "Administrative Assistant"}, {"company": "Stark Industries", "phone": "(570) 578-0097", "unit": "lb", "signup_date": "2023-06-20", "job_title": "Administrative Assistant"}, {"company": "Soylent Corp", "phone": "(516) 322-2287", "unit": "L", "signup_date": "2023-07-07", "job_title": "Administrative Assistant"}, {"company": "Vehement", "phone": "(577) 263-2260", "unit": "L", "signup_date": "2023-05-17", "job_title": "Administrative Assistant"}, {"company": "Acme Inc", "phone": "(560) 054-4027", "unit": "L", "signup_date": "2023-08-07", "job_title": "Administrative Assistant"}, {"company": "Soylent Corp", "phone": "(545) 120-8342", "unit": "mL", "signup_date": "2023-10-15", "job_title": "Administrative Assistant"}, {"company": "Acme Inc", "phone": "(552) 048-7785", "unit": "mL", "signup_date": "2023-09-01", "job_title": "Chief Technology Officer"}, {"company": "Umbrella", "phone": "(504) 941-1414", "unit": "L", "signup_date": "2023-03-05", "job_title": "Chief Technology Officer"}, {"company": "Umbrella", "phone": "(555) 996-7184", "unit": "L", "signup_date": "2023-04-21", "job_title": "Chief Technology Officer"}, {"company": "Stark Industries", "phone": "(522) 277-7501", "unit": "lb", "signup_date": "2023-01-24", "job_title": "Chief Technology Officer"}, {"company": "Stark Industries", "phone": "(587) 024-4024", "unit": "mL", "signup_date": "2023-08-22", "job_title": "Chief Technology Officer"}, {"company": "Globex", "phone": "(503) 866-9425", "unit": "lb", "signup_date": "2023-06-11", "job_title": "Administrative Assistant"}, {"company": "Umbrella", "phone": "(595) 668-6705", "unit": "lb", "signup_date": "2023-01-01", "job_title": "Administrative Assistant"}, {"company": "Wonka Co", "phone": "(547) 402-4741", "unit": "lb", "signup_date": "2023-10-09", "job_title": "Administrative Assistant"}, {"company": "Hooli", "phone": "(523) 262-6765", "unit": "mi", "signup_date": "2023-05-02", "job_title": "Product Manager"}, {"company": "Wonka Co", "phone": "(552) 131-3878", "unit": "lb", "signup_date": "2023-03-10", "job_title": "Administrative Assistant"}, {"company": "Wonka Co", "phone": "(540) 651-8073", "unit": "L", "signup_date": "2023-04-11", "job_title": "Product Manager"}, {"company": "Wonka Co", "phone": "(583) 403-0199", "unit": "lb", "signup_date": "2023-09-22", "job_title": "Product Manager"}, {"company": "Vehement", "phone": "(561) 904-6933", "unit": "mL", "signup_date": "2023-02-07", "job_title": "Administrative Assistant"}, {"company": "Vehement", "phone": "(508) 636-8853", "unit": "L", "signup_date": "2023-03-13", "job_title": "Administrative Assistant"}, {"company": "Vehement", "phone": "(522) 093-0729", "unit": "lb", "signup_date": "2023-12-20", "job_title": "Product Manager"}], "dirty_cols": ["company", "phone", "unit", "signup_date", "job_title"], "clean_cols": ["company", "phone", "unit", "signup_date", "job_title"], "plan": {"dataset_summary": "34 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"milliliter": "mL", "litre": "L", "mls": "mL", "ltre": "L", "miile": "mi", "pounds": "lb", "mipe": "mi", "MILE": "mi", "LITERS": "L", "ml": "mL", "Litre": "L", "l": "L", "Lb": "lb", "pound": "lb", "Mile": "mi", "lbs": "lb", "LB": "lb"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Admni Assistant": "Administrative Assistant", "cvt.o.": "Chief Technology Officer", "Admin Asst": "Administrative Assistant", "Admin": "Administrative Assistant", "C.T.O.": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "admin asst": "Administrative Assistant", "Admin Assistant": "Administrative Assistant", "ADMIN ASST": "Administrative Assistant", "ADMIN ASSISTANT": "Administrative Assistant", "cto": "Chief Technology Officer", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "CTO": "Chief Technology Officer", "administrative assistant": "Administrative Assistant", "PROD MANAGER": "Product Manager", "Adinistrative Assistant": "Administrative Assistant", "PROD MGR": "Product Manager", "ADMNI": "Administrative Assistant", "PRODUCT MANAGER": "Product Manager"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "  Republic of Senegal", "name": "Priya Johnson ", "company": "Umbrella", "currency": "   Swiss Franc", "email": "GEOFND@TEST.ORG", "amount": "7.269,90"}, {"country": "republic of the sudan", "name": "Carol  Cruz", "company": "Cyberdyne", "currency": " Cdf ", "email": "   diihg@mail.com", "amount": "$2,894.64"}, {"country": "PRY", "name": "   Karl Khan", "company": "Acme  Inc", "currency": "chf", "email": " mfjbbcp@mail.com ", "amount": "3.591,93"}, {"country": "SD", "name": "Sara Petrov   ", "company": "Cyberdyne", "currency": "CDF", "email": "HDCD@CORP.IO", "amount": "6.871,37"}, {"country": "Papua New Guinea", "name": "Heidi Lee", "company": "Wonka  Co", "currency": "COU", "email": "MDAEC@EXAMPLE.COM ", "amount": "7.342,60"}, {"country": "sdn", "name": "Karl Khan", "company": " Cyberdyne ", "currency": "unidad de valor real", "email": "   BOMK@EXAMPLE.COM", "amount": "1.565,37"}, {"country": "sen", "name": "Bob  Adams", "company": "Soylent Corp", "currency": "CHF   ", "email": " linihgg@corp.io", "amount": "4.224,08"}, {"country": "Republic of the Sudna", "name": "Alice Diaz", "company": "Stark  Industries", "currency": "  Swiss Franc", "email": "  HCGN@CORP.IO", "amount": "$4,226.46"}, {"country": "   Paraguay", "name": "Frank  Petrov", "company": "Cyberdyne", "currency": "unidad de valor real  ", "email": "   IPFC@CORP.IO", "amount": "N/A"}, {"country": "SDAN", "name": "Mona Park", "company": "   Wonka Co", "currency": "   cdf", "email": "efoaopk@mail.com", "amount": "6.852,72"}, {"country": "Sudan", "name": "Eve Park", "company": "  Soylent Corp", "currency": "Unidad de Valor  Rsal", "email": "NBIEN@CORP.IO", "amount": "5.302,84"}, {"country": "NOR", "name": "Bob Moore", "company": "Cyberdyne", "currency": "CHF", "email": "CNCH@EXAMPLE.COM", "amount": "$1,831.44"}, {"country": "NOR", "name": "Eve Diaz", "company": "Soylent Corp", "currency": "Swiss Franc", "email": "HBHB@TEST.ORG", "amount": "8.500,83"}, {"country": "PNG", "name": "Frank Johnson", "company": "Initech", "currency": " Swiss Franc", "email": "gmpo@example.com", "amount": "$7,341.03"}, {"country": " SEN", "name": "   Grace Novak", "company": "Vehement", "currency": "swiss franc", "email": "HMOICB@MAIL.COM", "amount": "$3,834.54"}, {"country": "independent state of papua new guinea", "name": "Omar Reyes", "company": "Hooli", "currency": "CDF", "email": " EDHNA@MAIL.COM", "amount": "na"}, {"country": "norway", "name": "Omar Adams", "company": "Stark Industries ", "currency": "CHF", "email": " dlicp@example.com ", "amount": "5.400,34"}, {"country": "", "name": "", "company": "", "currency": "", "email": "", "amount": ""}], "clean": [{"country": "Senegal", "name": "Priya Johnson", "company": "Umbrella", "currency": "CHF", "email": "geofnd@test.org", "amount": 7269.9}, {"country": "Sudan", "name": "Carol Cruz", "company": "Cyberdyne", "currency": "CDF", "email": "diihg@mail.com", "amount": 2894.64}, {"country": "Paraguay", "name": "Karl Khan", "company": "Acme Inc", "currency": "CHF", "email": "mfjbbcp@mail.com", "amount": 3591.93}, {"country": "Sudan", "name": "Sara Petrov", "company": "Cyberdyne", "currency": "CDF", "email": "hdcd@corp.io", "amount": 6871.37}, {"country": "Papua New Guinea", "name": "Heidi Lee", "company": "Wonka Co", "currency": "COU", "email": "mdaec@example.com", "amount": 7342.6}, {"country": "Sudan", "name": "Karl Khan", "company": "Cyberdyne", "currency": "COU", "email": "bomk@example.com", "amount": 1565.37}, {"country": "Senegal", "name": "Bob Adams", "company": "Soylent Corp", "currency": "CHF", "email": "linihgg@corp.io", "amount": 4224.08}, {"country": "Sudan", "name": "Alice Diaz", "company": "Stark Industries", "currency": "CHF", "email": "hcgn@corp.io", "amount": 4226.46}, {"country": "Paraguay", "name": "Frank Petrov", "company": "Cyberdyne", "currency": "COU", "email": "ipfc@corp.io", "amount": NaN}, {"country": "Sudan", "name": "Mona Park", "company": "Wonka Co", "currency": "CDF", "email": "efoaopk@mail.com", "amount": 6852.72}, {"country": "Sudan", "name": "Eve Park", "company": "Soylent Corp", "currency": "COU", "email": "nbien@corp.io", "amount": 5302.84}, {"country": "Norway", "name": "Bob Moore", "company": "Cyberdyne", "currency": "CHF", "email": "cnch@example.com", "amount": 1831.44}, {"country": "Norway", "name": "Eve Diaz", "company": "Soylent Corp", "currency": "CHF", "email": "hbhb@test.org", "amount": 8500.83}, {"country": "Papua New Guinea", "name": "Frank Johnson", "company": "Initech", "currency": "CHF", "email": "gmpo@example.com", "amount": 7341.03}, {"country": "Senegal", "name": "Grace Novak", "company": "Vehement", "currency": "CHF", "email": "hmoicb@mail.com", "amount": 3834.54}, {"country": "Papua New Guinea", "name": "Omar Reyes", "company": "Hooli", "currency": "CDF", "email": "edhna@mail.com", "amount": NaN}, {"country": "Norway", "name": "Omar Adams", "company": "Stark Industries", "currency": "CHF", "email": "dlicp@example.com", "amount": 5400.34}], "dirty_cols": ["country", "name", "company", "currency", "email", "amount"], "clean_cols": ["country", "name", "company", "currency", "email", "amount"], "plan": {"dataset_summary": "18 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Republic of Senegal": "Senegal", "republic of the sudan": "Sudan", "PRY": "Paraguay", "SD": "Sudan", "sdn": "Sudan", "sen": "Senegal", "Republic of the Sudna": "Sudan", "SDAN": "Sudan", "NOR": "Norway", "PNG": "Papua New Guinea", "SEN": "Senegal", "independent state of papua new guinea": "Papua New Guinea", "norway": "Norway"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Swiss Franc": "CHF", "Cdf": "CDF", "chf": "CHF", "unidad de valor real": "COU", "cdf": "CDF", "Unidad de Valor Rsal": "COU", "swiss franc": "CHF"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": []}}
{"dirty": [{"amount": "8.280,52", "city": "Huangmei", "state": "West  Virginia", "industry": "it", "signup_date": "45000", "country": "AS", "extra": ""}, {"amount": "TBD", "city": "huangmei", "state": "Mi", "industry": "Sofwtare", "signup_date": "45030", "country": "lux   ", "extra": ""}, {"amount": "1.452,75", "city": "Minna", "state": "HI", "industry": "comms", "signup_date": "2023-10-14", "country": "LUX", "extra": ""}, {"amount": "5.701,68", "city": " Kwinana", "state": " pa", "industry": "EdTegh", "signup_date": "2023-08-07", "country": "AS", "extra": ""}, {"amount": "$1,302.51", "city": " HUANGMEI ", "state": "Hawaii", "industry": "Edtech", "signup_date": "21 May 2023", "country": "LU", "extra": ""}, {"amount": "$238.17", "city": "MINNA", "state": " PA ", "industry": "EDTECH", "signup_date": "1/5/2023", "country": "AS", "extra": ""}, {"amount": "2.466,82", "city": " huangmei ", "state": "West Virginia", "industry": "Technology  ", "signup_date": "45124", "country": "As", "extra": ""}, {"amount": "$3,645.08", "city": "San Diego", "state": "ME", "industry": "  education", "signup_date": "2023-10-12", "country": "American Samoa", "extra": ""}, {"amount": "8.852,47", "city": "Huangmei", "state": "hawaii", "industry": "TECH   ", "signup_date": "44963", "country": "Grand Duchy of Luxembourg", "extra": ""}, {"amount": "6.008,03", "city": "San Diego", "state": "michihan", "industry": "COMSM", "signup_date": "2023-03-07", "country": " as", "extra": ""}, {"amount": "$2,968.51", "city": " Huangmei", "state": "PA", "industry": "education", "signup_date": "2023-11-22", "country": "luxembourg", "extra": ""}, {"amount": "5.311,78", "city": "Huangmei", "state": "hi", "industry": "Software", "signup_date": "45029", "country": "Grand Duchy of Luxembourg", "extra": ""}, {"amount": "4.105,88", "city": "minna", "state": "HI", "industry": "EDUCATION", "signup_date": "2023-03-04", "country": "  as", "extra": ""}, {"amount": "633,76", "city": "Kwinana", "state": "hawaii", "industry": "eduation", "signup_date": "3/18/2023", "country": "   luxembourg", "extra": ""}, {"amount": "8.534,43", "city": "San Diego", "state": " MI ", "industry": "education", "signup_date": "26 Aug 2023", "country": "as", "extra": ""}, {"amount": "5.392,14", "city": "huangmei", "state": "pa ", "industry": "Software", "signup_date": "45247", "country": "luxembourg", "extra": ""}, {"amount": "4.334,87", "city": "San Diego", "state": "HI", "industry": "Software", "signup_date": "2023-05-03", "country": " luxembourg ", "extra": ""}, {"amount": "634,11", "city": " Minna ", "state": "WEST VRGINIA", "industry": "TELECOMMUNICATIONS", "signup_date": "22 Jan 2023", "country": "american samoa ", "extra": ""}, {"amount": "$8,015.42", "city": "SAN DIEGO", "state": "West Virginia", "industry": "Ed  ", "signup_date": "7/8/2023", "country": "AS   ", "extra": ""}, {"amount": "6.584,57", "city": "Kwinana", "state": "PA", "industry": "Telecommunications", "signup_date": "45092", "country": "LU", "extra": ""}, {"amount": "6.040,22", "city": "minna", "state": " pennsylvania", "industry": "EdTech   ", "signup_date": "11/8/2023", "country": "as", "extra": ""}, {"amount": "$160.54", "city": "Huangmei", "state": "Michigan", "industry": "IT", "signup_date": "2023-07-07", "country": " Lu ", "extra": ""}, {"amount": "4.334,87", "city": "San Diego", "state": "HI", "industry": "Software", "signup_date": "2023-05-03", "country": " luxembourg ", "extra": ""}], "clean": [{"amount": 8280.52, "city": "Huangmei", "state": "West Virginia", "industry": "Technology", "signup_date": "2023-03-15", "country": "American Samoa"}, {"amount": NaN, "city": "Huangmei", "state": "Michigan", "industry": "Technology", "signup_date": "2023-04-14", "country": "Luxembourg"}, {"amount": 1452.75, "city": "Minna", "state": "Hawaii", "industry": "Telecommunications", "signup_date": "2023-10-14", "country": "Luxembourg"}, {"amount": 5701.68, "city": "Kwinana", "state": "Pennsylvania", "industry": "Education", "signup_date": "2023-08-07", "country": "American Samoa"}, {"amount": 1302.51, "city": "Huangmei", "state": "Hawaii", "industry": "Education", "signup_date": "2023-05-21", "country": "Luxembourg"}, {"amount": 238.17, "city": "Minna", "state": "Pennsylvania", "industry": "Education", "signup_date": "2023-01-05", "country": "American Samoa"}, {"amount": 2466.82, "city": "Huangmei", "state": "West Virginia", "industry": "Technology", "signup_date": "2023-07-17", "country": "American Samoa"}, {"amount": 3645.08, "city": "San Diego", "state": "Maine", "industry": "Education", "signup_date": "2023-10-12", "country": "American Samoa"}, {"amount": 8852.47, "city": "Huangmei", "state": "Hawaii", "industry": "Technology", "signup_date": "2023-02-06", "country": "Luxembourg"}, {"amount": 6008.03, "city": "San Diego", "state": "Michigan", "industry": "Telecommunications", "signup_date": "2023-03-07", "country": "American Samoa"}, {"amount": 2968.51, "city": "Huangmei", "state": "Pennsylvania", "industry": "Education", "signup_date": "2023-11-22", "country": "Luxembourg"}, {"amount": 5311.78, "city": "Huangmei", "state": "Hawaii", "industry": "Technology", "signup_date": "2023-04-13", "country": "Luxembourg"}, {"amount": 4105.88, "city": "Minna", "state": "Hawaii", "industry": "Education", "signup_date": "2023-03-04", "country": "American Samoa"}, {"amount": 633.76, "city": "Kwinana", "state": "Hawaii", "industry": "Education", "signup_date": "2023-03-18", "country": "Luxembourg"}, {"amount": 8534.43, "city": "San Diego", "state": "Michigan", "industry": "Education", "signup_date": "2023-08-26", "country": "American Samoa"}, {"amount": 5392.14, "city": "Huangmei", "state": "Pennsylvania", "industry": "Technology", "signup_date": "2023-11-17", "country": "Luxembourg"}, {"amount": 4334.87, "city": "San Diego", "state": "Hawaii", "industry": "Technology", "signup_date": "2023-05-03", "country": "Luxembourg"}, {"amount": 634.11, "city": "Minna", "state": "West Virginia", "industry": "Telecommunications", "signup_date": "2023-01-22", "country": "American Samoa"}, {"amount": 8015.42, "city": "San Diego", "state": "West Virginia", "industry": "Education", "signup_date": "2023-07-08", "country": "American Samoa"}, {"amount": 6584.57, "city": "Kwinana", "state": "Pennsylvania", "industry": "Telecommunications", "signup_date": "2023-06-15", "country": "Luxembourg"}, {"amount": 6040.22, "city": "Minna", "state": "Pennsylvania", "industry": "Education", "signup_date": "2023-11-08", "country": "American Samoa"}, {"amount": 160.54, "city": "Huangmei", "state": "Michigan", "industry": "Technology", "signup_date": "2023-07-07", "country": "Luxembourg"}], "dirty_cols": ["amount", "city", "state", "industry", "signup_date", "country", "extra"], "clean_cols": ["amount", "city", "state", "industry", "signup_date", "country"], "plan": {"dataset_summary": "23 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"huangmei": "Huangmei", "HUANGMEI": "Huangmei", "MINNA": "Minna", "minna": "Minna", "SAN DIEGO": "San Diego"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mi": "Michigan", "HI": "Hawaii", "pa": "Pennsylvania", "PA": "Pennsylvania", "ME": "Maine", "hawaii": "Hawaii", "michihan": "Michigan", "hi": "Hawaii", "MI": "Michigan", "WEST VRGINIA": "West Virginia", "pennsylvania": "Pennsylvania"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"it": "Technology", "Sofwtare": "Technology", "comms": "Telecommunications", "EdTegh": "Education", "Edtech": "Education", "EDTECH": "Education", "education": "Education", "TECH": "Technology", "COMSM": "Telecommunications", "Software": "Technology", "EDUCATION": "Education", "eduation": "Education", "TELECOMMUNICATIONS": "Telecommunications", "Ed": "Education", "EdTech": "Education", "IT": "Technology"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AS": "American Samoa", "lux": "Luxembourg", "LUX": "Luxembourg", "LU": "Luxembourg", "As": "American Samoa", "Grand Duchy of Luxembourg": "Luxembourg", "as": "American Samoa", "luxembourg": "Luxembourg", "american samoa": "American Samoa", "Lu": "Luxembourg"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "Admin Asst", "email": "FKJOFL@MAIL.COM", "unit": "miel", "phone": "566.080.4073", "industry": "  Bankigg", "signup_date": "14 Jul 2023"}, {"job_title": "CEO", "email": "  HCCIOMM@TEST.ORG", "unit": "Km", "phone": "(549)207-7993", "industry": "INDUSTRIAL", "signup_date": "5/3/2023"}, {"job_title": "chief executive officer", "email": "dmbia@example.com", "unit": "mile", "phone": "5132302694", "industry": "E-Commerce", "signup_date": "45155"}, {"job_title": "eng mgr", "email": "hdgd@mail.com", "unit": "   millilitvr", "phone": "526-121-9038", "industry": "Medical", "signup_date": "2023-12-20"}, {"job_title": "chief  executive officer", "email": "NHEJ@MAIL.COM", "unit": "KM", "phone": "524.046.3538", "industry": "Healthcare", "signup_date": "19 Jun 2023"}, {"job_title": "C..E.O.  ", "email": "  LKFLNK@MAIL.COM", "unit": "MLS", "phone": "(570)878-2600", "industry": "INDUSTRIAL", "signup_date": "45095"}, {"job_title": "Administrative Assistant", "email": "bfmdje@mail.com", "unit": "mls", "phone": "5648099412", "industry": "Industrial", "signup_date": "2 Oct 2023"}, {"job_title": "C..EO.", "email": "LPGCHIP@TEST.ORG", "unit": "mi", "phone": "5386546415", "industry": "Industrial", "signup_date": "45205"}, {"job_title": "Admin", "email": "madg@mail.com", "unit": "milliliter", "phone": "503.201.8843", "industry": "Financial Services", "signup_date": "2023-12-26"}, {"job_title": "c.e.o.", "email": "KKMNEM@EXAMPLE.COM", "unit": "Liter", "phone": "595-917-9804", "industry": "Retail & E-commerce", "signup_date": "2023-03-07"}, {"job_title": "SNR  ENGINEER", "email": "FNJEM@CORP.IO", "unit": "ml", "phone": "5420983269", "industry": "retail", "signup_date": "8/20/2023"}, {"job_title": "Sneior Engineer", "email": " ONPB@EXAMPLE.COM ", "unit": "MILES", "phone": "553-913-1127", "industry": "Finance", "signup_date": "8 Nov 2023"}, {"job_title": "Chief Executive Officer", "email": " EMCF@CORP.IO ", "unit": "milliliter", "phone": "596-971-4084", "industry": "Medical   ", "signup_date": "20 Sep 2023"}, {"job_title": "Senior Eng", "email": "magnk@mail.com", "unit": "ml", "phone": "(586)079-6832", "industry": "Medical", "signup_date": "2023-05-01"}, {"job_title": "Administrative Assistant", "email": "ghpp@test.org", "unit": "   ml", "phone": "500-376-9316", "industry": " retail ", "signup_date": "45241"}, {"job_title": "En Manager", "email": "ncmenl@corp.io", "unit": "ml", "phone": "515-983-8688", "industry": "  finserv", "signup_date": "10/7/2023"}, {"job_title": "Chief Executive Officer", "email": "epomikp@corp.io", "unit": "kvlometers", "phone": "5077309085", "industry": "RETAIL", "signup_date": "8 Jun 2023"}, {"job_title": "c.e.o.", "email": "  dfen@corp.io", "unit": "mL", "phone": "564.784.8350", "industry": "FINNANCE", "signup_date": "44932"}, {"job_title": "eng mgr", "email": "fhoppcb@corp.io", "unit": "mile", "phone": "(575)397-2633", "industry": "Finaoce", "signup_date": "2023-09-07"}, {"job_title": "ENGINEERING MGGR", "email": "bkad@mail.com", "unit": "  Km", "phone": "509.697.6229", "industry": "Health Care", "signup_date": "45037"}, {"job_title": "Eng Mgr", "email": "OBBDAL@MAIL.COM ", "unit": "kilometer", "phone": "566.954.7451", "industry": "  finance", "signup_date": "45150"}, {"job_title": "C.E.O.", "email": " HLMPJB@TEST.ORG ", "unit": "Ml", "phone": "542.082.8965", "industry": "Health  Care", "signup_date": "27 Mar 2023"}, {"job_title": "eng manaegr", "email": "kdcde@corp.io", "unit": "Mi", "phone": "(510)137-7122", "industry": "Finance", "signup_date": "2023-10-09"}, {"job_title": "ENGINEERING MANGER", "email": "hcepc@example.com", "unit": "mls", "phone": "524-023-2034", "industry": "health   ", "signup_date": "2023-07-12"}, {"job_title": "Eng Manager", "email": "   bbie@example.com", "unit": "   ml", "phone": "503.460.9396", "industry": "retail & e-commerce", "signup_date": "45086"}, {"job_title": " Engineering Manager", "email": "ciegg@mail.com", "unit": "l", "phone": "(524)005-7265", "industry": "RETAIL", "signup_date": "45257"}, {"job_title": "engineering manager", "email": " DJNL@CORP.IO ", "unit": "km", "phone": "5985318322", "industry": " Healthcare ", "signup_date": "2023-10-19"}, {"job_title": "C.E.. ", "email": "HPFD@CORP.IO", "unit": " miles", "phone": "5039761703", "industry": "Retail", "signup_date": "45267"}, {"job_title": " Amin", "email": "pmfpok@test.org", "unit": "liter   ", "phone": "(584)639-7383", "industry": "manufacturing", "signup_date": "15 Sep 2023"}, {"job_title": "CHIEF EXECUTIVE OFFICER", "email": "GJLPAAO@CORP.IO", "unit": "miles", "phone": "(511)238-8871", "industry": "Rtail", "signup_date": "4/11/2023"}, {"job_title": "Snr Engineer  ", "email": "dbjil@example.com", "unit": "L", "phone": "(505)049-9531", "industry": "E-commerce", "signup_date": "2023-12-06"}, {"job_title": "Admin Assistant", "email": "HFPH@TEST.ORG", "unit": " Mile", "phone": "502.420.0057", "industry": "Retail", "signup_date": "44952"}, {"job_title": "SR. ENGINEER", "email": "pecioj@example.com", "unit": "mL", "phone": "582.242.0351", "industry": "health  ", "signup_date": "23 Oct 2023"}, {"job_title": "  Administrative Assistant", "email": " NOFBDJ@MAIL.COM", "unit": "mi", "phone": "539.421.4934", "industry": "healthcare", "signup_date": "8 Jan 2023"}, {"job_title": "   ENGINEERING MANAGER", "email": " PBILL@EXAMPLE.COM ", "unit": "litre", "phone": "(510)727-2061", "industry": "finserv", "signup_date": "2023-12-25"}, {"job_title": "engineering manager", "email": "  BBDHFL@CORP.IO", "unit": "  mi", "phone": "563-727-2354", "industry": " Finserv", "signup_date": "24 May 2023"}, {"job_title": "Senior Engineer ", "email": "IJCBEAK@CORP.IO", "unit": "L", "phone": "586-236-9907", "industry": " retail", "signup_date": "2023-07-08"}, {"job_title": "Admin Assiistant ", "email": "AGLKI@CORP.IO", "unit": "miles", "phone": "535-192-1744", "industry": "Finance", "signup_date": "2023-03-10"}, {"job_title": "Engineering Manager", "email": " GBHODA@MAIL.COM", "unit": "ml", "phone": "5631564547", "industry": "Halthcare", "signup_date": "9/2/2023"}, {"job_title": "eng gr", "email": "DGEH@MAIL.COM", "unit": "mls", "phone": "563.041.5069", "industry": "Healtcare", "signup_date": "24 Mar 2023"}, {"job_title": "administrative assistant", "email": "afgeg@test.org", "unit": "mile", "phone": "583.711.1410", "industry": " Health ", "signup_date": "1 Dec 2023"}, {"job_title": "Sneior Engineer", "email": " ONPB@EXAMPLE.COM ", "unit": "MILES", "phone": "553-913-1127", "industry": "Finance", "signup_date": "8 Nov 2023"}, {"job_title": "SR. ENGINEER", "email": "pecioj@example.com", "unit": "mL", "phone": "582.242.0351", "industry": "health  ", "signup_date": "23 Oct 2023"}], "clean": [{"job_title": "Administrative Assistant", "email": "fkjofl@mail.com", "unit": "mi", "phone": "(566) 080-4073", "industry": "Financial Services", "signup_date": "2023-07-14"}, {"job_title": "Chief Executive Officer", "email": "hcciomm@test.org", "unit": "km", "phone": "(549) 207-7993", "industry": "Manufacturing", "signup_date": "2023-05-03"}, {"job_title": "Chief Executive Officer", "email": "dmbia@example.com", "unit": "mi", "phone": "(513) 230-2694", "industry": "Retail", "signup_date": "2023-08-17"}, {"job_title": "Engineering Manager", "email": "hdgd@mail.com", "unit": "mL", "phone": "(526) 121-9038", "industry": "Healthcare", "signup_date": "2023-12-20"}, {"job_title": "Chief Executive Officer", "email": "nhej@mail.com", "unit": "km", "phone": "(524) 046-3538", "industry": "Healthcare", "signup_date": "2023-06-19"}, {"job_title": "Chief Executive Officer", "email": "lkflnk@mail.com", "unit": "mL", "phone": "(570) 878-2600", "industry": "Manufacturing", "signup_date": "2023-06-18"}, {"job_title": "Administrative Assistant", "email": "bfmdje@mail.com", "unit": "mL", "phone": "(564) 809-9412", "industry": "Manufacturing", "signup_date": "2023-10-02"}, {"job_title": "Chief Executive Officer", "email": "lpgchip@test.org", "unit": "mi", "phone": "(538) 654-6415", "industry": "Manufacturing", "signup_date": "2023-10-06"}, {"job_title": "Administrative Assistant", "email": "madg@mail.com", "unit": "mL", "phone": "(503) 201-8843", "industry": "Financial Services", "signup_date": "2023-12-26"}, {"job_title": "Chief Executive Officer", "email": "kkmnem@example.com", "unit": "L", "phone": "(595) 917-9804", "industry": "Retail", "signup_date": "2023-03-07"}, {"job_title": "Senior Engineer", "email": "fnjem@corp.io", "unit": "mL", "phone": "(542) 098-3269", "industry": "Retail", "signup_date": "2023-08-20"}, {"job_title": "Senior Engineer", "email": "onpb@example.com", "unit": "mi", "phone": "(553) 913-1127", "industry": "Financial Services", "signup_date": "2023-11-08"}, {"job_title": "Chief Executive Officer", "email": "emcf@corp.io", "unit": "mL", "phone": "(596) 971-4084", "industry": "Healthcare", "signup_date": "2023-09-20"}, {"job_title": "Senior Engineer", "email": "magnk@mail.com", "unit": "mL", "phone": "(586) 079-6832", "industry": "Healthcare", "signup_date": "2023-05-01"}, {"job_title": "Administrative Assistant", "email": "ghpp@test.org", "unit": "mL", "phone": "(500) 376-9316", "industry": "Retail", "signup_date": "2023-11-11"}, {"job_title": "Engineering Manager", "email": "ncmenl@corp.io", "unit": "mL", "phone": "(515) 983-8688", "industry": "Financial Services", "signup_date": "2023-10-07"}, {"job_title": "Chief Executive Officer", "email": "epomikp@corp.io", "unit": "km", "phone": "(507) 730-9085", "industry": "Retail", "signup_date": "2023-06-08"}, {"job_title": "Chief Executive Officer", "email": "dfen@corp.io", "unit": "mL", "phone": "(564) 784-8350", "industry": "Financial Services", "signup_date": "2023-01-06"}, {"job_title": "Engineering Manager", "email": "fhoppcb@corp.io", "unit": "mi", "phone": "(575) 397-2633", "industry": "Financial Services", "signup_date": "2023-09-07"}, {"job_title": "Engineering Manager", "email": "bkad@mail.com", "unit": "km", "phone": "(509) 697-6229", "industry": "Healthcare", "signup_date": "2023-04-21"}, {"job_title": "Engineering Manager", "email": "obbdal@mail.com", "unit": "km", "phone": "(566) 954-7451", "industry": "Financial Services", "signup_date": "2023-08-12"}, {"job_title": "Chief Executive Officer", "email": "hlmpjb@test.org", "unit": "mL", "phone": "(542) 082-8965", "industry": "Healthcare", "signup_date": "2023-03-27"}, {"job_title": "Engineering Manager", "email": "kdcde@corp.io", "unit": "mi", "phone": "(510) 137-7122", "industry": "Financial Services", "signup_date": "2023-10-09"}, {"job_title": "Engineering Manager", "email": "hcepc@example.com", "unit": "mL", "phone": "(524) 023-2034", "industry": "Healthcare", "signup_date": "2023-07-12"}, {"job_title": "Engineering Manager", "email": "bbie@example.com", "unit": "mL", "phone": "(503) 460-9396", "industry": "Retail", "signup_date": "2023-06-09"}, {"job_title": "Engineering Manager", "email": "ciegg@mail.com", "unit": "L", "phone": "(524) 005-7265", "industry": "Retail", "signup_date": "2023-11-27"}, {"job_title": "Engineering Manager", "email": "djnl@corp.io", "unit": "km", "phone": "(598) 531-8322", "industry": "Healthcare", "signup_date": "2023-10-19"}, {"job_title": "Chief Executive Officer", "email": "hpfd@corp.io", "unit": "mi", "phone": "(503) 976-1703", "industry": "Retail", "signup_date": "2023-12-07"}, {"job_title": "Administrative Assistant", "email": "pmfpok@test.org", "unit": "L", "phone": "(584) 639-7383", "industry": "Manufacturing", "signup_date": "2023-09-15"}, {"job_title": "Chief Executive Officer", "email": "gjlpaao@corp.io", "unit": "mi", "phone": "(511) 238-8871", "industry": "Retail", "signup_date": "2023-04-11"}, {"job_title": "Senior Engineer", "email": "dbjil@example.com", "unit": "L", "phone": "(505) 049-9531", "industry": "Retail", "signup_date": "2023-12-06"}, {"job_title": "Administrative Assistant", "email": "hfph@test.org", "unit": "mi", "phone": "(502) 420-0057", "industry": "Retail", "signup_date": "2023-01-26"}, {"job_title": "Senior Engineer", "email": "pecioj@example.com", "unit": "mL", "phone": "(582) 242-0351", "industry": "Healthcare", "signup_date": "2023-10-23"}, {"job_title": "Administrative Assistant", "email": "nofbdj@mail.com", "unit": "mi", "phone": "(539) 421-4934", "industry": "Healthcare", "signup_date": "2023-01-08"}, {"job_title": "Engineering Manager", "email": "pbill@example.com", "unit": "L", "phone": "(510) 727-2061", "industry": "Financial Services", "signup_date": "2023-12-25"}, {"job_title": "Engineering Manager", "email": "bbdhfl@corp.io", "unit": "mi", "phone": "(563) 727-2354", "industry": "Financial Services", "signup_date": "2023-05-24"}, {"job_title": "Senior Engineer", "email": "ijcbeak@corp.io", "unit": "L", "phone": "(586) 236-9907", "industry": "Retail", "signup_date": "2023-07-08"}, {"job_title": "Administrative Assistant", "email": "aglki@corp.io", "unit": "mi", "phone": "(535) 192-1744", "industry": "Financial Services", "signup_date": "2023-03-10"}, {"job_title": "Engineering Manager", "email": "gbhoda@mail.com", "unit": "mL", "phone": "(563) 156-4547", "industry": "Healthcare", "signup_date": "2023-09-02"}, {"job_title": "Engineering Manager", "email": "dgeh@mail.com", "unit": "mL", "phone": "(563) 041-5069", "industry": "Healthcare", "signup_date": "2023-03-24"}, {"job_title": "Administrative Assistant", "email": "afgeg@test.org", "unit": "mi", "phone": "(583) 711-1410", "industry": "Healthcare", "signup_date": "2023-12-01"}], "dirty_cols": ["job_title", "email", "unit", "phone", "industry", "signup_date"], "clean_cols": ["job_title", "email", "unit", "phone", "industry", "signup_date"], "plan": {"dataset_summary": "43 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Admin Asst": "Administrative Assistant", "CEO": "Chief Executive Officer", "chief executive officer": "Chief Executive Officer", "eng mgr": "Engineering Manager", "chief  executive officer": "Chief Executive Officer", "C..E.O.": "Chief Executive Officer", "C..EO.": "Chief Executive Officer", "Admin": "Administrative Assistant", "c.e.o.": "Chief Executive Officer", "SNR ENGINEER": "Senior Engineer", "Sneior Engineer": "Senior Engineer", "Senior Eng": "Senior Engineer", "En Manager": "Engineering Manager", "ENGINEERING MGGR": "Engineering Manager", "Eng Mgr": "Engineering Manager", "C.E.O.": "Chief Executive Officer", "eng manaegr": "Engineering Manager", "ENGINEERING MANGER": "Engineering Manager", "Eng Manager": "Engineering Manager", "engineering manager": "Engineering Manager", "C.E..": "Chief Executive Officer", "Amin": "Administrative Assistant", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "Snr Engineer": "Senior Engineer", "Admin Assistant": "Administrative Assistant", "SR. ENGINEER": "Senior Engineer", "ENGINEERING MANAGER": "Engineering Manager", "Admin Assiistant": "Administrative Assistant", "eng gr": "Engineering Manager", "administrative assistant": "Administrative Assistant"}, "rationale": "Unified 30 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miel": "mi", "Km": "km", "mile": "mi", "millilitvr": "mL", "KM": "km", "MLS": "mL", "mls": "mL", "milliliter": "mL", "Liter": "L", "ml": "mL", "MILES": "mi", "kvlometers": "km", "kilometer": "km", "Ml": "mL", "Mi": "mi", "l": "L", "miles": "mi", "liter": "L", "Mile": "mi", "litre": "L"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Bankigg": "Financial Services", "INDUSTRIAL": "Manufacturing", "E-Commerce": "Retail", "Medical": "Healthcare", "Industrial": "Manufacturing", "Retail & E-commerce": "Retail", "retail": "Retail", "Finance": "Financial Services", "finserv": "Financial Services", "RETAIL": "Retail", "FINNANCE": "Financial Services", "Finaoce": "Financial Services", "Health Care": "Healthcare", "finance": "Financial Services", "health": "Healthcare", "retail & e-commerce": "Retail", "manufacturing": "Manufacturing", "Rtail": "Retail", "E-commerce": "Retail", "healthcare": "Healthcare", "Finserv": "Financial Services", "Halthcare": "Healthcare", "Healtcare": "Healthcare", "Health": "Healthcare"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"currency": "CNY", "job_title": "Vice President", "status": "unpaid", "company": "Initech", "rate": "52.1%", "industry": "MFG"}, {"currency": "Yuan Renminbi", "job_title": "   Rep", "status": " Refunded", "company": "Acme  Inc", "rate": "14.9%", "industry": "Industrial"}, {"currency": "cny", "job_title": "Sales Representative", "status": " Paid ", "company": " Vehement ", "rate": "65.6%", "industry": "  real estate"}, {"currency": "yuan renminbi", "job_title": "Acct Exec", "status": "paid", "company": " Umbrella", "rate": "30.2%", "industry": "realestate"}, {"currency": " BIF ", "job_title": "  Acct Exec", "status": "OVERDUE", "company": "Acme Inc ", "rate": "31.1%", "industry": "PROPERTY"}, {"currency": "  bif", "job_title": " Ae ", "status": "Overdue", "company": "Stark Industries ", "rate": "84.2%", "industry": "property"}, {"currency": "DKK", "job_title": "Sales Representative", "status": "Paid", "company": " Vehement ", "rate": "4.4%", "industry": "RealEstate"}, {"currency": "Bif", "job_title": "Account Executive", "status": "refunded", "company": "Cyberdyne", "rate": "40.4%", "industry": "Poptech"}, {"currency": "BIF ", "job_title": "Rep  ", "status": "Paid", "company": "  Stark Industries", "rate": "0.4%", "industry": "PropTech"}, {"currency": "Burundi Franc", "job_title": "rep", "status": "Refunded ", "company": "Vehement", "rate": "6.3%", "industry": "PROPERTY"}, {"currency": "BIF", "job_title": "Vice President", "status": " PENDING ", "company": "Acme  Inc", "rate": "55.1%", "industry": "Real EEstate"}, {"currency": "Cny", "job_title": "Vice President", "status": "Pending", "company": "Soylent Corp", "rate": "7.1%", "industry": " Industriaal "}, {"currency": "BIF  ", "job_title": "ae   ", "status": "pendnig", "company": "Soylent Corp", "rate": "36.5%", "industry": " MFG"}, {"currency": " CNY ", "job_title": "sales representative  ", "status": "pending", "company": "Stark Industries", "rate": "44.4%", "industry": "MANUFACTURING"}, {"currency": "dkk", "job_title": "VICE  PRES", "status": "due", "company": "Hooli", "rate": "89.1%", "industry": "REAL ESTATE"}, {"currency": "DKK  ", "job_title": "Acct  Exec", "status": "Overdue", "company": "  Hooli", "rate": "79.5%", "industry": "Industrial"}, {"currency": "BIF", "job_title": "SALES REPRESENTATIVE", "status": "Paid", "company": "Soylent Corp", "rate": "73.4%", "industry": "Property"}, {"currency": "Yuan Renminbi", "job_title": "V.P.", "status": "pending", "company": " Initech ", "rate": "12.6%", "industry": "real esstate"}, {"currency": "Yuan Renminbi", "job_title": " ae ", "status": "Pending", "company": "Acme  Inc", "rate": "19.1%", "industry": "Industrial"}, {"currency": "yuan renminbi", "job_title": "acct bxec", "status": " Paid ", "company": "Vehement  ", "rate": "92.3%", "industry": "Industrial "}, {"currency": "Yuan  Renminbi", "job_title": "account executive", "status": "refunded", "company": "Umbrella", "rate": "99.3%", "industry": "Real Estate"}, {"currency": "yuan renminbi", "job_title": "Rep", "status": "  Late", "company": "Stark Industries", "rate": "4.5%", "industry": "Idustrial"}, {"currency": "CNY", "job_title": "Vice President", "status": "unpaid", "company": "Initech", "rate": "52.1%", "industry": "MFG"}], "clean": [{"currency": "CNY", "job_title": "Vice President", "status": "Pending", "company": "Initech", "rate": 0.521, "industry": "Manufacturing"}, {"currency": "CNY", "job_title": "Sales Representative", "status": "Refunded", "company": "Acme Inc", "rate": 0.149, "industry": "Manufacturing"}, {"currency": "CNY", "job_title": "Sales Representative", "status": "Paid", "company": "Vehement", "rate": 0.6559999999999999, "industry": "Real Estate"}, {"currency": "CNY", "job_title": "Account Executive", "status": "Paid", "company": "Umbrella", "rate": 0.302, "industry": "Real Estate"}, {"currency": "BIF", "job_title": "Account Executive", "status": "Overdue", "company": "Acme Inc", "rate": 0.311, "industry": "Real Estate"}, {"currency": "BIF", "job_title": "Account Executive", "status": "Overdue", "company": "Stark Industries", "rate": 0.8420000000000001, "industry": "Real Estate"}, {"currency": "DKK", "job_title": "Sales Representative", "status": "Paid", "company": "Vehement", "rate": 0.044000000000000004, "industry": "Real Estate"}, {"currency": "BIF", "job_title": "Account Executive", "status": "Refunded", "company": "Cyberdyne", "rate": 0.40399999999999997, "industry": "Real Estate"}, {"currency": "BIF", "job_title": "Sales Representative", "status": "Paid", "company": "Stark Industries", "rate": 0.004, "industry": "Real Estate"}, {"currency": "BIF", "job_title": "Sales Representative", "status": "Refunded", "company": "Vehement", "rate": 0.063, "industry": "Real Estate"}, {"currency": "BIF", "job_title": "Vice President", "status": "Pending", "company": "Acme Inc", "rate": 0.551, "industry": "Real Estate"}, {"currency": "CNY", "job_title": "Vice President", "status": "Pending", "company": "Soylent Corp", "rate": 0.071, "industry": "Manufacturing"}, {"currency": "BIF", "job_title": "Account Executive", "status": "Pending", "company": "Soylent Corp", "rate": 0.365, "industry": "Manufacturing"}, {"currency": "CNY", "job_title": "Sales Representative", "status": "Pending", "company": "Stark Industries", "rate": 0.444, "industry": "Manufacturing"}, {"currency": "DKK", "job_title": "Vice President", "status": "Pending", "company": "Hooli", "rate": 0.8909999999999999, "industry": "Real Estate"}, {"currency": "DKK", "job_title": "Account Executive", "status": "Overdue", "company": "Hooli", "rate": 0.795, "industry": "Manufacturing"}, {"currency": "BIF", "job_title": "Sales Representative", "status": "Paid", "company": "Soylent Corp", "rate": 0.7340000000000001, "industry": "Real Estate"}, {"currency": "CNY", "job_title": "Vice President", "status": "Pending", "company": "Initech", "rate": 0.126, "industry": "Real Estate"}, {"currency": "CNY", "job_title": "Account Executive", "status": "Pending", "company": "Acme Inc", "rate": 0.191, "industry": "Manufacturing"}, {"currency": "CNY", "job_title": "Account Executive", "status": "Paid", "company": "Vehement", "rate": 0.9229999999999999, "industry": "Manufacturing"}, {"currency": "CNY", "job_title": "Account Executive", "status": "Refunded", "company": "Umbrella", "rate": 0.993, "industry": "Real Estate"}, {"currency": "CNY", "job_title": "Sales Representative", "status": "Overdue", "company": "Stark Industries", "rate": 0.045, "industry": "Manufacturing"}], "dirty_cols": ["currency", "job_title", "status", "company", "rate", "industry"], "clean_cols": ["currency", "job_title", "status", "company", "rate", "industry"], "plan": {"dataset_summary": "23 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Yuan Renminbi": "CNY", "cny": "CNY", "yuan renminbi": "CNY", "bif": "BIF", "Bif": "BIF", "Burundi Franc": "BIF", "Cny": "CNY", "dkk": "DKK"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Rep": "Sales Representative", "Acct Exec": "Account Executive", "Ae": "Account Executive", "rep": "Sales Representative", "ae": "Account Executive", "sales representative": "Sales Representative", "VICE PRES": "Vice President", "SALES REPRESENTATIVE": "Sales Representative", "V.P.": "Vice President", "acct bxec": "Account Executive", "account executive": "Account Executive"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"unpaid": "Pending", "paid": "Paid", "OVERDUE": "Overdue", "refunded": "Refunded", "PENDING": "Pending", "pendnig": "Pending", "pending": "Pending", "due": "Pending", "Late": "Overdue"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MFG": "Manufacturing", "Industrial": "Manufacturing", "real estate": "Real Estate", "realestate": "Real Estate", "PROPERTY": "Real Estate", "property": "Real Estate", "RealEstate": "Real Estate", "Poptech": "Real Estate", "PropTech": "Real Estate", "Real EEstate": "Real Estate", "Industriaal": "Manufacturing", "MANUFACTURING": "Manufacturing", "REAL ESTATE": "Real Estate", "Property": "Real Estate", "real esstate": "Real Estate", "Idustrial": "Manufacturing"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "hawaii", "name": "Bob Lee", "country": "BF", "city": "Valdebernardo  ", "currency": "guinean franc"}, {"state": "hwaii", "name": "  Frank Adams", "country": "Jamaica", "city": "Putte", "currency": "GNF"}, {"state": "Virgin Islands, U.S.", "name": "Mona Petrov", "country": "JAM", "city": " Putte ", "currency": "costa rican colon"}, {"state": "Virgin Islands, U.S.", "name": "Alice Moore  ", "country": "Jm", "city": "Valdebernardo", "currency": "Lempira"}, {"state": "hawaii", "name": "Alice Diaz", "country": "moz", "city": "Putte", "currency": "hnl"}, {"state": "hi   ", "name": "Judy Ali", "country": "Burkina Faso", "city": "Valdebernardo", "currency": "GMD"}, {"state": "Virgin Islands, U.S.", "name": "Omar Moore", "country": "Burkina Faso", "city": "Valdebernardo", "currency": "GMD"}, {"state": "hawaii", "name": "Alice Novak", "country": "Jamaica", "city": "Valdebernardo", "currency": "GMD"}, {"state": "Virgin Islands, U.S.", "name": "Ivan Ali  ", "country": "Jam", "city": " Delft ", "currency": "COSTA RICAN COLON"}, {"state": "   hawaii", "name": "Eve Novak  ", "country": "jamaicca", "city": "   Putte", "currency": "CRC"}, {"state": "Hi", "name": "Judy Novak   ", "country": "JAM", "city": "  PUTTE", "currency": "GNF"}, {"state": "Virgin Islands, U.S.", "name": "Omar Moore", "country": "Burkina Faso", "city": "Valdebernardo", "currency": "GMD"}, {"state": "", "name": "", "country": "", "city": "", "currency": ""}], "clean": [{"state": "Hawaii", "name": "Bob Lee", "country": "Burkina Faso", "city": "Valdebernardo", "currency": "GNF"}, {"state": "Hawaii", "name": "Frank Adams", "country": "Jamaica", "city": "Putte", "currency": "GNF"}, {"state": "Virgin Islands, U.S.", "name": "Mona Petrov", "country": "Jamaica", "city": "Putte", "currency": "CRC"}, {"state": "Virgin Islands, U.S.", "name": "Alice Moore", "country": "Jamaica", "city": "Valdebernardo", "currency": "HNL"}, {"state": "Hawaii", "name": "Alice Diaz", "country": "Mozambique", "city": "Putte", "currency": "HNL"}, {"state": "Hawaii", "name": "Judy Ali", "country": "Burkina Faso", "city": "Valdebernardo", "currency": "GMD"}, {"state": "Virgin Islands, U.S.", "name": "Omar Moore", "country": "Burkina Faso", "city": "Valdebernardo", "currency": "GMD"}, {"state": "Hawaii", "name": "Alice Novak", "country": "Jamaica", "city": "Valdebernardo", "currency": "GMD"}, {"state": "Virgin Islands, U.S.", "name": "Ivan Ali", "country": "Jamaica", "city": "Delft", "currency": "CRC"}, {"state": "Hawaii", "name": "Eve Novak", "country": "Jamaica", "city": "Putte", "currency": "CRC"}, {"state": "Hawaii", "name": "Judy Novak", "country": "Jamaica", "city": "Putte", "currency": "GNF"}], "dirty_cols": ["state", "name", "country", "city", "currency"], "clean_cols": ["state", "name", "country", "city", "currency"], "plan": {"dataset_summary": "13 rows × 5 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"hawaii": "Hawaii", "hwaii": "Hawaii", "hi": "Hawaii", "Hi": "Hawaii"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"BF": "Burkina Faso", "JAM": "Jamaica", "Jm": "Jamaica", "moz": "Mozambique", "Jam": "Jamaica", "jamaicca": "Jamaica"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PUTTE": "Putte"}, "rationale": "Unified 1 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"guinean franc": "GNF", "costa rican colon": "CRC", "Lempira": "HNL", "hnl": "HNL", "COSTA RICAN COLON": "CRC"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "Yes", "email": " hjig@example.com ", "phone": "(599)177-7167", "department": "Marketing", "amount": "4.321,46"}, {"is_active": "1", "email": "mfojod@mail.com ", "phone": "5510947429", "department": "marketing", "amount": "$1,015.58"}, {"is_active": "F", "email": "BKMDBI@EXAMPLE.COM", "phone": "5190259368", "department": "Marketing  ", "amount": "$1,057.37"}, {"is_active": "1", "email": "  LHCJ@CORP.IO", "phone": "549-782-4863", "department": "growth", "amount": "0"}, {"is_active": "1", "email": "cnmpjhb@test.org ", "phone": "(543)995-0783", "department": " FIN ", "amount": "517,81"}, {"is_active": "T", "email": "KNKDKB@CORP.IO   ", "phone": "571.430.7894", "department": "Accounting   ", "amount": "$1,852.16"}, {"is_active": "Y", "email": "bndnl@example.com", "phone": "598-817-6371", "department": "Mkt", "amount": "1,355"}, {"is_active": "0", "email": " OMKELAB@MAIL.COM ", "phone": "5316412650", "department": "Accounting", "amount": "$7,502.39"}, {"is_active": "0", "email": "kjadghg@corp.io   ", "phone": "502-846-5310", "department": "ACCOUNTING", "amount": "$649.03"}, {"is_active": "TRUE", "email": "kmfdllj@example.com", "phone": "(579)537-8824", "department": "Marketing", "amount": "7.300,22"}, {"is_active": "N", "email": "pncn@mail.com", "phone": "571.266.7340", "department": "mkt", "amount": "--"}, {"is_active": "No", "email": "OFLMFLI@MAIL.COM", "phone": "502.798.4217", "department": "Accounting", "amount": "null"}, {"is_active": "Y", "email": "bndnl@example.com", "phone": "598-817-6371", "department": "Mkt", "amount": "1,355"}], "clean": [{"is_active": true, "email": "hjig@example.com", "phone": "(599) 177-7167", "department": "Marketing", "amount": 4321.46}, {"is_active": true, "email": "mfojod@mail.com", "phone": "(551) 094-7429", "department": "Marketing", "amount": 1015.58}, {"is_active": false, "email": "bkmdbi@example.com", "phone": "(519) 025-9368", "department": "Marketing", "amount": 1057.37}, {"is_active": true, "email": "lhcj@corp.io", "phone": "(549) 782-4863", "department": "Marketing", "amount": 0.0}, {"is_active": true, "email": "cnmpjhb@test.org", "phone": "(543) 995-0783", "department": "Finance", "amount": 517.81}, {"is_active": true, "email": "knkdkb@corp.io", "phone": "(571) 430-7894", "department": "Finance", "amount": 1852.16}, {"is_active": true, "email": "bndnl@example.com", "phone": "(598) 817-6371", "department": "Marketing", "amount": 1355.0}, {"is_active": false, "email": "omkelab@mail.com", "phone": "(531) 641-2650", "department": "Finance", "amount": 7502.39}, {"is_active": false, "email": "kjadghg@corp.io", "phone": "(502) 846-5310", "department": "Finance", "amount": 649.03}, {"is_active": true, "email": "kmfdllj@example.com", "phone": "(579) 537-8824", "department": "Marketing", "amount": 7300.22}, {"is_active": false, "email": "pncn@mail.com", "phone": "(571) 266-7340", "department": "Marketing", "amount": NaN}, {"is_active": false, "email": "oflmfli@mail.com", "phone": "(502) 798-4217", "department": "Finance", "amount": NaN}], "dirty_cols": ["is_active", "email", "phone", "department", "amount"], "clean_cols": ["is_active", "email", "phone", "department", "amount"], "plan": {"dataset_summary": "13 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"marketing": "Marketing", "growth": "Marketing", "FIN": "Finance", "Accounting": "Finance", "Mkt": "Marketing", "ACCOUNTING": "Finance", "mkt": "Marketing"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"state": "nv", "currency": " BOB ", "department": "it", "name": "Alice Adams", "is_active": "Y", "signup_date": "5 Jun 2023"}, {"state": "AK", "currency": "BOB", "department": "   Information Technology", "name": " Lena Park", "is_active": "N", "signup_date": "7/15/2023"}, {"state": "nsvada", "currency": "  bob", "department": " I..", "name": "Judy  Moore", "is_active": "T", "signup_date": "9/26/2023"}, {"state": "Massachusetts", "currency": "BOLIFIANO  ", "department": "Legal & Comliance", "name": "Carol Smith   ", "is_active": "1", "signup_date": "45103"}, {"state": "AK", "currency": "Aed", "department": "Information Technology", "name": "Eve Wong", "is_active": "FALSE", "signup_date": "2023-04-05"}, {"state": " pr ", "currency": "UAE  DIRHAM", "department": "Legal & Compliance", "name": "Carol Moore", "is_active": "1", "signup_date": "2023-01-21"}, {"state": "ma", "currency": "GYD", "department": " Legal ", "name": "Omar Novak", "is_active": "No", "signup_date": "2023-08-26"}, {"state": "PR", "currency": " DKK ", "department": "Information Technology", "name": "Heidi Wong ", "is_active": "0", "signup_date": "45111"}, {"state": "NV", "currency": " BOB", "department": "IT", "name": "Lena Moore", "is_active": "N", "signup_date": "2023-10-04"}, {"state": "Puerto Rico", "currency": "guyana dollar", "department": "legal", "name": "Ivan  Adams", "is_active": "T", "signup_date": "8/24/2023"}, {"state": "   Alaska", "currency": "Gyd", "department": "It", "name": "Grace Ali  ", "is_active": "F", "signup_date": "27 Feb 2023"}, {"state": "ak", "currency": "UAAE Dirham", "department": "Legal", "name": " Omar Park", "is_active": "0", "signup_date": "2023-01-15"}, {"state": "Vermont ", "currency": " dkk ", "department": "Legal & Compliznce", "name": "Eve Reyes", "is_active": "Y", "signup_date": "4/10/2023"}, {"state": "pr", "currency": "BOB", "department": "Legal", "name": "Carol Brandt", "is_active": "No", "signup_date": "2023-01-16"}, {"state": "ma", "currency": "  AED", "department": "informatino technology", "name": "Sara Diaz  ", "is_active": "FALSE", "signup_date": "21 Jun 2023"}, {"state": "ALASKA", "currency": "DKK   ", "department": "Legal & Compliacne", "name": "Mona Petrov", "is_active": "N", "signup_date": "7/16/2023"}, {"state": "ma", "currency": "AED", "department": "Information Tecnology", "name": "   Carol Reyes", "is_active": "Yes", "signup_date": "2023-02-01"}, {"state": "NV", "currency": " BOB", "department": "IT", "name": "Lena Moore", "is_active": "N", "signup_date": "2023-10-04"}, {"state": "AK", "currency": "BOB", "department": "   Information Technology", "name": " Lena Park", "is_active": "N", "signup_date": "7/15/2023"}, {"state": "", "currency": "", "department": "", "name": "", "is_active": "", "signup_date": ""}], "clean": [{"state": "Nevada", "currency": "BOB", "department": "Information Technology", "name": "Alice Adams", "is_active": true, "signup_date": "2023-06-05"}, {"state": "Alaska", "currency": "BOB", "department": "Information Technology", "name": "Lena Park", "is_active": false, "signup_date": "2023-07-15"}, {"state": "Nevada", "currency": "BOB", "department": "Information Technology", "name": "Judy Moore", "is_active": true, "signup_date": "2023-09-26"}, {"state": "Massachusetts", "currency": "BOB", "department": "Legal", "name": "Carol Smith", "is_active": true, "signup_date": "2023-06-26"}, {"state": "Alaska", "currency": "AED", "department": "Information Technology", "name": "Eve Wong", "is_active": false, "signup_date": "2023-04-05"}, {"state": "Puerto Rico", "currency": "AED", "department": "Legal", "name": "Carol Moore", "is_active": true, "signup_date": "2023-01-21"}, {"state": "Massachusetts", "currency": "GYD", "department": "Legal", "name": "Omar Novak", "is_active": false, "signup_date": "2023-08-26"}, {"state": "Puerto Rico", "currency": "DKK", "department": "Information Technology", "name": "Heidi Wong", "is_active": false, "signup_date": "2023-07-04"}, {"state": "Nevada", "currency": "BOB", "department": "Information Technology", "name": "Lena Moore", "is_active": false, "signup_date": "2023-10-04"}, {"state": "Puerto Rico", "currency": "GYD", "department": "Legal", "name": "Ivan Adams", "is_active": true, "signup_date": "2023-08-24"}, {"state": "Alaska", "currency": "GYD", "department": "Information Technology", "name": "Grace Ali", "is_active": false, "signup_date": "2023-02-27"}, {"state": "Alaska", "currency": "AED", "department": "Legal", "name": "Omar Park", "is_active": false, "signup_date": "2023-01-15"}, {"state": "Vermont", "currency": "DKK", "department": "Legal", "name": "Eve Reyes", "is_active": true, "signup_date": "2023-04-10"}, {"state": "Puerto Rico", "currency": "BOB", "department": "Legal", "name": "Carol Brandt", "is_active": false, "signup_date": "2023-01-16"}, {"state": "Massachusetts", "currency": "AED", "department": "Information Technology", "name": "Sara Diaz", "is_active": false, "signup_date": "2023-06-21"}, {"state": "Alaska", "currency": "DKK", "department": "Legal", "name": "Mona Petrov", "is_active": false, "signup_date": "2023-07-16"}, {"state": "Massachusetts", "currency": "AED", "department": "Information Technology", "name": "Carol Reyes", "is_active": true, "signup_date": "2023-02-01"}], "dirty_cols": ["state", "currency", "department", "name", "is_active", "signup_date"], "clean_cols": ["state", "currency", "department", "name", "is_active", "signup_date"], "plan": {"dataset_summary": "20 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"nv": "Nevada", "AK": "Alaska", "nsvada": "Nevada", "pr": "Puerto Rico", "ma": "Massachusetts", "PR": "Puerto Rico", "NV": "Nevada", "ak": "Alaska", "ALASKA": "Alaska"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bob": "BOB", "BOLIFIANO": "BOB", "Aed": "AED", "UAE DIRHAM": "AED", "guyana dollar": "GYD", "Gyd": "GYD", "UAAE Dirham": "AED", "dkk": "DKK"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"it": "Information Technology", "I..": "Information Technology", "Legal & Comliance": "Legal", "Legal & Compliance": "Legal", "IT": "Information Technology", "legal": "Legal", "It": "Information Technology", "Legal & Compliznce": "Legal", "informatino technology": "Information Technology", "Legal & Compliacne": "Legal", "Information Tecnology": "Information Technology"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"name": " David Diaz", "company": "Hooli", "signup_date": "2023-12-26", "amount": "2.388,23", "country": "Bosnia and Herzegovina", "rate": "96.0%", "email": "jjgapb@example.com"}, {"name": "   David Brandt", "company": " Vehement ", "signup_date": "27 Nov 2023", "amount": "6.110,46", "country": "republic of bosnia and herzegovina", "rate": "68.3%", "email": "IGHGK@EXAMPLE.COM"}, {"name": "Eve Cruz", "company": "  Soylent Corp", "signup_date": "2023-03-09", "amount": "None", "country": "Sl", "rate": "65.1%", "email": "   HMGMGH@CORP.IO"}, {"name": "Judy  Ali", "company": "Globex  ", "signup_date": "2023-12-24", "amount": "$5,339.83", "country": "Togolese Republic", "rate": "52.8%", "email": "lbdfcb@example.com  "}, {"name": "Frank Smith", "company": " Wonka Co", "signup_date": "2023-11-23", "amount": "9999999", "country": "Omn", "rate": "14.2%", "email": "   KACMEG@CORP.IO"}, {"name": "   Ivan Ali", "company": "Initech", "signup_date": "8/1/2023", "amount": "$7,383.87", "country": " togo ", "rate": "25.5%", "email": "PHCKID@CORP.IO"}, {"name": "Sara Novak", "company": "Vehement  ", "signup_date": "2023-12-19", "amount": "1.264,09", "country": "SL", "rate": "88.5%", "email": "dpfkaho@test.org"}, {"name": "   Karl Park", "company": "Vehement", "signup_date": "8/26/2023", "amount": "TBD", "country": "BA", "rate": "16.8%", "email": "NELBKOH@TEST.ORG"}, {"name": "Grace Reyes   ", "company": "Globex", "signup_date": "2023-04-21", "amount": "921,38", "country": "Sultanate of Oman ", "rate": "62.8%", "email": "FKJM@EXAMPLE.COM"}, {"name": "Carol Khan", "company": "Globex", "signup_date": "45166", "amount": "$5,409.32", "country": "  Republic of Sierra Leone", "rate": "57.0%", "email": "hiakgn@test.org"}, {"name": " Ivan Smith", "company": "Vehement", "signup_date": "45113", "amount": "3.472,25", "country": "SL", "rate": "58.3%", "email": " ecejb@mail.com"}, {"name": "David Johnson   ", "company": "   Cyberdyne", "signup_date": "44998", "amount": "4.103,70", "country": "  bih", "rate": "60.8%", "email": "eflp@test.org"}, {"name": "Eve  Adams", "company": " Initech ", "signup_date": "2023-10-15", "amount": "4.595,20", "country": "REPUBLIC OF BOSNIA AND HERZEGOVINA", "rate": "56.2%", "email": "pdhlfho@corp.io"}, {"name": "Eve Smith", "company": "   Initech", "signup_date": "2023-08-14", "amount": "null", "country": "BIH", "rate": "28.4%", "email": "iince@corp.io"}, {"name": "  Eve Reyes", "company": "Vehement", "signup_date": "5 Jul 2023", "amount": "5.460,92", "country": "SL", "rate": "65.8%", "email": "cdhocn@mail.com"}, {"name": "Priya Khan", "company": "   Soylent Corp", "signup_date": "8 Sep 2023", "amount": "None", "country": "bih", "rate": "76.3%", "email": "AIIINBJ@MAIL.COM"}, {"name": "Mona Park", "company": "Acme Inc", "signup_date": "2023-04-22", "amount": "139,22", "country": "  omn", "rate": "86.7%", "email": "APCKGJB@MAIL.COM"}, {"name": "Sara Reyes", "company": " Umbrella ", "signup_date": "12 Dec 2023", "amount": "354,74", "country": "BIH", "rate": "22.6%", "email": "iinb@test.org"}, {"name": "Ivan  Moore", "company": " Globex ", "signup_date": "9 Mar 2023", "amount": "$6,148.89", "country": "TGO", "rate": "61.8%", "email": "iadp@example.com"}, {"name": "Frank Park", "company": "Umbrella", "signup_date": "3/6/2023", "amount": "$3,710.78", "country": "sle", "rate": "70.4%", "email": "KEKM@MAIL.COM"}, {"name": " Frank Cruz", "company": "Wonka Co   ", "signup_date": "2023-08-26", "amount": "None", "country": " TGO ", "rate": "65.1%", "email": " cepgmim@corp.io "}, {"name": "  Sara Adams", "company": " Vehement", "signup_date": "22 Feb 2023", "amount": "$3,996.41", "country": "TG", "rate": "8.4%", "email": "kjnclh@corp.io"}, {"name": "Mona Johnson", "company": "Cyberdyne", "signup_date": "1/8/2023", "amount": "null", "country": "omn", "rate": "42.8%", "email": "ibipk@example.com"}, {"name": "Lena  Ali", "company": "  Cyberdyne", "signup_date": "20 Feb 2023", "amount": "$4,597.03", "country": "SLE", "rate": "50.4%", "email": "gdpnmm@corp.io"}, {"name": "Frank Park", "company": "Acme Inc", "signup_date": "45262", "amount": "8.318,92", "country": "Omna", "rate": "67.9%", "email": "lopcafc@mail.com"}, {"name": "Ivan  Ali", "company": "Soylent Corp", "signup_date": "2023-06-28", "amount": "$1,838.01", "country": " SIERRA LEONE", "rate": "98.6%", "email": " mblfoal@mail.com "}, {"name": "Karl Novak", "company": "  Initech", "signup_date": "6/9/2023", "amount": "8.325,28", "country": "republic of sierra leone", "rate": "56.7%", "email": "GPPLNPG@MAIL.COM"}, {"name": "Sara Smith   ", "company": "Acme  Inc", "signup_date": "25 Jan 2023", "amount": "$840.43", "country": "Bomnia And Herzegovina", "rate": "80.7%", "email": "kdgje@corp.io"}, {"name": "Lena Khan   ", "company": "Cyberdyne", "signup_date": "5/24/2023", "amount": "4.821,04", "country": "Oamn", "rate": "94.9%", "email": "ffkhjm@corp.io"}, {"name": "Heidi Fischer", "company": "Wonka  Co", "signup_date": "44930", "amount": "$6,739.04", "country": "Sultanate of Oman", "rate": "43.0%", "email": "ffpnd@example.com"}, {"name": "   Priya Lee", "company": "  Umbrella", "signup_date": "2023-02-15", "amount": "$3,567.59", "country": "tg", "rate": "41.0%", "email": " EGMDAI@MAIL.COM "}, {"name": " Eve Ali", "company": "Cyberdyne", "signup_date": "10/26/2023", "amount": "#N/A", "country": " TGO ", "rate": "76.8%", "email": "GDBBK@EXAMPLE.COM"}, {"name": "David Wong", "company": " Hooli ", "signup_date": "2023-10-06", "amount": "1.975,08", "country": "Republic of Bosnia and Herzegovina", "rate": "80.1%", "email": "nflppnh@corp.io"}, {"name": "   Omar Petrov", "company": "Stark Industries", "signup_date": "2023-12-16", "amount": "#N/A", "country": "Sierra Leone", "rate": "55.5%", "email": "MKOM@EXAMPLE.COM"}, {"name": "Karl Johnson", "company": "Umbrella", "signup_date": "8 Feb 2023", "amount": "3.808,75", "country": "Sultanate of Oman", "rate": "11.3%", "email": "kkoo@example.com"}, {"name": "Judy Wong", "company": "Wonka Co", "signup_date": "2023-12-24", "amount": "$2,527.40", "country": "SULTTANATE OF  OMAN", "rate": "57.2%", "email": "aajmi@example.com  "}, {"name": "Priya Wong", "company": "Acme Inc   ", "signup_date": "2023-05-11", "amount": "8.005,76", "country": "OMN", "rate": "47.9%", "email": "LLMDG@EXAMPLE.COM"}, {"name": "Priya Fischer", "company": "Vehement ", "signup_date": "2 Jun 2023", "amount": "7.881,39", "country": "tgo", "rate": "87.1%", "email": "migmbkl@test.org"}, {"name": "Karl  Smith", "company": "Acme Inc ", "signup_date": "45206", "amount": "$2,217.02", "country": "togo", "rate": "41.5%", "email": "enma@corp.io  "}, {"name": "Eve Lee", "company": "Stark Industries", "signup_date": "2/4/2023", "amount": "$7,404.41", "country": "togolesebrepublic", "rate": "42.8%", "email": "oenedia@mail.com"}, {"name": "Carol Petrov", "company": "Stark  Industries", "signup_date": "8/26/2023", "amount": "2.513,84", "country": "tgo", "rate": "3.5%", "email": "dgikp@example.com"}, {"name": "Omar  Reyes", "company": "Globex", "signup_date": "6/21/2023", "amount": "$2,182.10", "country": "OM", "rate": "4.6%", "email": " EKEIIHM@EXAMPLE.COM"}, {"name": "Eve Diaz", "company": " Initech ", "signup_date": "45170", "amount": "4.999,09", "country": "Republic  of Sierra Leone", "rate": "3.2%", "email": "MPFOAJP@CORP.IO"}, {"name": "Eve Adams", "company": "Initech", "signup_date": "5/14/2023", "amount": "6.502,65", "country": "sultanate of oman  ", "rate": "31.3%", "email": " dedkbhf@mail.com"}, {"name": "Grace Smith", "company": "Acme Inc", "signup_date": "5/24/2023", "amount": "7.193,54", "country": "Republic Of Bosnia And Herzegovina", "rate": "86.9%", "email": "LHBNIB@TEST.ORG"}, {"name": "Frank Ali", "company": "Globex", "signup_date": "12 Mar 2023", "amount": "None", "country": "Bosnia and Herzegovina", "rate": "34.1%", "email": "   OHAAHGP@CORP.IO"}, {"name": "  Karl Ali", "company": "Initech", "signup_date": "3/28/2023", "amount": "8.165,76", "country": "oman", "rate": "22.6%", "email": " LAEH@EXAMPLE.COM "}, {"name": " Omar Park", "company": "Wonka Co", "signup_date": "2 Jun 2023", "amount": "$6,377.36", "country": "SL", "rate": "47.5%", "email": "DCHJD@TEST.ORG"}, {"name": "Mona Khan", "company": "Initech", "signup_date": "11/22/2023", "amount": "$6,593.79", "country": "TGO", "rate": "39.5%", "email": " PACC@TEST.ORG "}, {"name": "   Carol Park", "company": "Initech", "signup_date": "2023-03-20", "amount": "886,14", "country": "SL", "rate": "44.7%", "email": "ggajjgj@corp.io"}, {"name": "  Ivan Novak", "company": " Globex", "signup_date": "8/13/2023", "amount": "3.504,15", "country": "togo", "rate": "42.0%", "email": "ifnh@example.com"}, {"name": "Alice Wong", "company": "Cyberdyne", "signup_date": "44963", "amount": "5.899,44", "country": "  Sultanate of Oman", "rate": "29.3%", "email": "  peck@example.com"}, {"name": "Mona  Fischer", "company": "Globex", "signup_date": "9/12/2023", "amount": "$5,327.90", "country": "sl", "rate": "25.7%", "email": "bllpfgd@mail.com"}, {"name": "Mona Johnson", "company": "Initech", "signup_date": "6 Jan 2023", "amount": "7.071,07", "country": "repubic of bosnia and herzegovina", "rate": "78.3%", "email": "JOGIGI@MAIL.COM"}, {"name": "Judy Lee", "company": "  Acme Inc", "signup_date": "9/9/2023", "amount": "8.940,42", "country": "BOSNIA AND HERZEGOVINA", "rate": "43.8%", "email": "BPKFCLI@TEST.ORG"}, {"name": "Karl  Smith", "company": "Acme Inc ", "signup_date": "45206", "amount": "$2,217.02", "country": "togo", "rate": "41.5%", "email": "enma@corp.io  "}], "clean": [{"name": "David Diaz", "company": "Hooli", "signup_date": "2023-12-26", "amount": 2388.23, "country": "Bosnia and Herzegovina", "rate": 0.96, "email": "jjgapb@example.com"}, {"name": "David Brandt", "company": "Vehement", "signup_date": "2023-11-27", "amount": 6110.46, "country": "Bosnia and Herzegovina", "rate": 0.6829999999999999, "email": "ighgk@example.com"}, {"name": "Eve Cruz", "company": "Soylent Corp", "signup_date": "2023-03-09", "amount": NaN, "country": "Sierra Leone", "rate": 0.6509999999999999, "email": "hmgmgh@corp.io"}, {"name": "Judy Ali", "company": "Globex", "signup_date": "2023-12-24", "amount": 5339.83, "country": "Togo", "rate": 0.528, "email": "lbdfcb@example.com"}, {"name": "Frank Smith", "company": "Wonka Co", "signup_date": "2023-11-23", "amount": 9999999.0, "country": "Oman", "rate": 0.142, "email": "kacmeg@corp.io"}, {"name": "Ivan Ali", "company": "Initech", "signup_date": "2023-08-01", "amount": 7383.87, "country": "Togo", "rate": 0.255, "email": "phckid@corp.io"}, {"name": "Sara Novak", "company": "Vehement", "signup_date": "2023-12-19", "amount": 1264.09, "country": "Sierra Leone", "rate": 0.885, "email": "dpfkaho@test.org"}, {"name": "Karl Park", "company": "Vehement", "signup_date": "2023-08-26", "amount": NaN, "country": "Bosnia and Herzegovina", "rate": 0.168, "email": "nelbkoh@test.org"}, {"name": "Grace Reyes", "company": "Globex", "signup_date": "2023-04-21", "amount": 921.38, "country": "Oman", "rate": 0.628, "email": "fkjm@example.com"}, {"name": "Carol Khan", "company": "Globex", "signup_date": "2023-08-28", "amount": 5409.32, "country": "Sierra Leone", "rate": 0.57, "email": "hiakgn@test.org"}, {"name": "Ivan Smith", "company": "Vehement", "signup_date": "2023-07-06", "amount": 3472.25, "country": "Sierra Leone", "rate": 0.583, "email": "ecejb@mail.com"}, {"name": "David Johnson", "company": "Cyberdyne", "signup_date": "2023-03-13", "amount": 4103.7, "country": "Bosnia and Herzegovina", "rate": 0.608, "email": "eflp@test.org"}, {"name": "Eve Adams", "company": "Initech", "signup_date": "2023-10-15", "amount": 4595.2, "country": "Bosnia and Herzegovina", "rate": 0.562, "email": "pdhlfho@corp.io"}, {"name": "Eve Smith", "company": "Initech", "signup_date": "2023-08-14", "amount": NaN, "country": "Bosnia and Herzegovina", "rate": 0.284, "email": "iince@corp.io"}, {"name": "Eve Reyes", "company": "Vehement", "signup_date": "2023-07-05", "amount": 5460.92, "country": "Sierra Leone", "rate": 0.6579999999999999, "email": "cdhocn@mail.com"}, {"name": "Priya Khan", "company": "Soylent Corp", "signup_date": "2023-09-08", "amount": NaN, "country": "Bosnia and Herzegovina", "rate": 0.763, "email": "aiiinbj@mail.com"}, {"name": "Mona Park", "company": "Acme Inc", "signup_date": "2023-04-22", "amount": 139.22, "country": "Oman", "rate": 0.867, "email": "apckgjb@mail.com"}, {"name": "Sara Reyes", "company": "Umbrella", "signup_date": "2023-12-12", "amount": 354.74, "country": "Bosnia and Herzegovina", "rate": 0.226, "email": "iinb@test.org"}, {"name": "Ivan Moore", "company": "Globex", "signup_date": "2023-03-09", "amount": 6148.89, "country": "Togo", "rate": 0.618, "email": "iadp@example.com"}, {"name": "Frank Park", "company": "Umbrella", "signup_date": "2023-03-06", "amount": 3710.78, "country": "Sierra Leone", "rate": 0.7040000000000001, "email": "kekm@mail.com"}, {"name": "Frank Cruz", "company": "Wonka Co", "signup_date": "2023-08-26", "amount": NaN, "country": "Togo", "rate": 0.6509999999999999, "email": "cepgmim@corp.io"}, {"name": "Sara Adams", "company": "Vehement", "signup_date": "2023-02-22", "amount": 3996.41, "country": "Togo", "rate": 0.084, "email": "kjnclh@corp.io"}, {"name": "Mona Johnson", "company": "Cyberdyne", "signup_date": "2023-01-08", "amount": NaN, "country": "Oman", "rate": 0.428, "email": "ibipk@example.com"}, {"name": "Lena Ali", "company": "Cyberdyne", "signup_date": "2023-02-20", "amount": 4597.03, "country": "Sierra Leone", "rate": 0.504, "email": "gdpnmm@corp.io"}, {"name": "Frank Park", "company": "Acme Inc", "signup_date": "2023-12-02", "amount": 8318.92, "country": "Oman", "rate": 0.679, "email": "lopcafc@mail.com"}, {"name": "Ivan Ali", "company": "Soylent Corp", "signup_date": "2023-06-28", "amount": 1838.01, "country": "Sierra Leone", "rate": 0.986, "email": "mblfoal@mail.com"}, {"name": "Karl Novak", "company": "Initech", "signup_date": "2023-06-09", "amount": 8325.28, "country": "Sierra Leone", "rate": 0.5670000000000001, "email": "gpplnpg@mail.com"}, {"name": "Sara Smith", "company": "Acme Inc", "signup_date": "2023-01-25", "amount": 840.43, "country": "Bosnia and Herzegovina", "rate": 0.807, "email": "kdgje@corp.io"}, {"name": "Lena Khan", "company": "Cyberdyne", "signup_date": "2023-05-24", "amount": 4821.04, "country": "Oman", "rate": 0.9490000000000001, "email": "ffkhjm@corp.io"}, {"name": "Heidi Fischer", "company": "Wonka Co", "signup_date": "2023-01-04", "amount": 6739.04, "country": "Oman", "rate": 0.43, "email": "ffpnd@example.com"}, {"name": "Priya Lee", "company": "Umbrella", "signup_date": "2023-02-15", "amount": 3567.59, "country": "Togo", "rate": 0.41, "email": "egmdai@mail.com"}, {"name": "Eve Ali", "company": "Cyberdyne", "signup_date": "2023-10-26", "amount": NaN, "country": "Togo", "rate": 0.768, "email": "gdbbk@example.com"}, {"name": "David Wong", "company": "Hooli", "signup_date": "2023-10-06", "amount": 1975.08, "country": "Bosnia and Herzegovina", "rate": 0.8009999999999999, "email": "nflppnh@corp.io"}, {"name": "Omar Petrov", "company": "Stark Industries", "signup_date": "2023-12-16", "amount": NaN, "country": "Sierra Leone", "rate": 0.555, "email": "mkom@example.com"}, {"name": "Karl Johnson", "company": "Umbrella", "signup_date": "2023-02-08", "amount": 3808.75, "country": "Oman", "rate": 0.113, "email": "kkoo@example.com"}, {"name": "Judy Wong", "company": "Wonka Co", "signup_date": "2023-12-24", "amount": 2527.4, "country": "Oman", "rate": 0.5720000000000001, "email": "aajmi@example.com"}, {"name": "Priya Wong", "company": "Acme Inc", "signup_date": "2023-05-11", "amount": 8005.76, "country": "Oman", "rate": 0.479, "email": "llmdg@example.com"}, {"name": "Priya Fischer", "company": "Vehement", "signup_date": "2023-06-02", "amount": 7881.39, "country": "Togo", "rate": 0.871, "email": "migmbkl@test.org"}, {"name": "Karl Smith", "company": "Acme Inc", "signup_date": "2023-10-07", "amount": 2217.02, "country": "Togo", "rate": 0.415, "email": "enma@corp.io"}, {"name": "Eve Lee", "company": "Stark Industries", "signup_date": "2023-02-04", "amount": 7404.41, "country": "Togo", "rate": 0.428, "email": "oenedia@mail.com"}, {"name": "Carol Petrov", "company": "Stark Industries", "signup_date": "2023-08-26", "amount": 2513.84, "country": "Togo", "rate": 0.035, "email": "dgikp@example.com"}, {"name": "Omar Reyes", "company": "Globex", "signup_date": "2023-06-21", "amount": 2182.1, "country": "Oman", "rate": 0.046, "email": "ekeiihm@example.com"}, {"name": "Eve Diaz", "company": "Initech", "signup_date": "2023-09-01", "amount": 4999.09, "country": "Sierra Leone", "rate": 0.032, "email": "mpfoajp@corp.io"}, {"name": "Eve Adams", "company": "Initech", "signup_date": "2023-05-14", "amount": 6502.65, "country": "Oman", "rate": 0.313, "email": "dedkbhf@mail.com"}, {"name": "Grace Smith", "company": "Acme Inc", "signup_date": "2023-05-24", "amount": 7193.54, "country": "Bosnia and Herzegovina", "rate": 0.8690000000000001, "email": "lhbnib@test.org"}, {"name": "Frank Ali", "company": "Globex", "signup_date": "2023-03-12", "amount": NaN, "country": "Bosnia and Herzegovina", "rate": 0.341, "email": "ohaahgp@corp.io"}, {"name": "Karl Ali", "company": "Initech", "signup_date": "2023-03-28", "amount": 8165.76, "country": "Oman", "rate": 0.226, "email": "laeh@example.com"}, {"name": "Omar Park", "company": "Wonka Co", "signup_date": "2023-06-02", "amount": 6377.36, "country": "Sierra Leone", "rate": 0.475, "email": "dchjd@test.org"}, {"name": "Mona Khan", "company": "Initech", "signup_date": "2023-11-22", "amount": 6593.79, "country": "Togo", "rate": 0.395, "email": "pacc@test.org"}, {"name": "Carol Park", "company": "Initech", "signup_date": "2023-03-20", "amount": 886.14, "country": "Sierra Leone", "rate": 0.447, "email": "ggajjgj@corp.io"}, {"name": "Ivan Novak", "company": "Globex", "signup_date": "2023-08-13", "amount": 3504.15, "country": "Togo", "rate": 0.42, "email": "ifnh@example.com"}, {"name": "Alice Wong", "company": "Cyberdyne", "signup_date": "2023-02-06", "amount": 5899.44, "country": "Oman", "rate": 0.293, "email": "peck@example.com"}, {"name": "Mona Fischer", "company": "Globex", "signup_date": "2023-09-12", "amount": 5327.9, "country": "Sierra Leone", "rate": 0.257, "email": "bllpfgd@mail.com"}, {"name": "Mona Johnson", "company": "Initech", "signup_date": "2023-01-06", "amount": 7071.07, "country": "Bosnia and Herzegovina", "rate": 0.7829999999999999, "email": "jogigi@mail.com"}, {"name": "Judy Lee", "company": "Acme Inc", "signup_date": "2023-09-09", "amount": 8940.42, "country": "Bosnia and Herzegovina", "rate": 0.43799999999999994, "email": "bpkfcli@test.org"}], "dirty_cols": ["name", "company", "signup_date", "amount", "country", "rate", "email"], "clean_cols": ["name", "company", "signup_date", "amount", "country", "rate", "email"], "plan": {"dataset_summary": "56 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"republic of bosnia and herzegovina": "Bosnia and Herzegovina", "Sl": "Sierra Leone", "Togolese Republic": "Togo", "Omn": "Oman", "togo": "Togo", "SL": "Sierra Leone", "BA": "Bosnia and Herzegovina", "Sultanate of Oman": "Oman", "Republic of Sierra Leone": "Sierra Leone", "bih": "Bosnia and Herzegovina", "REPUBLIC OF BOSNIA AND HERZEGOVINA": "Bosnia and Herzegovina", "BIH": "Bosnia and Herzegovina", "omn": "Oman", "TGO": "Togo", "sle": "Sierra Leone", "TG": "Togo", "SLE": "Sierra Leone", "Omna": "Oman", "SIERRA LEONE": "Sierra Leone", "republic of sierra leone": "Sierra Leone", "Bomnia And Herzegovina": "Bosnia and Herzegovina", "Oamn": "Oman", "tg": "Togo", "Republic of Bosnia and Herzegovina": "Bosnia and Herzegovina", "SULTTANATE OF OMAN": "Oman", "OMN": "Oman", "tgo": "Togo", "togolesebrepublic": "Togo", "OM": "Oman", "sultanate of oman": "Oman", "Republic Of Bosnia And Herzegovina": "Bosnia and Herzegovina", "oman": "Oman", "sl": "Sierra Leone", "repubic of bosnia and herzegovina": "Bosnia and Herzegovina", "BOSNIA AND HERZEGOVINA": "Bosnia and Herzegovina"}, "rationale": "Unified 35 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"company": "Vehement", "job_title": "CEO", "amount": "--", "rate": "41.3%", "status": "h", "country": "swz", "notes2": ""}, {"company": "Acme Inc", "job_title": "ceo", "amount": "5.793,09", "rate": "2.6%", "status": "M", "country": "SWZ", "notes2": ""}, {"company": "Soylent Corp", "job_title": "c..o.", "amount": "$4,749.32", "rate": "78.7%", "status": "Medium", "country": " SG ", "notes2": ""}, {"company": "Acme Inc", "job_title": "chief executive officer", "amount": "1.525,74", "rate": "18.9%", "status": "medium", "country": "Republic of Singapore", "notes2": ""}, {"company": "Cyberdyne  ", "job_title": "chief executive officer  ", "amount": "$626.44", "rate": "92.8%", "status": "P2", "country": "SWZ", "notes2": ""}, {"company": "Wonka Co", "job_title": "ADMI ASSISTANT", "amount": "$4,571.80", "rate": "7.7%", "status": " Medium", "country": " SZ ", "notes2": ""}, {"company": "Stark Industries", "job_title": "chief executive officer ", "amount": "$2,944.21", "rate": "36.4%", "status": "HMGH", "country": "sg", "notes2": ""}, {"company": "Globex  ", "job_title": " ceo ", "amount": "655,23", "rate": "37.0%", "status": "   high", "country": "Eswatini ", "notes2": ""}, {"company": "Vehement   ", "job_title": "ceo", "amount": "$7,357.16", "rate": "28.1%", "status": "high", "country": "SG", "notes2": ""}, {"company": "   Vehement", "job_title": "Administrative Assistant", "amount": "$284.12", "rate": "91.0%", "status": "High", "country": "SGP", "notes2": ""}, {"company": "Initech ", "job_title": "Chief Executive Officer", "amount": "$5,382.61", "rate": "31.4%", "status": "H", "country": "ESWATINI", "notes2": ""}, {"company": "Vehement", "job_title": "ADMIN", "amount": "$1,561.33", "rate": "60.4%", "status": "high", "country": "kingdom of eswatini", "notes2": ""}, {"company": "   Wonka Co", "job_title": "ADMINISTRATIVE AASSISTANT", "amount": "6.512,90", "rate": "13.0%", "status": "  M", "country": "SWZ", "notes2": ""}, {"company": " Globex ", "job_title": "ADMIN ASSISTANT", "amount": "$7,692.87", "rate": "9.2%", "status": "Meedium", "country": "swz", "notes2": ""}], "clean": [{"company": "Vehement", "job_title": "Chief Executive Officer", "amount": NaN, "rate": 0.413, "status": "High", "country": "Eswatini"}, {"company": "Acme Inc", "job_title": "Chief Executive Officer", "amount": 5793.09, "rate": 0.026000000000000002, "status": "Medium", "country": "Eswatini"}, {"company": "Soylent Corp", "job_title": "Chief Executive Officer", "amount": 4749.32, "rate": 0.787, "status": "Medium", "country": "Singapore"}, {"company": "Acme Inc", "job_title": "Chief Executive Officer", "amount": 1525.74, "rate": 0.18899999999999997, "status": "Medium", "country": "Singapore"}, {"company": "Cyberdyne", "job_title": "Chief Executive Officer", "amount": 626.44, "rate": 0.9279999999999999, "status": "Medium", "country": "Eswatini"}, {"company": "Wonka Co", "job_title": "Administrative Assistant", "amount": 4571.8, "rate": 0.077, "status": "Medium", "country": "Eswatini"}, {"company": "Stark Industries", "job_title": "Chief Executive Officer", "amount": 2944.21, "rate": 0.364, "status": "High", "country": "Singapore"}, {"company": "Globex", "job_title": "Chief Executive Officer", "amount": 655.23, "rate": 0.37, "status": "High", "country": "Eswatini"}, {"company": "Vehement", "job_title": "Chief Executive Officer", "amount": 7357.16, "rate": 0.281, "status": "High", "country": "Singapore"}, {"company": "Vehement", "job_title": "Administrative Assistant", "amount": 284.12, "rate": 0.91, "status": "High", "country": "Singapore"}, {"company": "Initech", "job_title": "Chief Executive Officer", "amount": 5382.61, "rate": 0.314, "status": "High", "country": "Eswatini"}, {"company": "Vehement", "job_title": "Administrative Assistant", "amount": 1561.33, "rate": 0.604, "status": "High", "country": "Eswatini"}, {"company": "Wonka Co", "job_title": "Administrative Assistant", "amount": 6512.9, "rate": 0.13, "status": "Medium", "country": "Eswatini"}, {"company": "Globex", "job_title": "Administrative Assistant", "amount": 7692.87, "rate": 0.092, "status": "Medium", "country": "Eswatini"}], "dirty_cols": ["company", "job_title", "amount", "rate", "status", "country", "notes2"], "clean_cols": ["company", "job_title", "amount", "rate", "status", "country"], "plan": {"dataset_summary": "14 rows × 7 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CEO": "Chief Executive Officer", "ceo": "Chief Executive Officer", "c..o.": "Chief Executive Officer", "chief executive officer": "Chief Executive Officer", "ADMI ASSISTANT": "Administrative Assistant", "ADMIN": "Administrative Assistant", "ADMINISTRATIVE AASSISTANT": "Administrative Assistant", "ADMIN ASSISTANT": "Administrative Assistant"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"h": "High", "M": "Medium", "medium": "Medium", "P2": "Medium", "HMGH": "High", "high": "High", "H": "High", "Meedium": "Medium"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"swz": "Eswatini", "SWZ": "Eswatini", "SG": "Singapore", "Republic of Singapore": "Singapore", "SZ": "Eswatini", "sg": "Singapore", "SGP": "Singapore", "ESWATINI": "Eswatini", "kingdom of eswatini": "Eswatini"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "577-719-5596", "currency": "colombian peso", "company": "Hooli"}, {"phone": "(541)672-1073", "currency": "   cop", "company": "Soylent Corp   "}, {"phone": "556-834-2944", "currency": "brazilian  real", "company": "Stark Industries  "}, {"phone": "595.898.5267", "currency": "Bahraini  Dinar", "company": "Cyberdyne"}, {"phone": "509.294.6699", "currency": "COP", "company": "   Stark Industries"}, {"phone": "5647847671", "currency": "Bahraini Dinar", "company": "Globex "}, {"phone": "502-355-5857", "currency": "BHD", "company": "Soylent Corp"}, {"phone": "541-983-9943", "currency": "BHD   ", "company": "Hooli"}, {"phone": "577-061-8076", "currency": "BRL", "company": " Umbrella "}, {"phone": "515.033.3246", "currency": "Colombian Peso", "company": " Umbrella "}, {"phone": "(522)118-2786", "currency": "Coxombian Peso", "company": " Globex"}, {"phone": "552.396.4323", "currency": "Bahraini Dinar", "company": "Cyberdyne"}, {"phone": "563-331-4067", "currency": "Cop", "company": "Umbrella"}, {"phone": "5471802830", "currency": "BHD", "company": "Wonka Co"}, {"phone": "5805357876", "currency": "brazilian real", "company": " Umbrella "}, {"phone": "5812633744", "currency": "Bahraini Dinar", "company": " Vehement "}, {"phone": "516.187.5239", "currency": "BRL", "company": "Hooli"}, {"phone": "543.004.0688", "currency": " Bhd", "company": "Stark  Industries"}, {"phone": "(586)573-5490", "currency": "BHD", "company": "Vehement"}, {"phone": "519.574.0026", "currency": " Cop ", "company": "  Globex"}, {"phone": "574-682-2110", "currency": "Bhd", "company": "Stark Industries   "}, {"phone": "547.359.0730", "currency": "Brazilian Real ", "company": " Initech"}, {"phone": "(523)643-6803", "currency": " BRL ", "company": "Vehement"}, {"phone": "577-061-8076", "currency": "BRL", "company": " Umbrella "}], "clean": [{"phone": "(577) 719-5596", "currency": "COP", "company": "Hooli"}, {"phone": "(541) 672-1073", "currency": "COP", "company": "Soylent Corp"}, {"phone": "(556) 834-2944", "currency": "BRL", "company": "Stark Industries"}, {"phone": "(595) 898-5267", "currency": "BHD", "company": "Cyberdyne"}, {"phone": "(509) 294-6699", "currency": "COP", "company": "Stark Industries"}, {"phone": "(564) 784-7671", "currency": "BHD", "company": "Globex"}, {"phone": "(502) 355-5857", "currency": "BHD", "company": "Soylent Corp"}, {"phone": "(541) 983-9943", "currency": "BHD", "company": "Hooli"}, {"phone": "(577) 061-8076", "currency": "BRL", "company": "Umbrella"}, {"phone": "(515) 033-3246", "currency": "COP", "company": "Umbrella"}, {"phone": "(522) 118-2786", "currency": "COP", "company": "Globex"}, {"phone": "(552) 396-4323", "currency": "BHD", "company": "Cyberdyne"}, {"phone": "(563) 331-4067", "currency": "COP", "company": "Umbrella"}, {"phone": "(547) 180-2830", "currency": "BHD", "company": "Wonka Co"}, {"phone": "(580) 535-7876", "currency": "BRL", "company": "Umbrella"}, {"phone": "(581) 263-3744", "currency": "BHD", "company": "Vehement"}, {"phone": "(516) 187-5239", "currency": "BRL", "company": "Hooli"}, {"phone": "(543) 004-0688", "currency": "BHD", "company": "Stark Industries"}, {"phone": "(586) 573-5490", "currency": "BHD", "company": "Vehement"}, {"phone": "(519) 574-0026", "currency": "COP", "company": "Globex"}, {"phone": "(574) 682-2110", "currency": "BHD", "company": "Stark Industries"}, {"phone": "(547) 359-0730", "currency": "BRL", "company": "Initech"}, {"phone": "(523) 643-6803", "currency": "BRL", "company": "Vehement"}], "dirty_cols": ["phone", "currency", "company"], "clean_cols": ["phone", "currency", "company"], "plan": {"dataset_summary": "24 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"colombian peso": "COP", "cop": "COP", "brazilian real": "BRL", "Bahraini Dinar": "BHD", "Colombian Peso": "COP", "Coxombian Peso": "COP", "Cop": "COP", "Bhd": "BHD", "Brazilian Real": "BRL"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"country": "Equaturial  Guinea", "currency": "aruban florin", "email": "ANHLE@EXAMPLE.COM", "rate": "24.7%"}, {"country": "CYP", "currency": "ALL", "email": "PPEMC@CORP.IO   ", "rate": "50.1%"}, {"country": "  GNQ", "currency": "Lek", "email": "PEKFGPJ@MAIL.COM", "rate": "15.0%"}, {"country": "   SJ", "currency": "LEK", "email": "   DMAABD@CORP.IO", "rate": "42.8%"}, {"country": "CURAAO", "currency": "ALL", "email": "NPNCABJ@EXAMPLE.COM", "rate": "32.5%"}, {"country": "RWA", "currency": "COP", "email": "IOJNLA@CORP.IO", "rate": "32.6%"}, {"country": "Republic of Equatorial Guinea", "currency": "awg", "email": "JMBHDL@MAIL.COM", "rate": "1.3%"}, {"country": "RWA", "currency": "dzd", "email": "  DFLJLLO@CORP.IO", "rate": "56.6%"}, {"country": "Rwandese Republic  ", "currency": "AWG", "email": "   LMJEE@MAIL.COM", "rate": "32.3%"}, {"country": "Gq", "currency": "Colombian Peso", "email": "HMPH@TEST.ORG", "rate": "92.3%"}, {"country": "republic of cyprus", "currency": "Algerian  Dinar", "email": "mfmpj@test.org", "rate": "75.4%"}, {"country": "RW", "currency": "dzd", "email": " CPIMC@TEST.ORG ", "rate": "84.3%"}, {"country": "Equatorial  Guinea", "currency": "  COP", "email": "iikiaoh@test.org ", "rate": "23.8%"}, {"country": "CURAAO", "currency": "ALL", "email": "NPNCABJ@EXAMPLE.COM", "rate": "32.5%"}, {"country": "RW", "currency": "dzd", "email": " CPIMC@TEST.ORG ", "rate": "84.3%"}], "clean": [{"country": "Equatorial Guinea", "currency": "AWG", "email": "anhle@example.com", "rate": 0.247}, {"country": "Cyprus", "currency": "ALL", "email": "ppemc@corp.io", "rate": 0.501}, {"country": "Equatorial Guinea", "currency": "ALL", "email": "pekfgpj@mail.com", "rate": 0.15}, {"country": "Svalbard and Jan Mayen", "currency": "ALL", "email": "dmaabd@corp.io", "rate": 0.428}, {"country": "Curaçao", "currency": "ALL", "email": "npncabj@example.com", "rate": 0.325}, {"country": "Rwanda", "currency": "COP", "email": "iojnla@corp.io", "rate": 0.326}, {"country": "Equatorial Guinea", "currency": "AWG", "email": "jmbhdl@mail.com", "rate": 0.013000000000000001}, {"country": "Rwanda", "currency": "DZD", "email": "dfljllo@corp.io", "rate": 0.5660000000000001}, {"country": "Rwanda", "currency": "AWG", "email": "lmjee@mail.com", "rate": 0.32299999999999995}, {"country": "Equatorial Guinea", "currency": "COP", "email": "hmph@test.org", "rate": 0.9229999999999999}, {"country": "Cyprus", "currency": "DZD", "email": "mfmpj@test.org", "rate": 0.754}, {"country": "Rwanda", "currency": "DZD", "email": "cpimc@test.org", "rate": 0.843}, {"country": "Equatorial Guinea", "currency": "COP", "email": "iikiaoh@test.org", "rate": 0.23800000000000002}], "dirty_cols": ["country", "currency", "email", "rate"], "clean_cols": ["country", "currency", "email", "rate"], "plan": {"dataset_summary": "15 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Equaturial Guinea": "Equatorial Guinea", "CYP": "Cyprus", "GNQ": "Equatorial Guinea", "SJ": "Svalbard and Jan Mayen", "CURAAO": "Curaçao", "RWA": "Rwanda", "Republic of Equatorial Guinea": "Equatorial Guinea", "Rwandese Republic": "Rwanda", "Gq": "Equatorial Guinea", "republic of cyprus": "Cyprus", "RW": "Rwanda"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"aruban florin": "AWG", "Lek": "ALL", "LEK": "ALL", "awg": "AWG", "dzd": "DZD", "Colombian Peso": "COP", "Algerian Dinar": "DZD"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"amount": "$5,107.94", "state": "south dakota", "city": "crespo", "rate": "71.2%", "unnamed": ""}, {"amount": "2.525,79", "state": " Sd ", "city": "Crrespo ", "rate": "51.2%", "unnamed": ""}, {"amount": "8.542,09", "state": " South Dakota", "city": "Pulivendla", "rate": "8.2%", "unnamed": ""}, {"amount": "N/A", "state": " ME", "city": "Crespo", "rate": "49.4%", "unnamed": ""}, {"amount": "5.921,01", "state": "ME", "city": "Pulivendla", "rate": "12.1%", "unnamed": ""}, {"amount": "$959.89", "state": "sd", "city": "  Cresop", "rate": "75.9%", "unnamed": ""}, {"amount": "$7,205.66", "state": "me", "city": "Pulivendla", "rate": "97.6%", "unnamed": ""}, {"amount": "8.141,23", "state": "south dakota  ", "city": "pulivendla  ", "rate": "82.5%", "unnamed": ""}, {"amount": "TBD", "state": "Maine", "city": "Pulivendla", "rate": "74.4%", "unnamed": ""}, {"amount": "7.480,59", "state": "South Dakota", "city": " Pulivendla", "rate": "63.2%", "unnamed": ""}, {"amount": "8.748,53", "state": "me", "city": "Pulivendla", "rate": "24.5%", "unnamed": ""}, {"amount": "4.044,55", "state": "Maine", "city": "crespo", "rate": "91.9%", "unnamed": ""}, {"amount": "7.204,57", "state": "ME", "city": "Crespo", "rate": "69.8%", "unnamed": ""}, {"amount": "$7,144.50", "state": "South Dakota", "city": "cresao  ", "rate": "80.7%", "unnamed": ""}, {"amount": "0", "state": "South Dakota", "city": " Crespo ", "rate": "11.1%", "unnamed": ""}, {"amount": "2.213,61", "state": "ME", "city": "CRESPO", "rate": "20.6%", "unnamed": ""}], "clean": [{"amount": 5107.94, "state": "South Dakota", "city": "Crespo", "rate": 0.7120000000000001}, {"amount": 2525.79, "state": "South Dakota", "city": "Crespo", "rate": 0.512}, {"amount": 8542.09, "state": "South Dakota", "city": "Pulivendla", "rate": 0.08199999999999999}, {"amount": NaN, "state": "Maine", "city": "Crespo", "rate": 0.494}, {"amount": 5921.01, "state": "Maine", "city": "Pulivendla", "rate": 0.121}, {"amount": 959.89, "state": "South Dakota", "city": "Crespo", "rate": 0.759}, {"amount": 7205.66, "state": "Maine", "city": "Pulivendla", "rate": 0.976}, {"amount": 8141.23, "state": "South Dakota", "city": "Pulivendla", "rate": 0.825}, {"amount": NaN, "state": "Maine", "city": "Pulivendla", "rate": 0.7440000000000001}, {"amount": 7480.59, "state": "South Dakota", "city": "Pulivendla", "rate": 0.632}, {"amount": 8748.53, "state": "Maine", "city": "Pulivendla", "rate": 0.245}, {"amount": 4044.55, "state": "Maine", "city": "Crespo", "rate": 0.919}, {"amount": 7204.57, "state": "Maine", "city": "Crespo", "rate": 0.698}, {"amount": 7144.5, "state": "South Dakota", "city": "Crespo", "rate": 0.807}, {"amount": 0.0, "state": "South Dakota", "city": "Crespo", "rate": 0.111}, {"amount": 2213.61, "state": "Maine", "city": "Crespo", "rate": 0.20600000000000002}], "dirty_cols": ["amount", "state", "city", "rate", "unnamed"], "clean_cols": ["amount", "state", "city", "rate"], "plan": {"dataset_summary": "16 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"south dakota": "South Dakota", "Sd": "South Dakota", "ME": "Maine", "sd": "South Dakota", "me": "Maine"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"crespo": "Crespo", "Crrespo": "Crespo", "Cresop": "Crespo", "pulivendla": "Pulivendla", "cresao": "Crespo", "CRESPO": "Crespo"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"status": "in progress", "company": "Umbrella", "phone": "5225913643", "amount": "$2,440.66", "industry": "Hospitality", "notes2": ""}, {"status": "Won", "company": "   Soylent Corp", "phone": "521.007.2540", "amount": "$8,297.03", "industry": "Hospitality", "notes2": ""}, {"status": " closed-won ", "company": " Initech", "phone": "533.724.8315", "amount": "--", "industry": "HOSPITALITY", "notes2": ""}, {"status": "closed-won", "company": "   Acme Inc", "phone": "553.977.7939", "amount": "8.385,72", "industry": "Trvael", "notes2": ""}, {"status": "ongoing", "company": "Soylent Corp   ", "phone": "5667498952", "amount": "3.630,36", "industry": "ed", "notes2": ""}, {"status": "   LOST", "company": "Umbrella  ", "phone": "5168768084", "amount": "$1,134.85", "industry": "education", "notes2": ""}, {"status": "Closed-Won", "company": " Hooli ", "phone": "(593)048-6585", "amount": "6.156,31", "industry": " Ed ", "notes2": ""}, {"status": "in prgress", "company": " Globex ", "phone": "5206632664", "amount": "7.753,69", "industry": "education", "notes2": ""}, {"status": "ongoing  ", "company": "Wonka Co   ", "phone": "553.927.3018", "amount": "-", "industry": "education", "notes2": ""}, {"status": "won", "company": "Acme  Inc", "phone": "517-494-3996", "amount": "$5,181.09", "industry": "TRAVEL", "notes2": ""}, {"status": "Won", "company": "   Umbrella", "phone": "526-121-2421", "amount": "2.596,61", "industry": "hospitality", "notes2": ""}, {"status": "closed lost ", "company": "Cyberdyne", "phone": "5605307442", "amount": "4.572,08", "industry": "Travel  ", "notes2": ""}, {"status": "Lost", "company": " Vehement ", "phone": "511-755-7146", "amount": "$949.54", "industry": " Travel ", "notes2": ""}, {"status": "ongoing", "company": "Soylent Corp", "phone": "(588)286-1641", "amount": "5.941,21", "industry": "  medical", "notes2": ""}, {"status": "lost", "company": "Wonka Co", "phone": "500.903.3588", "amount": "None", "industry": "EdTech", "notes2": ""}, {"status": "in progress", "company": "Acme Inc ", "phone": "5505859662", "amount": "#N/A", "industry": "MEDICAL", "notes2": ""}, {"status": "WIP", "company": " Globex", "phone": "531.105.6823", "amount": "$3,976.99", "industry": "hospitality ", "notes2": ""}, {"status": "Closed Lost", "company": "Wonka Co", "phone": "513-638-1579", "amount": "2.828,28", "industry": "   EdTecch", "notes2": ""}, {"status": "won", "company": "Globex", "phone": "577.001.2181", "amount": "2.951,90", "industry": "health care", "notes2": ""}, {"status": "ongoing", "company": " Globex ", "phone": "(524)218-0541", "amount": "TBD", "industry": "health care", "notes2": ""}, {"status": " ongoing ", "company": "Hooli   ", "phone": "575-115-5242", "amount": "na", "industry": " EDTECH", "notes2": ""}, {"status": " Closed Won", "company": "Cyberdyne", "phone": "(575)895-8617", "amount": "$6,774.70", "industry": "healh", "notes2": ""}, {"status": "in-progress", "company": "Stark Industries", "phone": "545.424.2422", "amount": "$8,629.66", "industry": "   Healthcare", "notes2": ""}, {"status": "IN-PROGRESS", "company": "Globex", "phone": "521.616.4273", "amount": "-", "industry": "education", "notes2": ""}], "clean": [{"status": "In Progress", "company": "Umbrella", "phone": "(522) 591-3643", "amount": 2440.66, "industry": "Hospitality"}, {"status": "Won", "company": "Soylent Corp", "phone": "(521) 007-2540", "amount": 8297.03, "industry": "Hospitality"}, {"status": "Won", "company": "Initech", "phone": "(533) 724-8315", "amount": NaN, "industry": "Hospitality"}, {"status": "Won", "company": "Acme Inc", "phone": "(553) 977-7939", "amount": 8385.72, "industry": "Hospitality"}, {"status": "In Progress", "company": "Soylent Corp", "phone": "(566) 749-8952", "amount": 3630.36, "industry": "Education"}, {"status": "Lost", "company": "Umbrella", "phone": "(516) 876-8084", "amount": 1134.85, "industry": "Education"}, {"status": "Won", "company": "Hooli", "phone": "(593) 048-6585", "amount": 6156.31, "industry": "Education"}, {"status": "In Progress", "company": "Globex", "phone": "(520) 663-2664", "amount": 7753.69, "industry": "Education"}, {"status": "In Progress", "company": "Wonka Co", "phone": "(553) 927-3018", "amount": NaN, "industry": "Education"}, {"status": "Won", "company": "Acme Inc", "phone": "(517) 494-3996", "amount": 5181.09, "industry": "Hospitality"}, {"status": "Won", "company": "Umbrella", "phone": "(526) 121-2421", "amount": 2596.61, "industry": "Hospitality"}, {"status": "Lost", "company": "Cyberdyne", "phone": "(560) 530-7442", "amount": 4572.08, "industry": "Hospitality"}, {"status": "Lost", "company": "Vehement", "phone": "(511) 755-7146", "amount": 949.54, "industry": "Hospitality"}, {"status": "In Progress", "company": "Soylent Corp", "phone": "(588) 286-1641", "amount": 5941.21, "industry": "Healthcare"}, {"status": "Lost", "company": "Wonka Co", "phone": "(500) 903-3588", "amount": NaN, "industry": "Education"}, {"status": "In Progress", "company": "Acme Inc", "phone": "(550) 585-9662", "amount": NaN, "industry": "Healthcare"}, {"status": "In Progress", "company": "Globex", "phone": "(531) 105-6823", "amount": 3976.99, "industry": "Hospitality"}, {"status": "Lost", "company": "Wonka Co", "phone": "(513) 638-1579", "amount": 2828.28, "industry": "Education"}, {"status": "Won", "company": "Globex", "phone": "(577) 001-2181", "amount": 2951.9, "industry": "Healthcare"}, {"status": "In Progress", "company": "Globex", "phone": "(524) 218-0541", "amount": NaN, "industry": "Healthcare"}, {"status": "In Progress", "company": "Hooli", "phone": "(575) 115-5242", "amount": NaN, "industry": "Education"}, {"status": "Won", "company": "Cyberdyne", "phone": "(575) 895-8617", "amount": 6774.7, "industry": "Healthcare"}, {"status": "In Progress", "company": "Stark Industries", "phone": "(545) 424-2422", "amount": 8629.66, "industry": "Healthcare"}, {"status": "In Progress", "company": "Globex", "phone": "(521) 616-4273", "amount": NaN, "industry": "Education"}], "dirty_cols": ["status", "company", "phone", "amount", "industry", "notes2"], "clean_cols": ["status", "company", "phone", "amount", "industry"], "plan": {"dataset_summary": "24 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"in progress": "In Progress", "closed-won": "Won", "ongoing": "In Progress", "LOST": "Lost", "Closed-Won": "Won", "in prgress": "In Progress", "won": "Won", "closed lost": "Lost", "lost": "Lost", "WIP": "In Progress", "Closed Lost": "Lost", "Closed Won": "Won", "in-progress": "In Progress", "IN-PROGRESS": "In Progress"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HOSPITALITY": "Hospitality", "Trvael": "Hospitality", "ed": "Education", "education": "Education", "Ed": "Education", "TRAVEL": "Hospitality", "hospitality": "Hospitality", "Travel": "Hospitality", "medical": "Healthcare", "EdTech": "Education", "MEDICAL": "Healthcare", "EdTecch": "Education", "health care": "Healthcare", "EDTECH": "Education", "healh": "Healthcare"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "6.388,99", "department": "Suppoort", "is_active": "No", "country": "gtm", "job_title": "PM", "unnamed": ""}, {"amount": "TBD", "department": "Finance", "is_active": "F", "country": "LBR ", "job_title": "PROD MANAGER", "unnamed": ""}, {"amount": "2.595,15", "department": "CS", "is_active": "true", "country": "QA", "job_title": "cto", "unnamed": ""}, {"amount": "$3,060.79", "department": "accounting", "is_active": "Y", "country": "Palau", "job_title": "Dir", "unnamed": ""}, {"amount": "TBD", "department": " Finance ", "is_active": "F", "country": "Liberia", "job_title": "Chief Technology Officer", "unnamed": ""}, {"amount": "-100", "department": "Fin", "is_active": "false", "country": "CF", "job_title": "DIRECTOR", "unnamed": ""}, {"amount": "$3,471.93", "department": "accounting", "is_active": "FALSE", "country": "  lr", "job_title": "Dir", "unnamed": ""}, {"amount": "6.355,48", "department": "customer support  ", "is_active": "N", "country": "REPUBLIC OF GUATEMALA", "job_title": "PM", "unnamed": ""}, {"amount": "961,91", "department": "CUSTOMER SUPPORT", "is_active": "No", "country": "Guatemala", "job_title": "c.t.o.", "unnamed": ""}, {"amount": "TBD", "department": "CS", "is_active": "Y", "country": "  GTM", "job_title": "C.T..O", "unnamed": ""}, {"amount": "4.833,77", "department": "fin", "is_active": "false", "country": "Pw", "job_title": "Prod Manager", "unnamed": ""}, {"amount": "4.465,88", "department": "Accounting", "is_active": "Yes", "country": "republic of guatemala", "job_title": "CTO  ", "unnamed": ""}, {"amount": "$2,787.67", "department": "fin", "is_active": "F", "country": "PLW", "job_title": "Chief Technology Officer", "unnamed": ""}, {"amount": "None", "department": "fin", "is_active": "Y", "country": "Qatar", "job_title": "Chief Technology Officer", "unnamed": ""}, {"amount": "2.754,32", "department": "suppott", "is_active": "No", "country": "PW", "job_title": "Prod Mgr", "unnamed": ""}, {"amount": "$5,489.96", "department": "finance", "is_active": "false", "country": "PW", "job_title": "c.t.o.", "unnamed": ""}, {"amount": "$8,440.09", "department": "Support", "is_active": "true", "country": "Gtm  ", "job_title": "Chief Technology Officer", "unnamed": ""}, {"amount": "$5,040.60", "department": "support", "is_active": "Yes", "country": "qa", "job_title": "director", "unnamed": ""}, {"amount": "8.303,69", "department": "   accounting", "is_active": "N", "country": "LR", "job_title": "product manager", "unnamed": ""}, {"amount": "$8,984.62", "department": "CS", "is_active": "true", "country": "Central African Republic", "job_title": "cto  ", "unnamed": ""}, {"amount": "$2,177.54", "department": "FNIANCE", "is_active": "No", "country": "Republic Of Palau", "job_title": "director", "unnamed": ""}, {"amount": "$1,935.56", "department": "Customer Support   ", "is_active": "1", "country": "CF", "job_title": " Chief Technolog Officer", "unnamed": ""}, {"amount": "$642.86", "department": "cs", "is_active": "0", "country": "Central African Republic", "job_title": "cto", "unnamed": ""}, {"amount": "None", "department": "finance", "is_active": "T", "country": "GUATTEMALA", "job_title": "CTO", "unnamed": ""}, {"amount": "524,38", "department": "Cust Support", "is_active": "No", "country": "LR", "job_title": "   c.t.o.", "unnamed": ""}, {"amount": "3.017,93", "department": "Customer Support", "is_active": "T", "country": "  Central African Republic", "job_title": " CTO", "unnamed": ""}, {"amount": "$7,346.50", "department": "FIN", "is_active": "N", "country": " Guatemala", "job_title": "cto", "unnamed": ""}, {"amount": "$8,064.87", "department": "Fin", "is_active": "FALSE", "country": " QAT ", "job_title": "PM", "unnamed": ""}, {"amount": "5.373,09", "department": "cust suport", "is_active": "true", "country": "   Lr", "job_title": " C.T.O. ", "unnamed": ""}, {"amount": "851,92", "department": "customer support", "is_active": "Y", "country": "   Liberia", "job_title": "Dir.", "unnamed": ""}, {"amount": "null", "department": "Fin", "is_active": "true", "country": "Liberia", "job_title": "PRODUCT MANWGER", "unnamed": ""}, {"amount": "5.182,19", "department": "Fin", "is_active": "FALSE", "country": "REPUBLIC OF PALAU", "job_title": "Dir", "unnamed": ""}, {"amount": "$7,508.48", "department": "Accounting", "is_active": "No", "country": "gtm", "job_title": "cto", "unnamed": ""}, {"amount": "4.732,49", "department": "Support", "is_active": "true", "country": "   GTM", "job_title": "PM", "unnamed": ""}, {"amount": "2.192,29", "department": "FIN   ", "is_active": "Y", "country": "Republic of Liberia", "job_title": "Pm   ", "unnamed": ""}, {"amount": "--", "department": "fin", "is_active": "0", "country": "qataar", "job_title": "prod anager", "unnamed": ""}], "clean": [{"amount": 6388.99, "department": "Customer Support", "is_active": false, "country": "Guatemala", "job_title": "Product Manager"}, {"amount": NaN, "department": "Finance", "is_active": false, "country": "Liberia", "job_title": "Product Manager"}, {"amount": 2595.15, "department": "Customer Support", "is_active": true, "country": "Qatar", "job_title": "Chief Technology Officer"}, {"amount": 3060.79, "department": "Finance", "is_active": true, "country": "Palau", "job_title": "Director"}, {"amount": NaN, "department": "Finance", "is_active": false, "country": "Liberia", "job_title": "Chief Technology Officer"}, {"amount": -100.0, "department": "Finance", "is_active": false, "country": "Central African Republic", "job_title": "Director"}, {"amount": 3471.93, "department": "Finance", "is_active": false, "country": "Liberia", "job_title": "Director"}, {"amount": 6355.48, "department": "Customer Support", "is_active": false, "country": "Guatemala", "job_title": "Product Manager"}, {"amount": 961.91, "department": "Customer Support", "is_active": false, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": NaN, "department": "Customer Support", "is_active": true, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": 4833.77, "department": "Finance", "is_active": false, "country": "Palau", "job_title": "Product Manager"}, {"amount": 4465.88, "department": "Finance", "is_active": true, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": 2787.67, "department": "Finance", "is_active": false, "country": "Palau", "job_title": "Chief Technology Officer"}, {"amount": NaN, "department": "Finance", "is_active": true, "country": "Qatar", "job_title": "Chief Technology Officer"}, {"amount": 2754.32, "department": "Customer Support", "is_active": false, "country": "Palau", "job_title": "Product Manager"}, {"amount": 5489.96, "department": "Finance", "is_active": false, "country": "Palau", "job_title": "Chief Technology Officer"}, {"amount": 8440.09, "department": "Customer Support", "is_active": true, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": 5040.6, "department": "Customer Support", "is_active": true, "country": "Qatar", "job_title": "Director"}, {"amount": 8303.69, "department": "Finance", "is_active": false, "country": "Liberia", "job_title": "Product Manager"}, {"amount": 8984.62, "department": "Customer Support", "is_active": true, "country": "Central African Republic", "job_title": "Chief Technology Officer"}, {"amount": 2177.54, "department": "Finance", "is_active": false, "country": "Palau", "job_title": "Director"}, {"amount": 1935.56, "department": "Customer Support", "is_active": true, "country": "Central African Republic", "job_title": "Chief Technology Officer"}, {"amount": 642.86, "department": "Customer Support", "is_active": false, "country": "Central African Republic", "job_title": "Chief Technology Officer"}, {"amount": NaN, "department": "Finance", "is_active": true, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": 524.38, "department": "Customer Support", "is_active": false, "country": "Liberia", "job_title": "Chief Technology Officer"}, {"amount": 3017.93, "department": "Customer Support", "is_active": true, "country": "Central African Republic", "job_title": "Chief Technology Officer"}, {"amount": 7346.5, "department": "Finance", "is_active": false, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": 8064.87, "department": "Finance", "is_active": false, "country": "Qatar", "job_title": "Product Manager"}, {"amount": 5373.09, "department": "Customer Support", "is_active": true, "country": "Liberia", "job_title": "Chief Technology Officer"}, {"amount": 851.92, "department": "Customer Support", "is_active": true, "country": "Liberia", "job_title": "Director"}, {"amount": NaN, "department": "Finance", "is_active": true, "country": "Liberia", "job_title": "Product Manager"}, {"amount": 5182.19, "department": "Finance", "is_active": false, "country": "Palau", "job_title": "Director"}, {"amount": 7508.48, "department": "Finance", "is_active": false, "country": "Guatemala", "job_title": "Chief Technology Officer"}, {"amount": 4732.49, "department": "Customer Support", "is_active": true, "country": "Guatemala", "job_title": "Product Manager"}, {"amount": 2192.29, "department": "Finance", "is_active": true, "country": "Liberia", "job_title": "Product Manager"}, {"amount": NaN, "department": "Finance", "is_active": false, "country": "Qatar", "job_title": "Product Manager"}], "dirty_cols": ["amount", "department", "is_active", "country", "job_title", "unnamed"], "clean_cols": ["amount", "department", "is_active", "country", "job_title"], "plan": {"dataset_summary": "36 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Suppoort": "Customer Support", "CS": "Customer Support", "accounting": "Finance", "Fin": "Finance", "customer support": "Customer Support", "CUSTOMER SUPPORT": "Customer Support", "fin": "Finance", "Accounting": "Finance", "suppott": "Customer Support", "finance": "Finance", "Support": "Customer Support", "support": "Customer Support", "FNIANCE": "Finance", "cs": "Customer Support", "Cust Support": "Customer Support", "FIN": "Finance", "cust suport": "Customer Support"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"gtm": "Guatemala", "LBR": "Liberia", "QA": "Qatar", "CF": "Central African Republic", "lr": "Liberia", "REPUBLIC OF GUATEMALA": "Guatemala", "GTM": "Guatemala", "Pw": "Palau", "republic of guatemala": "Guatemala", "PLW": "Palau", "PW": "Palau", "Gtm": "Guatemala", "qa": "Qatar", "LR": "Liberia", "Republic Of Palau": "Palau", "GUATTEMALA": "Guatemala", "QAT": "Qatar", "Lr": "Liberia", "REPUBLIC OF PALAU": "Palau", "Republic of Liberia": "Liberia", "qataar": "Qatar"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PM": "Product Manager", "PROD MANAGER": "Product Manager", "cto": "Chief Technology Officer", "Dir": "Director", "DIRECTOR": "Director", "c.t.o.": "Chief Technology Officer", "C.T..O": "Chief Technology Officer", "Prod Manager": "Product Manager", "CTO": "Chief Technology Officer", "Prod Mgr": "Product Manager", "director": "Director", "product manager": "Product Manager", "Chief Technolog Officer": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "Dir.": "Director", "PRODUCT MANWGER": "Product Manager", "Pm": "Product Manager", "prod anager": "Product Manager"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"name": "Sara Wong   ", "company": "Globex", "industry": "comms", "notes2": ""}, {"name": "Lena Adams ", "company": "Stark Industries ", "industry": "RETAIL", "notes2": ""}, {"name": "Grace Reyes", "company": "Acme Inc   ", "industry": "telecommunications", "notes2": ""}, {"name": "Alice Cruz ", "company": "Stark  Industries", "industry": " retail ", "notes2": ""}, {"name": "David Diaz", "company": "Wonka Co", "industry": " PROPTECH ", "notes2": ""}, {"name": "Sara Reyes", "company": "Hooli", "industry": "  REALESTATE", "notes2": ""}, {"name": "Frank Brandt  ", "company": " Globex ", "industry": "Retail", "notes2": ""}, {"name": "Priya  Park", "company": "Vehement", "industry": " Retail & E-commerce", "notes2": ""}, {"name": "Mona Fischer", "company": "Acme Inc ", "industry": "proptech", "notes2": ""}, {"name": "  Grace Brandt", "company": "Stark Industries", "industry": "Telecommunications", "notes2": ""}, {"name": "Judy  Wong", "company": " Globex ", "industry": "E-commerce", "notes2": ""}, {"name": "Frank  Wong", "company": " Cyberdyne ", "industry": "Telecommunications", "notes2": ""}, {"name": "Ivan  Johnson", "company": "Acme Inc", "industry": "TELECOMMUNICATIONS", "notes2": ""}, {"name": "", "company": "", "industry": "", "notes2": ""}], "clean": [{"name": "Sara Wong", "company": "Globex", "industry": "Telecommunications"}, {"name": "Lena Adams", "company": "Stark Industries", "industry": "Retail"}, {"name": "Grace Reyes", "company": "Acme Inc", "industry": "Telecommunications"}, {"name": "Alice Cruz", "company": "Stark Industries", "industry": "Retail"}, {"name": "David Diaz", "company": "Wonka Co", "industry": "Real Estate"}, {"name": "Sara Reyes", "company": "Hooli", "industry": "Real Estate"}, {"name": "Frank Brandt", "company": "Globex", "industry": "Retail"}, {"name": "Priya Park", "company": "Vehement", "industry": "Retail"}, {"name": "Mona Fischer", "company": "Acme Inc", "industry": "Real Estate"}, {"name": "Grace Brandt", "company": "Stark Industries", "industry": "Telecommunications"}, {"name": "Judy Wong", "company": "Globex", "industry": "Retail"}, {"name": "Frank Wong", "company": "Cyberdyne", "industry": "Telecommunications"}, {"name": "Ivan Johnson", "company": "Acme Inc", "industry": "Telecommunications"}], "dirty_cols": ["name", "company", "industry", "notes2"], "clean_cols": ["name", "company", "industry"], "plan": {"dataset_summary": "14 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"comms": "Telecommunications", "RETAIL": "Retail", "telecommunications": "Telecommunications", "retail": "Retail", "PROPTECH": "Real Estate", "REALESTATE": "Real Estate", "Retail & E-commerce": "Retail", "proptech": "Real Estate", "E-commerce": "Retail", "TELECOMMUNICATIONS": "Telecommunications"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "45.0%", "email": "DJDAJ@CORP.IO ", "city": "wiset chaichan", "unit": "km"}, {"rate": "38.7%", "email": "  LIFL@MAIL.COM", "city": "Nguk", "unit": "   Mi"}, {"rate": "33.5%", "email": "LGAB@EXAMPLE.COM", "city": "kampung cheng", "unit": "km"}, {"rate": "55.9%", "email": " AKBH@MAIL.COM ", "city": "NSOATRE", "unit": "  miles"}, {"rate": "69.0%", "email": "ijjicpd@test.org ", "city": "wiset chaichan", "unit": "Kilometers"}, {"rate": "18.9%", "email": "MGLIPK@TEST.ORG", "city": "Nsoatre", "unit": "mi"}, {"rate": "78.5%", "email": "ALFIMI@MAIL.COM", "city": "Nuuk", "unit": "miles"}, {"rate": "49.2%", "email": "afmb@mail.com  ", "city": "Nuuk", "unit": "km   "}, {"rate": "52.9%", "email": "gnofnmg@corp.io", "city": "wiset chaichan", "unit": "kilometers"}, {"rate": "17.1%", "email": "BOCO@CORP.IO ", "city": "Wiset Chaichan", "unit": "lier   "}, {"rate": "34.9%", "email": "JJIIDO@EXAMPLE.COM", "city": "Nsoatre", "unit": "Milys"}, {"rate": "", "email": "", "city": "", "unit": ""}], "clean": [{"rate": 0.45, "email": "djdaj@corp.io", "city": "Wiset Chaichan", "unit": "km"}, {"rate": 0.387, "email": "lifl@mail.com", "city": "Nuuk", "unit": "mi"}, {"rate": 0.335, "email": "lgab@example.com", "city": "Kampung Cheng", "unit": "km"}, {"rate": 0.5589999999999999, "email": "akbh@mail.com", "city": "Nsoatre", "unit": "mi"}, {"rate": 0.69, "email": "ijjicpd@test.org", "city": "Wiset Chaichan", "unit": "km"}, {"rate": 0.18899999999999997, "email": "mglipk@test.org", "city": "Nsoatre", "unit": "mi"}, {"rate": 0.785, "email": "alfimi@mail.com", "city": "Nuuk", "unit": "mi"}, {"rate": 0.49200000000000005, "email": "afmb@mail.com", "city": "Nuuk", "unit": "km"}, {"rate": 0.529, "email": "gnofnmg@corp.io", "city": "Wiset Chaichan", "unit": "km"}, {"rate": 0.171, "email": "boco@corp.io", "city": "Wiset Chaichan", "unit": "L"}, {"rate": 0.349, "email": "jjiido@example.com", "city": "Nsoatre", "unit": "mi"}], "dirty_cols": ["rate", "email", "city", "unit"], "clean_cols": ["rate", "email", "city", "unit"], "plan": {"dataset_summary": "12 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"wiset chaichan": "Wiset Chaichan", "Nguk": "Nuuk", "kampung cheng": "Kampung Cheng", "NSOATRE": "Nsoatre"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mi": "mi", "miles": "mi", "Kilometers": "km", "kilometers": "km", "lier": "L", "Milys": "mi"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"unit": "km  ", "status": "   new", "state": "Louisiana", "city": "Rossendale", "rate": "97.1%", "name": "Judy Lee"}, {"unit": "KILOMETERS   ", "status": "  new", "state": "New  Mexico", "city": "Highton", "rate": "26.3%", "name": "Lena Diaz"}, {"unit": "l", "status": "won", "state": " ca", "city": "ROSSENDALE", "rate": "62.0%", "name": "   Sara Cruz"}, {"unit": "inch", "status": "closed-won", "state": " SD", "city": "Highton", "rate": "90.3%", "name": "Priya  Reyes"}, {"unit": "Mi", "status": " new ", "state": "California", "city": "highton", "rate": "3.2%", "name": "   Karl Fischer"}, {"unit": "\"", "status": "WON", "state": "South Dakota", "city": "   rossendale", "rate": "66.3%", "name": "Alice Park  "}, {"unit": "   km", "status": "IN PROGRESS", "state": "CA", "city": "highton", "rate": "65.9%", "name": "Heidi Diaz"}, {"unit": "Mi", "status": "ONGOING", "state": "LOUISIANA", "city": "rossendale", "rate": "6.4%", "name": "Mona  Ali"}, {"unit": "miles", "status": "open", "state": "california", "city": "hihton", "rate": "51.1%", "name": "Judy  Johnson"}, {"unit": "kipometers ", "status": "Open", "state": "Louisiana", "city": " Rossendale", "rate": "27.4%", "name": "Carol Fischer"}, {"unit": "LITER", "status": "WIP", "state": "nm", "city": "Rossendale", "rate": "30.5%", "name": " Carol Reyes"}, {"unit": "kilometer", "status": "new", "state": " LA", "city": "Rossendale", "rate": "50.3%", "name": "   Ivan Fischer"}, {"unit": "mile", "status": "Closed Won", "state": "sd", "city": "Rostendale", "rate": "25.5%", "name": "  Carol Park"}, {"unit": " km", "status": "Open", "state": "ca", "city": "highton", "rate": "2.0%", "name": "Mona Moore"}, {"unit": "MILE", "status": " Won", "state": "louisiana", "city": "Highton ", "rate": "41.2%", "name": "Mona  Novak"}, {"unit": "MILE ", "status": "wip", "state": "New Mexico", "city": "highton", "rate": "48.3%", "name": "David  Smith"}, {"unit": "KM", "status": "open", "state": "south dakota", "city": " Highton ", "rate": "26.2%", "name": "Lena Moore"}, {"unit": "l", "status": "OPEN", "state": "Soutth Dakota", "city": "Highton  ", "rate": "55.1%", "name": "Bob Lee"}, {"unit": "L", "status": "closed-won", "state": "Sd", "city": "Rossendale", "rate": "28.0%", "name": " Priya Park"}, {"unit": "MILES", "status": "WIP", "state": "nm", "city": " Rossendale ", "rate": "49.9%", "name": "Heidi Wong"}, {"unit": "  mi", "status": "Oen", "state": "CALIFORNIA", "city": "  Rossendale", "rate": "2.1%", "name": "Sara Novak   "}, {"unit": " mi ", "status": " open ", "state": "SD", "city": "Highotn", "rate": "15.4%", "name": "Lena  Smith"}, {"unit": "   inch", "status": "won ", "state": "New Mexico", "city": "ROSSENDALE", "rate": "61.7%", "name": "Omar  Cruz"}, {"unit": "km", "status": "in progress", "state": "   New Mexico", "city": "Rossendale", "rate": "15.8%", "name": "Mona Lee  "}, {"unit": "in", "status": "in  progress", "state": "new  mxeico", "city": "ROSSEDNALE", "rate": "68.1%", "name": "Alice Moore "}, {"unit": "litre", "status": "WON", "state": "  South Dakota", "city": "Rossendale", "rate": "11.1%", "name": "Heidi Cruz"}, {"unit": "liter", "status": "closed-wno", "state": "SD", "city": "Rossendale", "rate": "36.8%", "name": "Bob Cruz"}, {"unit": "Mi", "status": "open", "state": "new mexico", "city": "  Rossedale", "rate": "31.1%", "name": "Karl Park"}, {"unit": " Litre", "status": "new", "state": "NM", "city": "highton", "rate": "22.6%", "name": "Omar Johnson"}, {"unit": "l", "status": "won  ", "state": "nm", "city": "rossendale", "rate": "97.1%", "name": "Mona Adams  "}, {"unit": "liters", "status": " open ", "state": " La", "city": "Rossendale", "rate": "24.6%", "name": "Omar Adams"}, {"unit": "kilometer", "status": "Closed Won", "state": "sd", "city": "highton", "rate": "20.9%", "name": " Mona Reyes"}, {"unit": "LITRE", "status": "NEW", "state": "Nm", "city": "highton", "rate": "21.3%", "name": "Karl Johnson"}, {"unit": " kilometers ", "status": "ongoing", "state": "CA", "city": "Rossendale", "rate": "44.2%", "name": "Priya Brandt"}, {"unit": "kipometers ", "status": "Open", "state": "Louisiana", "city": " Rossendale", "rate": "27.4%", "name": "Carol Fischer"}, {"unit": "l", "status": "won", "state": " ca", "city": "ROSSENDALE", "rate": "62.0%", "name": "   Sara Cruz"}, {"unit": "", "status": "", "state": "", "city": "", "rate": "", "name": ""}], "clean": [{"unit": "km", "status": "Open", "state": "Louisiana", "city": "Rossendale", "rate": 0.971, "name": "Judy Lee"}, {"unit": "km", "status": "Open", "state": "New Mexico", "city": "Highton", "rate": 0.263, "name": "Lena Diaz"}, {"unit": "L", "status": "Won", "state": "California", "city": "Rossendale", "rate": 0.62, "name": "Sara Cruz"}, {"unit": "in", "status": "Won", "state": "South Dakota", "city": "Highton", "rate": 0.903, "name": "Priya Reyes"}, {"unit": "mi", "status": "Open", "state": "California", "city": "Highton", "rate": 0.032, "name": "Karl Fischer"}, {"unit": "in", "status": "Won", "state": "South Dakota", "city": "Rossendale", "rate": 0.6629999999999999, "name": "Alice Park"}, {"unit": "km", "status": "In Progress", "state": "California", "city": "Highton", "rate": 0.659, "name": "Heidi Diaz"}, {"unit": "mi", "status": "In Progress", "state": "Louisiana", "city": "Rossendale", "rate": 0.064, "name": "Mona Ali"}, {"unit": "mi", "status": "Open", "state": "California", "city": "Highton", "rate": 0.511, "name": "Judy Johnson"}, {"unit": "km", "status": "Open", "state": "Louisiana", "city": "Rossendale", "rate": 0.27399999999999997, "name": "Carol Fischer"}, {"unit": "L", "status": "In Progress", "state": "New Mexico", "city": "Rossendale", "rate": 0.305, "name": "Carol Reyes"}, {"unit": "km", "status": "Open", "state": "Louisiana", "city": "Rossendale", "rate": 0.503, "name": "Ivan Fischer"}, {"unit": "mi", "status": "Won", "state": "South Dakota", "city": "Rossendale", "rate": 0.255, "name": "Carol Park"}, {"unit": "km", "status": "Open", "state": "California", "city": "Highton", "rate": 0.02, "name": "Mona Moore"}, {"unit": "mi", "status": "Won", "state": "Louisiana", "city": "Highton", "rate": 0.41200000000000003, "name": "Mona Novak"}, {"unit": "mi", "status": "In Progress", "state": "New Mexico", "city": "Highton", "rate": 0.483, "name": "David Smith"}, {"unit": "km", "status": "Open", "state": "South Dakota", "city": "Highton", "rate": 0.262, "name": "Lena Moore"}, {"unit": "L", "status": "Open", "state": "South Dakota", "city": "Highton", "rate": 0.551, "name": "Bob Lee"}, {"unit": "L", "status": "Won", "state": "South Dakota", "city": "Rossendale", "rate": 0.28, "name": "Priya Park"}, {"unit": "mi", "status": "In Progress", "state": "New Mexico", "city": "Rossendale", "rate": 0.499, "name": "Heidi Wong"}, {"unit": "mi", "status": "Open", "state": "California", "city": "Rossendale", "rate": 0.021, "name": "Sara Novak"}, {"unit": "mi", "status": "Open", "state": "South Dakota", "city": "Highton", "rate": 0.154, "name": "Lena Smith"}, {"unit": "in", "status": "Won", "state": "New Mexico", "city": "Rossendale", "rate": 0.617, "name": "Omar Cruz"}, {"unit": "km", "status": "In Progress", "state": "New Mexico", "city": "Rossendale", "rate": 0.158, "name": "Mona Lee"}, {"unit": "in", "status": "In Progress", "state": "New Mexico", "city": "Rossendale", "rate": 0.6809999999999999, "name": "Alice Moore"}, {"unit": "L", "status": "Won", "state": "South Dakota", "city": "Rossendale", "rate": 0.111, "name": "Heidi Cruz"}, {"unit": "L", "status": "Won", "state": "South Dakota", "city": "Rossendale", "rate": 0.368, "name": "Bob Cruz"}, {"unit": "mi", "status": "Open", "state": "New Mexico", "city": "Rossendale", "rate": 0.311, "name": "Karl Park"}, {"unit": "L", "status": "Open", "state": "New Mexico", "city": "Highton", "rate": 0.226, "name": "Omar Johnson"}, {"unit": "L", "status": "Won", "state": "New Mexico", "city": "Rossendale", "rate": 0.971, "name": "Mona Adams"}, {"unit": "L", "status": "Open", "state": "Louisiana", "city": "Rossendale", "rate": 0.24600000000000002, "name": "Omar Adams"}, {"unit": "km", "status": "Won", "state": "South Dakota", "city": "Highton", "rate": 0.209, "name": "Mona Reyes"}, {"unit": "L", "status": "Open", "state": "New Mexico", "city": "Highton", "rate": 0.213, "name": "Karl Johnson"}, {"unit": "km", "status": "In Progress", "state": "California", "city": "Rossendale", "rate": 0.442, "name": "Priya Brandt"}], "dirty_cols": ["unit", "status", "state", "city", "rate", "name"], "clean_cols": ["unit", "status", "state", "city", "rate", "name"], "plan": {"dataset_summary": "37 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KILOMETERS": "km", "l": "L", "inch": "in", "Mi": "mi", "\"": "in", "miles": "mi", "kipometers": "km", "LITER": "L", "kilometer": "km", "mile": "mi", "MILE": "mi", "KM": "km", "MILES": "mi", "litre": "L", "liter": "L", "Litre": "L", "liters": "L", "LITRE": "L", "kilometers": "km"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"new": "Open", "won": "Won", "closed-won": "Won", "WON": "Won", "IN PROGRESS": "In Progress", "ONGOING": "In Progress", "open": "Open", "WIP": "In Progress", "Closed Won": "Won", "wip": "In Progress", "OPEN": "Open", "Oen": "Open", "in progress": "In Progress", "closed-wno": "Won", "NEW": "Open", "ongoing": "In Progress"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ca": "California", "SD": "South Dakota", "CA": "California", "LOUISIANA": "Louisiana", "california": "California", "nm": "New Mexico", "LA": "Louisiana", "sd": "South Dakota", "louisiana": "Louisiana", "south dakota": "South Dakota", "Soutth Dakota": "South Dakota", "Sd": "South Dakota", "CALIFORNIA": "California", "new mxeico": "New Mexico", "new mexico": "New Mexico", "NM": "New Mexico", "La": "Louisiana", "Nm": "New Mexico"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ROSSENDALE": "Rossendale", "highton": "Highton", "rossendale": "Rossendale", "hihton": "Highton", "Rostendale": "Rossendale", "Highotn": "Highton", "ROSSEDNALE": "Rossendale", "Rossedale": "Rossendale"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"department": "PEOPLE", "country": " AZ", "amount": "$7,918.93", "signup_date": "45216", "rate": "87.4%", "industry": " Utilities ", "currency": "ETB", "unit": "Kg"}, {"department": "biz dev", "country": "irq", "amount": "2.419,90", "signup_date": "25 Jul 2023", "rate": "83.8%", "industry": "energy", "currency": "€", "unit": "kilograms"}, {"department": "HR", "country": "MNE", "amount": "$3,887.93", "signup_date": "45098", "rate": "56.2%", "industry": "ENERGY", "currency": "   €", "unit": "KLOGRAM"}, {"department": "HR", "country": "   Republic of Finland", "amount": "$8,449.77", "signup_date": "7/19/2023", "rate": "98.6%", "industry": "Bankng", "currency": "€", "unit": " cm"}, {"department": "PEOPLE", "country": "Republic of Finland", "amount": "1.894,67", "signup_date": "2023-04-06", "rate": "87.8%", "industry": "energy", "currency": "etb ", "unit": "lbs"}, {"department": "people ops", "country": "Montenegro", "amount": "$2,644.54", "signup_date": "7/22/2023", "rate": "98.9%", "industry": "Telco", "currency": "Euro   ", "unit": "PONUD"}, {"department": "biz  dev", "country": "iq", "amount": "--", "signup_date": "6/22/2023", "rate": "76.5%", "industry": "telecom", "currency": "aed", "unit": "centimeter"}, {"department": "biz dev", "country": "Montenegro", "amount": "$8,628.82", "signup_date": "15 Dec 2023", "rate": "83.2%", "industry": "energy", "currency": "etb", "unit": "lb"}, {"department": "sales", "country": "Monteegro", "amount": "88,59", "signup_date": "3 Jan 2023", "rate": "84.4%", "industry": "Telecom", "currency": "Ethiopian Birr", "unit": "Lb"}, {"department": "Saales", "country": "azerbaijan", "amount": "$6,426.11", "signup_date": "7/2/2023", "rate": "60.4%", "industry": "TELCO  ", "currency": " AED", "unit": "centimeters"}, {"department": "Sales", "country": "Az", "amount": "na", "signup_date": "14 Jun 2023", "rate": "7.4%", "industry": "oil & gas", "currency": "Cabo Verde Escudo", "unit": "kilogam"}, {"department": "Biz Dev", "country": "aze", "amount": "1.787,53", "signup_date": "2 Oct 2023", "rate": "7.2%", "industry": "Comms ", "currency": "etb", "unit": "cm   "}, {"department": "Biz Dev", "country": "ME", "amount": "5.231,35", "signup_date": "23 Dec 2023", "rate": "24.4%", "industry": "telecom", "currency": "€", "unit": "CM"}, {"department": "   HUMAN RESOURCES", "country": "me", "amount": "880,97", "signup_date": "10/2/2023", "rate": "51.2%", "industry": "Tleecommunications", "currency": "€", "unit": "cm"}, {"department": "sales", "country": "ME", "amount": "$5,360.35", "signup_date": "6/13/2023", "rate": "73.1%", "industry": "   Comms", "currency": "€", "unit": " centimeter "}, {"department": "People Ops", "country": "Mne", "amount": "$4,198.50", "signup_date": "10/17/2023", "rate": "95.2%", "industry": " comms ", "currency": "Euro", "unit": " Centimeter "}, {"department": "Sals", "country": "iraq", "amount": "$320.03", "signup_date": "45073", "rate": "33.5%", "industry": "   Energy", "currency": "ETB", "unit": " CENTIMETER"}, {"department": "people ops", "country": "MONTENEGRO", "amount": "$4,070.63", "signup_date": "20 Jun 2023", "rate": "76.9%", "industry": "oil & gas  ", "currency": "Ethiopiqn Birr", "unit": " KILOGRAM "}, {"department": "  Hr", "country": "IRAQ", "amount": "2.067,86", "signup_date": "11/4/2023", "rate": "64.8%", "industry": "Utilities", "currency": "Cabo Verde Escudo  ", "unit": "Cm"}, {"department": "People", "country": "iraq", "amount": "6.788,67", "signup_date": "6 Jun 2023", "rate": "28.4%", "industry": "Utdlities", "currency": "aed", "unit": "Cm"}, {"department": "people ops", "country": "Azerbaijan  ", "amount": "3,360", "signup_date": "45005", "rate": "8.1%", "industry": "Oil & Gas  ", "currency": "ethiopiann birr", "unit": "pounds"}, {"department": "hr", "country": "FIN", "amount": "2.133,23", "signup_date": "9 Jan 2023", "rate": "42.2%", "industry": "FINSERV   ", "currency": "etb", "unit": "cm"}, {"department": "people", "country": "MNE", "amount": "117,99", "signup_date": "2023-04-11", "rate": "27.5%", "industry": "Utilities", "currency": "ethiopian birr", "unit": "centimeter"}, {"department": "PEOPLE", "country": "FI", "amount": "4.320,86", "signup_date": "2023-01-03", "rate": "68.7%", "industry": "Bankig", "currency": "Ethiopian Birr", "unit": "LB"}, {"department": "People Ops ", "country": "IQ", "amount": "3.487,71", "signup_date": "45030", "rate": "58.8%", "industry": "  fintech", "currency": "Ethiopian Birr", "unit": "Lb"}, {"department": "Peopple Ops", "country": "AZE", "amount": "6.515,07", "signup_date": "6/17/2023", "rate": "88.8%", "industry": "Teleocm", "currency": "cabo verde escudo", "unit": "Lb"}, {"department": "Biz Dev", "country": "REPUBLIC OF AZERBAIJAN  ", "amount": "$236.13", "signup_date": "45189", "rate": "37.9%", "industry": "ENERGY ", "currency": " AED ", "unit": "KILOGRAMS"}, {"department": "Hr", "country": "Montenegro", "amount": "$6,778.35", "signup_date": "45190", "rate": "76.2%", "industry": "Fintech", "currency": "euro", "unit": "POUND"}, {"department": "Saels", "country": "mne", "amount": "1.602,68", "signup_date": "2023-01-08", "rate": "9.4%", "industry": "   Financial Services", "currency": "Cabo Verde Escudo ", "unit": "   Lb"}, {"department": "Sales", "country": "Iraq", "amount": "$4,953.53", "signup_date": "5 Aug 2023", "rate": "35.8%", "industry": "Telecommunications", "currency": "Cabo Verde Escudo", "unit": "cm"}, {"department": "Biz Dev", "country": "  irq", "amount": "None", "signup_date": "20 Jun 2023", "rate": "74.8%", "industry": "TELECOM", "currency": "Etb", "unit": " kilograms "}, {"department": "biz dev", "country": "republic of iraq", "amount": "na", "signup_date": "2/13/2023", "rate": "75.8%", "industry": " Utiliies ", "currency": "CVE", "unit": "lbs"}, {"department": "People Ops", "country": "mne   ", "amount": "2.495,11", "signup_date": "7 Sep 2023", "rate": "14.3%", "industry": " Energy ", "currency": "cve", "unit": "centimeters"}, {"department": "Salees", "country": "FI", "amount": "--", "signup_date": "2/13/2023", "rate": "48.4%", "industry": " fintech ", "currency": "€  ", "unit": "Cm"}, {"department": "HR   ", "country": "FIN", "amount": "$6,727.46", "signup_date": "45011", "rate": "37.4%", "industry": "Energy", "currency": "ETHIOPIAN BIRR", "unit": "CM"}, {"department": " biz dev", "country": "irq", "amount": "4.956,73", "signup_date": "18 Jul 2023", "rate": "23.2%", "industry": " Comms", "currency": "Ethiopian Birr", "unit": "cm"}, {"department": "sales", "country": "MONTENEGRO", "amount": "$4,792.35", "signup_date": "44937", "rate": "85.5%", "industry": "Comms", "currency": "Ethiopian  Brir", "unit": "lbs"}, {"department": "HR", "country": "MNE", "amount": "$3,887.93", "signup_date": "45098", "rate": "56.2%", "industry": "ENERGY", "currency": "   €", "unit": "KLOGRAM"}], "clean": [{"department": "Human Resources", "country": "Azerbaijan", "amount": 7918.93, "signup_date": "2023-10-17", "rate": 0.8740000000000001, "industry": "Energy", "currency": "ETB", "unit": "kg"}, {"department": "Sales", "country": "Iraq", "amount": 2419.9, "signup_date": "2023-07-25", "rate": 0.838, "industry": "Energy", "currency": "EUR", "unit": "kg"}, {"department": "Human Resources", "country": "Montenegro", "amount": 3887.93, "signup_date": "2023-06-21", "rate": 0.562, "industry": "Energy", "currency": "EUR", "unit": "kg"}, {"department": "Human Resources", "country": "Finland", "amount": 8449.77, "signup_date": "2023-07-19", "rate": 0.986, "industry": "Financial Services", "currency": "EUR", "unit": "cm"}, {"department": "Human Resources", "country": "Finland", "amount": 1894.67, "signup_date": "2023-04-06", "rate": 0.878, "industry": "Energy", "currency": "ETB", "unit": "lb"}, {"department": "Human Resources", "country": "Montenegro", "amount": 2644.54, "signup_date": "2023-07-22", "rate": 0.9890000000000001, "industry": "Telecommunications", "currency": "EUR", "unit": "lb"}, {"department": "Sales", "country": "Iraq", "amount": NaN, "signup_date": "2023-06-22", "rate": 0.765, "industry": "Telecommunications", "currency": "AED", "unit": "cm"}, {"department": "Sales", "country": "Montenegro", "amount": 8628.82, "signup_date": "2023-12-15", "rate": 0.8320000000000001, "industry": "Energy", "currency": "ETB", "unit": "lb"}, {"department": "Sales", "country": "Montenegro", "amount": 88.59, "signup_date": "2023-01-03", "rate": 0.8440000000000001, "industry": "Telecommunications", "currency": "ETB", "unit": "lb"}, {"department": "Sales", "country": "Azerbaijan", "amount": 6426.11, "signup_date": "2023-07-02", "rate": 0.604, "industry": "Telecommunications", "currency": "AED", "unit": "cm"}, {"department": "Sales", "country": "Azerbaijan", "amount": NaN, "signup_date": "2023-06-14", "rate": 0.07400000000000001, "industry": "Energy", "currency": "CVE", "unit": "kg"}, {"department": "Sales", "country": "Azerbaijan", "amount": 1787.53, "signup_date": "2023-10-02", "rate": 0.07200000000000001, "industry": "Telecommunications", "currency": "ETB", "unit": "cm"}, {"department": "Sales", "country": "Montenegro", "amount": 5231.35, "signup_date": "2023-12-23", "rate": 0.244, "industry": "Telecommunications", "currency": "EUR", "unit": "cm"}, {"department": "Human Resources", "country": "Montenegro", "amount": 880.97, "signup_date": "2023-10-02", "rate": 0.512, "industry": "Telecommunications", "currency": "EUR", "unit": "cm"}, {"department": "Sales", "country": "Montenegro", "amount": 5360.35, "signup_date": "2023-06-13", "rate": 0.731, "industry": "Telecommunications", "currency": "EUR", "unit": "cm"}, {"department": "Human Resources", "country": "Montenegro", "amount": 4198.5, "signup_date": "2023-10-17", "rate": 0.9520000000000001, "industry": "Telecommunications", "currency": "EUR", "unit": "cm"}, {"department": "Sales", "country": "Iraq", "amount": 320.03, "signup_date": "2023-05-27", "rate": 0.335, "industry": "Energy", "currency": "ETB", "unit": "cm"}, {"department": "Human Resources", "country": "Montenegro", "amount": 4070.63, "signup_date": "2023-06-20", "rate": 0.769, "industry": "Energy", "currency": "ETB", "unit": "kg"}, {"department": "Human Resources", "country": "Iraq", "amount": 2067.86, "signup_date": "2023-11-04", "rate": 0.648, "industry": "Energy", "currency": "CVE", "unit": "cm"}, {"department": "Human Resources", "country": "Iraq", "amount": 6788.67, "signup_date": "2023-06-06", "rate": 0.284, "industry": "Energy", "currency": "AED", "unit": "cm"}, {"department": "Human Resources", "country": "Azerbaijan", "amount": 3360.0, "signup_date": "2023-03-20", "rate": 0.081, "industry": "Energy", "currency": "ETB", "unit": "lb"}, {"department": "Human Resources", "country": "Finland", "amount": 2133.23, "signup_date": "2023-01-09", "rate": 0.42200000000000004, "industry": "Financial Services", "currency": "ETB", "unit": "cm"}, {"department": "Human Resources", "country": "Montenegro", "amount": 117.99, "signup_date": "2023-04-11", "rate": 0.275, "industry": "Energy", "currency": "ETB", "unit": "cm"}, {"department": "Human Resources", "country": "Finland", "amount": 4320.86, "signup_date": "2023-01-03", "rate": 0.687, "industry": "Financial Services", "currency": "ETB", "unit": "lb"}, {"department": "Human Resources", "country": "Iraq", "amount": 3487.71, "signup_date": "2023-04-14", "rate": 0.588, "industry": "Financial Services", "currency": "ETB", "unit": "lb"}, {"department": "Human Resources", "country": "Azerbaijan", "amount": 6515.07, "signup_date": "2023-06-17", "rate": 0.888, "industry": "Telecommunications", "currency": "CVE", "unit": "lb"}, {"department": "Sales", "country": "Azerbaijan", "amount": 236.13, "signup_date": "2023-09-20", "rate": 0.379, "industry": "Energy", "currency": "AED", "unit": "kg"}, {"department": "Human Resources", "country": "Montenegro", "amount": 6778.35, "signup_date": "2023-09-21", "rate": 0.762, "industry": "Financial Services", "currency": "EUR", "unit": "lb"}, {"department": "Sales", "country": "Montenegro", "amount": 1602.68, "signup_date": "2023-01-08", "rate": 0.094, "industry": "Financial Services", "currency": "CVE", "unit": "lb"}, {"department": "Sales", "country": "Iraq", "amount": 4953.53, "signup_date": "2023-08-05", "rate": 0.358, "industry": "Telecommunications", "currency": "CVE", "unit": "cm"}, {"department": "Sales", "country": "Iraq", "amount": NaN, "signup_date": "2023-06-20", "rate": 0.748, "industry": "Telecommunications", "currency": "ETB", "unit": "kg"}, {"department": "Sales", "country": "Iraq", "amount": NaN, "signup_date": "2023-02-13", "rate": 0.758, "industry": "Energy", "currency": "CVE", "unit": "lb"}, {"department": "Human Resources", "country": "Montenegro", "amount": 2495.11, "signup_date": "2023-09-07", "rate": 0.14300000000000002, "industry": "Energy", "currency": "CVE", "unit": "cm"}, {"department": "Sales", "country": "Finland", "amount": NaN, "signup_date": "2023-02-13", "rate": 0.484, "industry": "Financial Services", "currency": "EUR", "unit": "cm"}, {"department": "Human Resources", "country": "Finland", "amount": 6727.46, "signup_date": "2023-03-26", "rate": 0.374, "industry": "Energy", "currency": "ETB", "unit": "cm"}, {"department": "Sales", "country": "Iraq", "amount": 4956.73, "signup_date": "2023-07-18", "rate": 0.23199999999999998, "industry": "Telecommunications", "currency": "ETB", "unit": "cm"}, {"department": "Sales", "country": "Montenegro", "amount": 4792.35, "signup_date": "2023-01-11", "rate": 0.855, "industry": "Telecommunications", "currency": "ETB", "unit": "lb"}], "dirty_cols": ["department", "country", "amount", "signup_date", "rate", "industry", "currency", "unit"], "clean_cols": ["department", "country", "amount", "signup_date", "rate", "industry", "currency", "unit"], "plan": {"dataset_summary": "38 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PEOPLE": "Human Resources", "biz dev": "Sales", "HR": "Human Resources", "people ops": "Human Resources", "sales": "Sales", "Saales": "Sales", "Biz Dev": "Sales", "HUMAN RESOURCES": "Human Resources", "People Ops": "Human Resources", "Sals": "Sales", "Hr": "Human Resources", "People": "Human Resources", "hr": "Human Resources", "people": "Human Resources", "Peopple Ops": "Human Resources", "Saels": "Sales", "Salees": "Sales"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AZ": "Azerbaijan", "irq": "Iraq", "MNE": "Montenegro", "Republic of Finland": "Finland", "iq": "Iraq", "Monteegro": "Montenegro", "azerbaijan": "Azerbaijan", "Az": "Azerbaijan", "aze": "Azerbaijan", "ME": "Montenegro", "me": "Montenegro", "Mne": "Montenegro", "iraq": "Iraq", "MONTENEGRO": "Montenegro", "IRAQ": "Iraq", "FIN": "Finland", "FI": "Finland", "IQ": "Iraq", "AZE": "Azerbaijan", "REPUBLIC OF AZERBAIJAN": "Azerbaijan", "mne": "Montenegro", "republic of iraq": "Iraq"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Utilities": "Energy", "energy": "Energy", "ENERGY": "Energy", "Bankng": "Financial Services", "Telco": "Telecommunications", "telecom": "Telecommunications", "Telecom": "Telecommunications", "TELCO": "Telecommunications", "oil & gas": "Energy", "Comms": "Telecommunications", "Tleecommunications": "Telecommunications", "comms": "Telecommunications", "Utdlities": "Energy", "Oil & Gas": "Energy", "FINSERV": "Financial Services", "Bankig": "Financial Services", "fintech": "Financial Services", "Teleocm": "Telecommunications", "Fintech": "Financial Services", "TELECOM": "Telecommunications", "Utiliies": "Energy"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"€": "EUR", "etb": "ETB", "Euro": "EUR", "aed": "AED", "Ethiopian Birr": "ETB", "Cabo Verde Escudo": "CVE", "Ethiopiqn Birr": "ETB", "ethiopiann birr": "ETB", "ethiopian birr": "ETB", "cabo verde escudo": "CVE", "euro": "EUR", "Etb": "ETB", "cve": "CVE", "ETHIOPIAN BIRR": "ETB", "Ethiopian Brir": "ETB"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Kg": "kg", "kilograms": "kg", "KLOGRAM": "kg", "lbs": "lb", "PONUD": "lb", "centimeter": "cm", "Lb": "lb", "centimeters": "cm", "kilogam": "kg", "CM": "cm", "Centimeter": "cm", "CENTIMETER": "cm", "KILOGRAM": "kg", "Cm": "cm", "pounds": "lb", "LB": "lb", "KILOGRAMS": "kg", "POUND": "lb"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"industry": "fintceh", "job_title": "engineering mr  ", "rate": "39.8%", "notes2": ""}, {"industry": "  FINSERV", "job_title": "vice president", "rate": "79.0%", "notes2": ""}, {"industry": "software", "job_title": "VP", "rate": "63.8%", "notes2": ""}, {"industry": "Bakning", "job_title": "vp", "rate": "15.0%", "notes2": ""}, {"industry": "tech", "job_title": "vp", "rate": "11.0%", "notes2": ""}, {"industry": "tech", "job_title": "vice pres", "rate": "70.4%", "notes2": ""}, {"industry": "FinServ", "job_title": "Vice President", "rate": "42.7%", "notes2": ""}, {"industry": "TECHNOLOGY", "job_title": " Eng Mgr", "rate": "83.5%", "notes2": ""}, {"industry": "Tech", "job_title": "v.p.  ", "rate": "43.4%", "notes2": ""}, {"industry": "Financial Services", "job_title": "Vice Pres", "rate": "97.6%", "notes2": ""}, {"industry": "Tech   ", "job_title": "vice  president", "rate": "71.5%", "notes2": ""}, {"industry": "Fnance", "job_title": "V..", "rate": "68.5%", "notes2": ""}, {"industry": "Fnancial Services", "job_title": "engineering manager", "rate": "16.0%", "notes2": ""}, {"industry": "Financial Services", "job_title": "Engineering Mgr", "rate": "48.3%", "notes2": ""}, {"industry": "  tech", "job_title": "vice pres", "rate": "90.2%", "notes2": ""}, {"industry": "fintech   ", "job_title": "ENG MGR", "rate": "40.7%", "notes2": ""}, {"industry": "TECH", "job_title": "eng  mgr", "rate": "6.5%", "notes2": ""}, {"industry": "tech", "job_title": "Etg Manager", "rate": "55.9%", "notes2": ""}, {"industry": "TECH", "job_title": "Eng Mgr", "rate": "61.9%", "notes2": ""}, {"industry": "FINANCE", "job_title": "Eng Mgr   ", "rate": "11.3%", "notes2": ""}, {"industry": "Fintech", "job_title": "V.P.", "rate": "19.4%", "notes2": ""}, {"industry": " Technology ", "job_title": "   ENGINEERING MGR", "rate": "93.0%", "notes2": ""}, {"industry": "tech  ", "job_title": "Vice Pres", "rate": "78.4%", "notes2": ""}, {"industry": "Financial Services", "job_title": "vice president", "rate": "58.7%", "notes2": ""}, {"industry": "Banking", "job_title": "ENG  MANAGER", "rate": "9.5%", "notes2": ""}, {"industry": "tech", "job_title": "Etg Manager", "rate": "55.9%", "notes2": ""}, {"industry": "Tech   ", "job_title": "vice  president", "rate": "71.5%", "notes2": ""}], "clean": [{"industry": "Financial Services", "job_title": "Engineering Manager", "rate": 0.39799999999999996}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.79}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.638}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.15}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.11}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.7040000000000001}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.42700000000000005}, {"industry": "Technology", "job_title": "Engineering Manager", "rate": 0.835}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.434}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.976}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.715}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.685}, {"industry": "Financial Services", "job_title": "Engineering Manager", "rate": 0.16}, {"industry": "Financial Services", "job_title": "Engineering Manager", "rate": 0.483}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.902}, {"industry": "Financial Services", "job_title": "Engineering Manager", "rate": 0.40700000000000003}, {"industry": "Technology", "job_title": "Engineering Manager", "rate": 0.065}, {"industry": "Technology", "job_title": "Engineering Manager", "rate": 0.5589999999999999}, {"industry": "Technology", "job_title": "Engineering Manager", "rate": 0.619}, {"industry": "Financial Services", "job_title": "Engineering Manager", "rate": 0.113}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.19399999999999998}, {"industry": "Technology", "job_title": "Engineering Manager", "rate": 0.93}, {"industry": "Technology", "job_title": "Vice President", "rate": 0.784}, {"industry": "Financial Services", "job_title": "Vice President", "rate": 0.5870000000000001}, {"industry": "Financial Services", "job_title": "Engineering Manager", "rate": 0.095}], "dirty_cols": ["industry", "job_title", "rate", "notes2"], "clean_cols": ["industry", "job_title", "rate"], "plan": {"dataset_summary": "27 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"fintceh": "Financial Services", "FINSERV": "Financial Services", "software": "Technology", "Bakning": "Financial Services", "tech": "Technology", "FinServ": "Financial Services", "TECHNOLOGY": "Technology", "Tech": "Technology", "Fnance": "Financial Services", "Fnancial Services": "Financial Services", "fintech": "Financial Services", "TECH": "Technology", "FINANCE": "Financial Services", "Fintech": "Financial Services", "Banking": "Financial Services"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"engineering mr": "Engineering Manager", "vice president": "Vice President", "VP": "Vice President", "vp": "Vice President", "vice pres": "Vice President", "Eng Mgr": "Engineering Manager", "v.p.": "Vice President", "Vice Pres": "Vice President", "V..": "Vice President", "engineering manager": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "ENG MGR": "Engineering Manager", "eng mgr": "Engineering Manager", "Etg Manager": "Engineering Manager", "V.P.": "Vice President", "ENGINEERING MGR": "Engineering Manager", "ENG MANAGER": "Engineering Manager"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"email": "kngai@example.com", "state": "Florida", "city": "razgrad", "status": "High", "rate": "87.7%", "country": "Tonga", "amount": "1.780,87", "extra": ""}, {"email": "FKPCH@CORP.IO ", "state": "CT", "city": "Mbinga", "status": "P3", "rate": "36.8%", "country": " TO ", "amount": "N/A", "extra": ""}, {"email": " AMKMMJ@EXAMPLE.COM", "state": "DC", "city": "Razgrad", "status": "P1", "rate": "56.8%", "country": "SAINT LUCIA", "amount": "#N/A", "extra": ""}, {"email": "   kaajde@test.org", "state": "Guam", "city": "Razgrad ", "status": "L", "rate": "24.0%", "country": "Guernsey", "amount": "-", "extra": ""}, {"email": " hbfk@test.org", "state": "fl", "city": "Razgrad", "status": "medium", "rate": "92.8%", "country": "TON", "amount": "3.163,71", "extra": ""}, {"email": "  pgkkefi@test.org", "state": "Guam ", "city": "Mbinga", "status": "Medium ", "rate": "48.2%", "country": "  TO", "amount": "8.861,28", "extra": ""}, {"email": "fbdb@corp.io", "state": "pr", "city": "Razgrad", "status": "low", "rate": "34.3%", "country": "  Gg", "amount": "?", "extra": ""}, {"email": "KKNP@MAIL.COM  ", "state": "DC", "city": "Razgrad", "status": "P3", "rate": "95.0%", "country": "Guernsey", "amount": "2.192,91", "extra": ""}, {"email": "   JCIIEKC@EXAMPLE.COM", "state": "PR", "city": "Razgrad", "status": "H", "rate": "59.9%", "country": "Romania", "amount": "#N/A", "extra": ""}, {"email": " pkbj@example.com ", "state": " pr ", "city": "Mbinga", "status": "HIGH", "rate": "72.6%", "country": "GUERNSEY", "amount": "6.325,02", "extra": ""}, {"email": "bkbhmb@mail.com", "state": " GU ", "city": "Mbinga   ", "status": "Medium", "rate": "52.4%", "country": "romania  ", "amount": "8.336,92", "extra": ""}, {"email": "boaoh@corp.io", "state": "FL", "city": " rzgrad ", "status": "M", "rate": "30.8%", "country": " GUERNSEY", "amount": "7.285,62", "extra": ""}, {"email": "  occleb@example.com", "state": "FL", "city": "Mbinga", "status": "P3 ", "rate": "69.9%", "country": "Romania", "amount": "6.777,91", "extra": ""}, {"email": "blnbmh@mail.com", "state": "dc", "city": "razgrad", "status": "hih", "rate": "41.0%", "country": "Guernsey", "amount": "6.859,29", "extra": ""}, {"email": " gpejcj@corp.io ", "state": "PR", "city": "razgrad", "status": " MED ", "rate": "72.6%", "country": "rou", "amount": "7.320,51", "extra": ""}, {"email": "hejfa@corp.io", "state": "PR", "city": "mbinga", "status": " P3 ", "rate": "94.3%", "country": "   Guernsey", "amount": "7.543,14", "extra": ""}, {"email": "  dbbbh@corp.io", "state": "Connecticut", "city": "   Razgrad", "status": "P3", "rate": "85.3%", "country": "lc", "amount": "$3,049.29", "extra": ""}, {"email": "anjkae@example.com   ", "state": "ct ", "city": "Mbinga", "status": "HIGH", "rate": "94.2%", "country": "ro", "amount": "3.736,11", "extra": ""}, {"email": "AMAJG@CORP.IO", "state": "florida", "city": "Mbinga", "status": "  M", "rate": "56.3%", "country": "ROU ", "amount": "4.749,99", "extra": ""}, {"email": "  MLAEMBI@MAIL.COM", "state": "Guam", "city": "razgrad", "status": "P2", "rate": "12.2%", "country": "LCA", "amount": "4.157,66", "extra": ""}, {"email": "DEJNHL@MAIL.COM", "state": " DC", "city": "mbinga", "status": "mefium", "rate": "37.3%", "country": "gg", "amount": "$674.78", "extra": ""}, {"email": "DGENMF@EXAMPLE.COM  ", "state": "CT", "city": "razgrad", "status": "MEDIUM", "rate": "36.1%", "country": "kingdmo of  tonga", "amount": "1.415,55", "extra": ""}, {"email": "BCONLP@TEST.ORG", "state": "Puerto Rico", "city": "Mbinga", "status": "p1", "rate": "55.5%", "country": "to", "amount": "-", "extra": ""}, {"email": "CDCCC@TEST.ORG", "state": "gu", "city": "Mbinga   ", "status": "L", "rate": "63.5%", "country": "LC", "amount": "$3,400.03", "extra": ""}, {"email": "  lkjbb@example.com", "state": "connecticut", "city": "Mbinga ", "status": "P2", "rate": "61.1%", "country": "   GGY", "amount": "2.279,87", "extra": ""}, {"email": " amcaege@test.org ", "state": "DC  ", "city": "Mbinga", "status": "P1", "rate": "57.7%", "country": "Romania", "amount": "null", "extra": ""}, {"email": "dbfce@corp.io", "state": "Ct", "city": " RAZGRAD", "status": "l", "rate": "24.8%", "country": "Marshall Islands", "amount": "4.911,41", "extra": ""}, {"email": "bnhdlm@test.org", "state": "District of Columbia", "city": " Mbinga ", "status": "p3", "rate": "31.2%", "country": "Saint Lucia", "amount": "$4,685.86", "extra": ""}, {"email": "GJJK@MAIL.COM ", "state": " CT ", "city": "razgrad", "status": "MEDIUM", "rate": "81.3%", "country": "Republic of the Marshall Islands  ", "amount": "$6,035.27", "extra": ""}, {"email": "hikcp@corp.io", "state": "gu", "city": "Razgrad", "status": " L ", "rate": "35.3%", "country": " GGY ", "amount": "6,618", "extra": ""}, {"email": " HBJJM@MAIL.COM ", "state": "DISTRICT OF COLUMBIA", "city": "Mbinga", "status": "p2", "rate": "1.8%", "country": "RO", "amount": "2.002,85", "extra": ""}, {"email": "fpnkcbb@corp.io", "state": "PR", "city": "Mbinga", "status": "P1", "rate": "72.9%", "country": "LC", "amount": "7.719,34", "extra": ""}, {"email": "GIIBMKM@MAIL.COM", "state": "District  of Columbia", "city": "mbinga", "status": " med ", "rate": "29.1%", "country": "Mh", "amount": "2.125,89", "extra": ""}, {"email": "hbpmih@test.org", "state": "PR", "city": "Razgrad", "status": "Low", "rate": "49.4%", "country": "  Lc", "amount": "7.796,01", "extra": ""}, {"email": "   HAPEGKL@CORP.IO", "state": "District of Columbia", "city": "MBINGA", "status": "m", "rate": "74.0%", "country": "to", "amount": "6.629,74", "extra": ""}, {"email": "ccpkahb@corp.io", "state": "DC", "city": "Razgrad", "status": "low   ", "rate": "67.4%", "country": " ro ", "amount": "$4,311.04", "extra": ""}, {"email": "KLOH@MAIL.COM", "state": "fl", "city": "mbinga", "status": " P3 ", "rate": "50.7%", "country": "Guernsey", "amount": "--", "extra": ""}, {"email": "JBGNMH@TEST.ORG", "state": "dc", "city": "mbinga", "status": "M", "rate": "76.6%", "country": "MARSHALL ISLANDS", "amount": "$1,801.70", "extra": ""}, {"email": "FMBJKD@TEST.ORG", "state": "FLORIDA", "city": "   Mbinga", "status": "low", "rate": "47.5%", "country": "Mhl", "amount": "7.624,58", "extra": ""}, {"email": "EDBMNKL@TEST.ORG", "state": "District of Columbia", "city": "Mbinga", "status": "P2", "rate": "26.3%", "country": "ROU", "amount": "$4,214.90", "extra": ""}, {"email": "FALAL@CORP.IO", "state": "Florida", "city": " razgrad", "status": "low", "rate": "71.9%", "country": "LCA", "amount": "#N/A", "extra": ""}, {"email": "nhcfl@test.org", "state": "Connecticut", "city": "RAZGRAD", "status": "High  ", "rate": "14.9%", "country": "   lc", "amount": "1.896,13", "extra": ""}, {"email": " JPDFBBM@MAIL.COM", "state": " dc ", "city": "  razgrad", "status": "P2", "rate": "29.9%", "country": " TON ", "amount": "#N/A", "extra": ""}, {"email": "JGIG@EXAMPLE.COM", "state": "Puerto Rico", "city": "razgrad", "status": " P1 ", "rate": "9.8%", "country": "guernsey", "amount": "TBD", "extra": ""}, {"email": "LJDDBL@EXAMPLE.COM ", "state": "PR ", "city": "Mbinga", "status": "medium", "rate": "5.7%", "country": "KINGDOM OF TONGA", "amount": "$1,863.75", "extra": ""}], "clean": [{"email": "kngai@example.com", "state": "Florida", "city": "Razgrad", "status": "High", "rate": 0.877, "country": "Tonga", "amount": 1780.87}, {"email": "fkpch@corp.io", "state": "Connecticut", "city": "Mbinga", "status": "Low", "rate": 0.368, "country": "Tonga", "amount": NaN}, {"email": "amkmmj@example.com", "state": "District of Columbia", "city": "Razgrad", "status": "High", "rate": 0.568, "country": "Saint Lucia", "amount": NaN}, {"email": "kaajde@test.org", "state": "Guam", "city": "Razgrad", "status": "Low", "rate": 0.24, "country": "Guernsey", "amount": NaN}, {"email": "hbfk@test.org", "state": "Florida", "city": "Razgrad", "status": "Medium", "rate": 0.9279999999999999, "country": "Tonga", "amount": 3163.71}, {"email": "pgkkefi@test.org", "state": "Guam", "city": "Mbinga", "status": "Medium", "rate": 0.48200000000000004, "country": "Tonga", "amount": 8861.28}, {"email": "fbdb@corp.io", "state": "Puerto Rico", "city": "Razgrad", "status": "Low", "rate": 0.34299999999999997, "country": "Guernsey", "amount": NaN}, {"email": "kknp@mail.com", "state": "District of Columbia", "city": "Razgrad", "status": "Low", "rate": 0.95, "country": "Guernsey", "amount": 2192.91}, {"email": "jciiekc@example.com", "state": "Puerto Rico", "city": "Razgrad", "status": "High", "rate": 0.599, "country": "Romania", "amount": NaN}, {"email": "pkbj@example.com", "state": "Puerto Rico", "city": "Mbinga", "status": "High", "rate": 0.726, "country": "Guernsey", "amount": 6325.02}, {"email": "bkbhmb@mail.com", "state": "Guam", "city": "Mbinga", "status": "Medium", "rate": 0.524, "country": "Romania", "amount": 8336.92}, {"email": "boaoh@corp.io", "state": "Florida", "city": "Razgrad", "status": "Medium", "rate": 0.308, "country": "Guernsey", "amount": 7285.62}, {"email": "occleb@example.com", "state": "Florida", "city": "Mbinga", "status": "Low", "rate": 0.6990000000000001, "country": "Romania", "amount": 6777.91}, {"email": "blnbmh@mail.com", "state": "District of Columbia", "city": "Razgrad", "status": "High", "rate": 0.41, "country": "Guernsey", "amount": 6859.29}, {"email": "gpejcj@corp.io", "state": "Puerto Rico", "city": "Razgrad", "status": "Medium", "rate": 0.726, "country": "Romania", "amount": 7320.51}, {"email": "hejfa@corp.io", "state": "Puerto Rico", "city": "Mbinga", "status": "Low", "rate": 0.943, "country": "Guernsey", "amount": 7543.14}, {"email": "dbbbh@corp.io", "state": "Connecticut", "city": "Razgrad", "status": "Low", "rate": 0.853, "country": "Saint Lucia", "amount": 3049.29}, {"email": "anjkae@example.com", "state": "Connecticut", "city": "Mbinga", "status": "High", "rate": 0.9420000000000001, "country": "Romania", "amount": 3736.11}, {"email": "amajg@corp.io", "state": "Florida", "city": "Mbinga", "status": "Medium", "rate": 0.563, "country": "Romania", "amount": 4749.99}, {"email": "mlaembi@mail.com", "state": "Guam", "city": "Razgrad", "status": "Medium", "rate": 0.122, "country": "Saint Lucia", "amount": 4157.66}, {"email": "dejnhl@mail.com", "state": "District of Columbia", "city": "Mbinga", "status": "Medium", "rate": 0.373, "country": "Guernsey", "amount": 674.78}, {"email": "dgenmf@example.com", "state": "Connecticut", "city": "Razgrad", "status": "Medium", "rate": 0.361, "country": "Tonga", "amount": 1415.55}, {"email": "bconlp@test.org", "state": "Puerto Rico", "city": "Mbinga", "status": "High", "rate": 0.555, "country": "Tonga", "amount": NaN}, {"email": "cdccc@test.org", "state": "Guam", "city": "Mbinga", "status": "Low", "rate": 0.635, "country": "Saint Lucia", "amount": 3400.03}, {"email": "lkjbb@example.com", "state": "Connecticut", "city": "Mbinga", "status": "Medium", "rate": 0.611, "country": "Guernsey", "amount": 2279.87}, {"email": "amcaege@test.org", "state": "District of Columbia", "city": "Mbinga", "status": "High", "rate": 0.5770000000000001, "country": "Romania", "amount": NaN}, {"email": "dbfce@corp.io", "state": "Connecticut", "city": "Razgrad", "status": "Low", "rate": 0.248, "country": "Marshall Islands", "amount": 4911.41}, {"email": "bnhdlm@test.org", "state": "District of Columbia", "city": "Mbinga", "status": "Low", "rate": 0.312, "country": "Saint Lucia", "amount": 4685.86}, {"email": "gjjk@mail.com", "state": "Connecticut", "city": "Razgrad", "status": "Medium", "rate": 0.813, "country": "Marshall Islands", "amount": 6035.27}, {"email": "hikcp@corp.io", "state": "Guam", "city": "Razgrad", "status": "Low", "rate": 0.353, "country": "Guernsey", "amount": 6618.0}, {"email": "hbjjm@mail.com", "state": "District of Columbia", "city": "Mbinga", "status": "Medium", "rate": 0.018000000000000002, "country": "Romania", "amount": 2002.85}, {"email": "fpnkcbb@corp.io", "state": "Puerto Rico", "city": "Mbinga", "status": "High", "rate": 0.7290000000000001, "country": "Saint Lucia", "amount": 7719.34}, {"email": "giibmkm@mail.com", "state": "District of Columbia", "city": "Mbinga", "status": "Medium", "rate": 0.29100000000000004, "country": "Marshall Islands", "amount": 2125.89}, {"email": "hbpmih@test.org", "state": "Puerto Rico", "city": "Razgrad", "status": "Low", "rate": 0.494, "country": "Saint Lucia", "amount": 7796.01}, {"email": "hapegkl@corp.io", "state": "District of Columbia", "city": "Mbinga", "status": "Medium", "rate": 0.74, "country": "Tonga", "amount": 6629.74}, {"email": "ccpkahb@corp.io", "state": "District of Columbia", "city": "Razgrad", "status": "Low", "rate": 0.674, "country": "Romania", "amount": 4311.04}, {"email": "kloh@mail.com", "state": "Florida", "city": "Mbinga", "status": "Low", "rate": 0.507, "country": "Guernsey", "amount": NaN}, {"email": "jbgnmh@test.org", "state": "District of Columbia", "city": "Mbinga", "status": "Medium", "rate": 0.7659999999999999, "country": "Marshall Islands", "amount": 1801.7}, {"email": "fmbjkd@test.org", "state": "Florida", "city": "Mbinga", "status": "Low", "rate": 0.475, "country": "Marshall Islands", "amount": 7624.58}, {"email": "edbmnkl@test.org", "state": "District of Columbia", "city": "Mbinga", "status": "Medium", "rate": 0.263, "country": "Romania", "amount": 4214.9}, {"email": "falal@corp.io", "state": "Florida", "city": "Razgrad", "status": "Low", "rate": 0.7190000000000001, "country": "Saint Lucia", "amount": NaN}, {"email": "nhcfl@test.org", "state": "Connecticut", "city": "Razgrad", "status": "High", "rate": 0.149, "country": "Saint Lucia", "amount": 1896.13}, {"email": "jpdfbbm@mail.com", "state": "District of Columbia", "city": "Razgrad", "status": "Medium", "rate": 0.299, "country": "Tonga", "amount": NaN}, {"email": "jgig@example.com", "state": "Puerto Rico", "city": "Razgrad", "status": "High", "rate": 0.098, "country": "Guernsey", "amount": NaN}, {"email": "ljddbl@example.com", "state": "Puerto Rico", "city": "Mbinga", "status": "Medium", "rate": 0.057, "country": "Tonga", "amount": 1863.75}], "dirty_cols": ["email", "state", "city", "status", "rate", "country", "amount", "extra"], "clean_cols": ["email", "state", "city", "status", "rate", "country", "amount"], "plan": {"dataset_summary": "45 rows × 8 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CT": "Connecticut", "DC": "District of Columbia", "fl": "Florida", "pr": "Puerto Rico", "PR": "Puerto Rico", "GU": "Guam", "FL": "Florida", "dc": "District of Columbia", "ct": "Connecticut", "florida": "Florida", "gu": "Guam", "connecticut": "Connecticut", "Ct": "Connecticut", "DISTRICT OF COLUMBIA": "District of Columbia", "FLORIDA": "Florida"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"razgrad": "Razgrad", "rzgrad": "Razgrad", "mbinga": "Mbinga", "RAZGRAD": "Razgrad", "MBINGA": "Mbinga"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P3": "Low", "P1": "High", "L": "Low", "medium": "Medium", "low": "Low", "H": "High", "HIGH": "High", "M": "Medium", "hih": "High", "MED": "Medium", "P2": "Medium", "mefium": "Medium", "MEDIUM": "Medium", "p1": "High", "l": "Low", "p3": "Low", "p2": "Medium", "med": "Medium", "m": "Medium"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TO": "Tonga", "SAINT LUCIA": "Saint Lucia", "TON": "Tonga", "Gg": "Guernsey", "GUERNSEY": "Guernsey", "romania": "Romania", "rou": "Romania", "lc": "Saint Lucia", "ro": "Romania", "ROU": "Romania", "LCA": "Saint Lucia", "gg": "Guernsey", "kingdmo of tonga": "Tonga", "to": "Tonga", "LC": "Saint Lucia", "GGY": "Guernsey", "Republic of the Marshall Islands": "Marshall Islands", "RO": "Romania", "Mh": "Marshall Islands", "Lc": "Saint Lucia", "MARSHALL ISLANDS": "Marshall Islands", "Mhl": "Marshall Islands", "guernsey": "Guernsey", "KINGDOM OF TONGA": "Tonga"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": []}}
{"dirty": [{"department": " Operations", "rate": "67.9%", "city": "te atatu south", "job_title": "Admin Asst", "col_x": ""}, {"department": " OPS ", "rate": "66.1%", "city": "Sydney", "job_title": "  vice president", "col_x": ""}, {"department": "   Operations", "rate": "78.0%", "city": "Maspaloyas", "job_title": "sr. engineer", "col_x": ""}, {"department": "eng.", "rate": "13.7%", "city": "maspalomas", "job_title": "Senior Engineer", "col_x": ""}, {"department": "MKTG", "rate": "17.0%", "city": "te atatu south", "job_title": "admin asst", "col_x": ""}, {"department": " ENGINEEERING", "rate": "92.1%", "city": "Te Atatu Soth", "job_title": "C.E.O.", "col_x": ""}, {"department": "  R&D", "rate": "58.1%", "city": "te atatu south ", "job_title": "c.e.o.", "col_x": ""}, {"department": " growth ", "rate": "62.4%", "city": "MASPALOMAS", "job_title": "sr edgineer", "col_x": ""}, {"department": "Operations", "rate": "28.2%", "city": "Sydney", "job_title": "CEO", "col_x": ""}, {"department": "ops   ", "rate": "39.2%", "city": "TE  ATATU SOUTH", "job_title": "  CHIEF EXECUTIVE OFFICER", "col_x": ""}, {"department": "eng", "rate": "36.2%", "city": "Sydney", "job_title": "Chief Executive Officer", "col_x": ""}, {"department": "mkt", "rate": "73.7%", "city": " Sydney", "job_title": "  VP", "col_x": ""}, {"department": "eng.", "rate": "86.3%", "city": "  Maspalomas", "job_title": "Administrative Assistant", "col_x": ""}, {"department": "Engineering", "rate": "53.0%", "city": "Maspalomas", "job_title": "Sr. Enggineer", "col_x": ""}, {"department": "Operations", "rate": "62.1%", "city": "MASPALOMAS ", "job_title": "SENIOR ENGINEER", "col_x": ""}, {"department": "Ops", "rate": "82.9%", "city": "Sydney ", "job_title": "vice pres", "col_x": ""}, {"department": "ops", "rate": "51.7%", "city": " Maspalomas", "job_title": "c.e.o.", "col_x": ""}, {"department": " Opeoations", "rate": "72.9%", "city": "sydney", "job_title": "admin assistant", "col_x": ""}, {"department": "Eng", "rate": "6.4%", "city": " Sydxey", "job_title": "Admin  Assistant", "col_x": ""}, {"department": "Dev", "rate": "86.3%", "city": "SYDNEY", "job_title": "admin assistant ", "col_x": ""}, {"department": "Mtkg", "rate": "64.7%", "city": "sydney", "job_title": "  vie president", "col_x": ""}, {"department": "Eng.  ", "rate": "83.7%", "city": "Sydney", "job_title": "Senior Engineer", "col_x": ""}, {"department": "operations", "rate": "65.3%", "city": "TE ATATU SOUTH", "job_title": "v.p.   ", "col_x": ""}, {"department": "Growth", "rate": "55.5%", "city": "sydney", "job_title": " Vp ", "col_x": ""}, {"department": "", "rate": "", "city": "", "job_title": "", "col_x": ""}], "clean": [{"department": "Operations", "rate": 0.679, "city": "Te Atatu South", "job_title": "Administrative Assistant"}, {"department": "Operations", "rate": 0.6609999999999999, "city": "Sydney", "job_title": "Vice President"}, {"department": "Operations", "rate": 0.78, "city": "Maspalomas", "job_title": "Senior Engineer"}, {"department": "Engineering", "rate": 0.13699999999999998, "city": "Maspalomas", "job_title": "Senior Engineer"}, {"department": "Marketing", "rate": 0.17, "city": "Te Atatu South", "job_title": "Administrative Assistant"}, {"department": "Engineering", "rate": 0.9209999999999999, "city": "Te Atatu South", "job_title": "Chief Executive Officer"}, {"department": "Engineering", "rate": 0.581, "city": "Te Atatu South", "job_title": "Chief Executive Officer"}, {"department": "Marketing", "rate": 0.624, "city": "Maspalomas", "job_title": "Senior Engineer"}, {"department": "Operations", "rate": 0.282, "city": "Sydney", "job_title": "Chief Executive Officer"}, {"department": "Operations", "rate": 0.392, "city": "Te Atatu South", "job_title": "Chief Executive Officer"}, {"department": "Engineering", "rate": 0.36200000000000004, "city": "Sydney", "job_title": "Chief Executive Officer"}, {"department": "Marketing", "rate": 0.737, "city": "Sydney", "job_title": "Vice President"}, {"department": "Engineering", "rate": 0.863, "city": "Maspalomas", "job_title": "Administrative Assistant"}, {"department": "Engineering", "rate": 0.53, "city": "Maspalomas", "job_title": "Senior Engineer"}, {"department": "Operations", "rate": 0.621, "city": "Maspalomas", "job_title": "Senior Engineer"}, {"department": "Operations", "rate": 0.8290000000000001, "city": "Sydney", "job_title": "Vice President"}, {"department": "Operations", "rate": 0.517, "city": "Maspalomas", "job_title": "Chief Executive Officer"}, {"department": "Operations", "rate": 0.7290000000000001, "city": "Sydney", "job_title": "Administrative Assistant"}, {"department": "Engineering", "rate": 0.064, "city": "Sydney", "job_title": "Administrative Assistant"}, {"department": "Engineering", "rate": 0.863, "city": "Sydney", "job_title": "Administrative Assistant"}, {"department": "Marketing", "rate": 0.647, "city": "Sydney", "job_title": "Vice President"}, {"department": "Engineering", "rate": 0.8370000000000001, "city": "Sydney", "job_title": "Senior Engineer"}, {"department": "Operations", "rate": 0.653, "city": "Te Atatu South", "job_title": "Vice President"}, {"department": "Marketing", "rate": 0.555, "city": "Sydney", "job_title": "Vice President"}], "dirty_cols": ["department", "rate", "city", "job_title", "col_x"], "clean_cols": ["department", "rate", "city", "job_title"], "plan": {"dataset_summary": "25 rows × 5 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OPS": "Operations", "eng.": "Engineering", "MKTG": "Marketing", "ENGINEEERING": "Engineering", "R&D": "Engineering", "growth": "Marketing", "ops": "Operations", "eng": "Engineering", "mkt": "Marketing", "Ops": "Operations", "Opeoations": "Operations", "Eng": "Engineering", "Dev": "Engineering", "Mtkg": "Marketing", "Eng.": "Engineering", "operations": "Operations", "Growth": "Marketing"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"te atatu south": "Te Atatu South", "Maspaloyas": "Maspalomas", "maspalomas": "Maspalomas", "Te Atatu Soth": "Te Atatu South", "MASPALOMAS": "Maspalomas", "TE ATATU SOUTH": "Te Atatu South", "sydney": "Sydney", "Sydxey": "Sydney", "SYDNEY": "Sydney"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Admin Asst": "Administrative Assistant", "vice president": "Vice President", "sr. engineer": "Senior Engineer", "admin asst": "Administrative Assistant", "C.E.O.": "Chief Executive Officer", "c.e.o.": "Chief Executive Officer", "sr edgineer": "Senior Engineer", "CEO": "Chief Executive Officer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "VP": "Vice President", "Sr. Enggineer": "Senior Engineer", "SENIOR ENGINEER": "Senior Engineer", "vice pres": "Vice President", "admin assistant": "Administrative Assistant", "Admin Assistant": "Administrative Assistant", "vie president": "Vice President", "v.p.": "Vice President", "Vp": "Vice President"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "  TACLOBAN", "status": "canceled", "email": "MJOKAPL@MAIL.COM", "department": "support", "job_title": "   VP", "unit": "kg", "company": "Acme Inc", "phone": "(568)619-8349"}, {"city": "Trenque Lauquen", "status": "active", "email": " MDKFN@TEST.ORG", "department": "Information Technology", "job_title": "Snr Engineer", "unit": "   km", "company": "Stark Industries", "phone": "533.689.5324"}, {"city": "Taclloban", "status": " active ", "email": "kdkhj@mail.com", "department": "  Mktg", "job_title": "Senior Eng", "unit": "kg ", "company": "Umbrella   ", "phone": "557-011-2542"}, {"city": "Jeddah   ", "status": "on hold", "email": "fnfaic@test.org", "department": "Customer  Support", "job_title": "Senior Engineer", "unit": "KG", "company": "Globex", "phone": "(599)211-7806"}, {"city": "Jeddah", "status": "Active", "email": " ALNLNPG@TEST.ORG ", "department": "Cust Support", "job_title": "VP", "unit": "  KG", "company": "Wonka Co", "phone": "527.846.0272"}, {"city": "   Tacloban", "status": "active", "email": "BHMC@CORP.IO", "department": "  Information Technology", "job_title": "product manager", "unit": "mi", "company": " Initech", "phone": "5479047788"}, {"city": "jeddah", "status": "Churned", "email": "FBBCJ@MAIL.COM", "department": "groxth", "job_title": "Sr. Enggineer", "unit": " kg ", "company": "Initech  ", "phone": "(599)087-0185"}, {"city": "Trenque Lauquen", "status": "free trial", "email": "lcml@test.org", "department": "information technology", "job_title": "prrod mgr", "unit": "Km", "company": "Wonka Co  ", "phone": "579-170-4535"}, {"city": "tresque lauquen", "status": "PAUSED", "email": "ANBM@MAIL.COM", "department": "CS", "job_title": "Product Manager", "unit": "Kiloogram", "company": "Soylent Corp", "phone": "(583)517-8695"}, {"city": "Trenque Lauquen", "status": "active", "email": "kdhpean@mail.com", "department": "mkt", "job_title": "sennior eng", "unit": "km", "company": "Soylent Corp", "phone": "(525)934-5531"}, {"city": " jeddah ", "status": "  ACTIVE", "email": " IMNG@MAIL.COM ", "department": "Mkt", "job_title": "Vice Pres", "unit": "mi", "company": "Vehement", "phone": "503.076.9002"}, {"city": "TRENQUE LAUQUEN  ", "status": "free  trial", "email": "  JFNI@TEST.ORG", "department": "Support", "job_title": " Sr Engineer", "unit": "kg", "company": " Initech", "phone": "567-953-9662"}, {"city": "Trenque Lauquen", "status": "paused", "email": "JLJI@MAIL.COM", "department": "Marketing", "job_title": "Sr Engneer", "unit": "kg", "company": "Umbrella", "phone": "(598)333-9713"}, {"city": "trenque eauquen", "status": "on hold", "email": " oihmi@example.com ", "department": " support ", "job_title": "  Sr Engineer", "unit": "kilometer", "company": "Vehement", "phone": "510.566.2670"}, {"city": "Tacloban", "status": "freetrial", "email": "JLBPOEA@CORP.IO", "department": "I.T.", "job_title": "Sr. Engineer", "unit": "miles", "company": "Hooli", "phone": "5184235385"}, {"city": "Jeddah", "status": "cancelled  ", "email": "ikpg@mail.com", "department": "INFORMATION TECHNOLOGY", "job_title": "Sr Engineer", "unit": "mile ", "company": "Acme  Inc", "phone": "5767817723"}, {"city": " Jeddah", "status": "Actise", "email": "fhmini@mail.com", "department": "Cust Support", "job_title": " VP", "unit": "kg", "company": "Acme  Inc", "phone": "(563)202-4652"}, {"city": "Trenque Lauquen", "status": "cancelled", "email": "   ENFNI@EXAMPLE.COM", "department": "CS", "job_title": "Product Manager", "unit": "mile", "company": "Vehement", "phone": "(574)542-8575"}, {"city": "jeddah  ", "status": "active", "email": " collhp@example.com ", "department": "Cust Support", "job_title": "Senior Engineer", "unit": "miles", "company": "Vehement ", "phone": "(542)616-8879"}, {"city": "Tacloban", "status": "canceled", "email": "bjia@test.org", "department": "support", "job_title": "V.P.", "unit": "kilometer", "company": " Hooli ", "phone": "541.019.9556"}, {"city": "Tacloban", "status": "on hold", "email": "  AMDK@EXAMPLE.COM", "department": "I.T.", "job_title": "Vice President", "unit": "kilometers", "company": "Acme Inc  ", "phone": "537.757.9287"}, {"city": "Treqnue Lauquen", "status": "cacneled", "email": "jleno@example.com", "department": "informatio ntechnology", "job_title": "  Vice Ptes", "unit": "km", "company": "Initech ", "phone": "(546)365-5903"}, {"city": " Trenque Lauquen", "status": "paused", "email": "oago@mail.com", "department": " it ", "job_title": "prod mgr", "unit": "mi", "company": " Umbrella", "phone": "521-373-5024"}, {"city": "Tacloban", "status": "  cagceled", "email": "OHOG@CORP.IO", "department": "Suport   ", "job_title": "  Product Manager", "unit": "KM", "company": "  Vehement", "phone": "(536)160-4306"}, {"city": "Trenquc Lauquen ", "status": "Active", "email": "JGOHPP@CORP.IO  ", "department": "Marketing ", "job_title": " Vice Pres", "unit": "Mi", "company": "   Vehement", "phone": "572-433-5543"}, {"city": "Trenque Lauquen", "status": "TRIAL", "email": " aijf@example.com", "department": "Supoprt", "job_title": "Sr Engineer", "unit": "Mi", "company": "Initech", "phone": "(511)500-1313"}, {"city": "Jeddah", "status": "paaused", "email": "mnmaghn@test.org", "department": "Information Technology", "job_title": "  Snr Engineer", "unit": "  killogram", "company": "Soylent  Corp", "phone": "5035297247"}, {"city": " Tacloban ", "status": "Trial", "email": "  MOJIFOP@MAIL.COM", "department": "Customer Support", "job_title": "Vice Pres", "unit": "kilograsm", "company": " Vehement ", "phone": "558.508.4233"}, {"city": "Tacloban", "status": "FREE TRIAL", "email": "elec@test.org", "department": "I.T.", "job_title": "PROD MGR", "unit": "Km", "company": "Wonka  Co", "phone": "(500)652-8930"}, {"city": "Trenque Lauquen", "status": "paused", "email": "LFPKJH@MAIL.COM", "department": "Marketing", "job_title": "vice pres", "unit": "milys", "company": "Wonka Co ", "phone": "527.455.1841"}, {"city": "Jeddah", "status": "CANCELD  ", "email": "CGON@MAIL.COM  ", "department": "Information Technolowy", "job_title": "snr engineer", "unit": "mi", "company": "Soylent Corp", "phone": "5462112942"}, {"city": "Trenque Lauquen", "status": "trial", "email": "  CDCFI@MAIL.COM", "department": "I.T.", "job_title": "SR. ENGINEER", "unit": "kilometer", "company": " Stark Industries", "phone": "5054613830"}, {"city": "   jeddah", "status": "TRIAL", "email": "GHPLFHF@CORP.IO", "department": "support", "job_title": "Prod  Mgr", "unit": "kilometer", "company": "Stark Industries", "phone": "5868438385"}, {"city": "   trenque lauuqen", "status": "trial", "email": "ppabb@example.com", "department": "I.T.", "job_title": "PROD MANAGER", "unit": "kg", "company": "Globex", "phone": "572-836-7773"}, {"city": "Jeddah", "status": "canceled", "email": "  MBAAPGA@TEST.ORG", "department": "IT", "job_title": "prod manager", "unit": "kilomettrs", "company": "Globex", "phone": "(503)208-1427"}, {"city": "tacloban", "status": "ACTIVE", "email": " ophljgg@example.com", "department": "MARKETING", "job_title": "vp", "unit": "kilogram", "company": "Vehement", "phone": "5986567923"}, {"city": " Tacloban ", "status": "paused", "email": "injelok@example.com", "department": "Mkt", "job_title": "SR EGINEER", "unit": "Km", "company": " Umbrella ", "phone": "501-382-2040"}, {"city": "Tacloban", "status": "TRIAL", "email": " PNNF@TEST.ORG ", "department": "CUSTOMER SUPPORT", "job_title": "Vice Prejident", "unit": "Mi", "company": "Globex", "phone": "5637747025"}, {"city": "Jeddah   ", "status": "paused", "email": "dkicoc@test.org", "department": "Marketing", "job_title": "Vice  President", "unit": "  kilometer", "company": "Umbrella   ", "phone": "(541)924-3173"}, {"city": "Trenque Lauquen", "status": "churned", "email": "bbnpk@corp.io", "department": "grlwth", "job_title": "senior engineer", "unit": "kg", "company": "Vehement", "phone": "5932008872"}, {"city": "Tacloban", "status": "paused", "email": " BPCP@TEST.ORG ", "department": "Cust  Support", "job_title": "VICE PRES", "unit": "miles   ", "company": "Wonka Co ", "phone": "504-198-3235"}, {"city": " Jeddah ", "status": " churned ", "email": " lhkee@example.com ", "department": " CUST SUPPORT", "job_title": " Prod Mgr", "unit": "mile ", "company": "Cyberdyne", "phone": "5777824272"}, {"city": "Trenquk Lauquen", "status": "Active", "email": "hcanah@corp.io ", "department": "I.T.", "job_title": "Vice  President", "unit": "km", "company": "Acme Inc", "phone": "510.073.8185"}, {"city": "Trenque  Lauquen", "status": "Trtal", "email": " OGIPD@MAIL.COM ", "department": "Growth", "job_title": "Senior Engineer", "unit": "mi", "company": "Soylent Corp  ", "phone": "537.812.8569"}, {"city": "Trenque Lauquen", "status": "paused", "email": "LOFCPI@EXAMPLE.COM", "department": "Mkt", "job_title": "Sr Engineer", "unit": "km  ", "company": " Cyberdyne ", "phone": "556.817.6799"}, {"city": "   Tacloban", "status": "active", "email": "BHMC@CORP.IO", "department": "  Information Technology", "job_title": "product manager", "unit": "mi", "company": " Initech", "phone": "5479047788"}, {"city": " Tacloban ", "status": "paused", "email": "injelok@example.com", "department": "Mkt", "job_title": "SR EGINEER", "unit": "Km", "company": " Umbrella ", "phone": "501-382-2040"}], "clean": [{"city": "Tacloban", "status": "Churned", "email": "mjokapl@mail.com", "department": "Customer Support", "job_title": "Vice President", "unit": "kg", "company": "Acme Inc", "phone": "(568) 619-8349"}, {"city": "Trenque Lauquen", "status": "Active", "email": "mdkfn@test.org", "department": "Information Technology", "job_title": "Senior Engineer", "unit": "km", "company": "Stark Industries", "phone": "(533) 689-5324"}, {"city": "Tacloban", "status": "Active", "email": "kdkhj@mail.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "kg", "company": "Umbrella", "phone": "(557) 011-2542"}, {"city": "Jeddah", "status": "Paused", "email": "fnfaic@test.org", "department": "Customer Support", "job_title": "Senior Engineer", "unit": "kg", "company": "Globex", "phone": "(599) 211-7806"}, {"city": "Jeddah", "status": "Active", "email": "alnlnpg@test.org", "department": "Customer Support", "job_title": "Vice President", "unit": "kg", "company": "Wonka Co", "phone": "(527) 846-0272"}, {"city": "Tacloban", "status": "Active", "email": "bhmc@corp.io", "department": "Information Technology", "job_title": "Product Manager", "unit": "mi", "company": "Initech", "phone": "(547) 904-7788"}, {"city": "Jeddah", "status": "Churned", "email": "fbbcj@mail.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "kg", "company": "Initech", "phone": "(599) 087-0185"}, {"city": "Trenque Lauquen", "status": "Trial", "email": "lcml@test.org", "department": "Information Technology", "job_title": "Product Manager", "unit": "km", "company": "Wonka Co", "phone": "(579) 170-4535"}, {"city": "Trenque Lauquen", "status": "Paused", "email": "anbm@mail.com", "department": "Customer Support", "job_title": "Product Manager", "unit": "kg", "company": "Soylent Corp", "phone": "(583) 517-8695"}, {"city": "Trenque Lauquen", "status": "Active", "email": "kdhpean@mail.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "km", "company": "Soylent Corp", "phone": "(525) 934-5531"}, {"city": "Jeddah", "status": "Active", "email": "imng@mail.com", "department": "Marketing", "job_title": "Vice President", "unit": "mi", "company": "Vehement", "phone": "(503) 076-9002"}, {"city": "Trenque Lauquen", "status": "Trial", "email": "jfni@test.org", "department": "Customer Support", "job_title": "Senior Engineer", "unit": "kg", "company": "Initech", "phone": "(567) 953-9662"}, {"city": "Trenque Lauquen", "status": "Paused", "email": "jlji@mail.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "kg", "company": "Umbrella", "phone": "(598) 333-9713"}, {"city": "Trenque Lauquen", "status": "Paused", "email": "oihmi@example.com", "department": "Customer Support", "job_title": "Senior Engineer", "unit": "km", "company": "Vehement", "phone": "(510) 566-2670"}, {"city": "Tacloban", "status": "Trial", "email": "jlbpoea@corp.io", "department": "Information Technology", "job_title": "Senior Engineer", "unit": "mi", "company": "Hooli", "phone": "(518) 423-5385"}, {"city": "Jeddah", "status": "Churned", "email": "ikpg@mail.com", "department": "Information Technology", "job_title": "Senior Engineer", "unit": "mi", "company": "Acme Inc", "phone": "(576) 781-7723"}, {"city": "Jeddah", "status": "Active", "email": "fhmini@mail.com", "department": "Customer Support", "job_title": "Vice President", "unit": "kg", "company": "Acme Inc", "phone": "(563) 202-4652"}, {"city": "Trenque Lauquen", "status": "Churned", "email": "enfni@example.com", "department": "Customer Support", "job_title": "Product Manager", "unit": "mi", "company": "Vehement", "phone": "(574) 542-8575"}, {"city": "Jeddah", "status": "Active", "email": "collhp@example.com", "department": "Customer Support", "job_title": "Senior Engineer", "unit": "mi", "company": "Vehement", "phone": "(542) 616-8879"}, {"city": "Tacloban", "status": "Churned", "email": "bjia@test.org", "department": "Customer Support", "job_title": "Vice President", "unit": "km", "company": "Hooli", "phone": "(541) 019-9556"}, {"city": "Tacloban", "status": "Paused", "email": "amdk@example.com", "department": "Information Technology", "job_title": "Vice President", "unit": "km", "company": "Acme Inc", "phone": "(537) 757-9287"}, {"city": "Trenque Lauquen", "status": "Churned", "email": "jleno@example.com", "department": "Information Technology", "job_title": "Vice President", "unit": "km", "company": "Initech", "phone": "(546) 365-5903"}, {"city": "Trenque Lauquen", "status": "Paused", "email": "oago@mail.com", "department": "Information Technology", "job_title": "Product Manager", "unit": "mi", "company": "Umbrella", "phone": "(521) 373-5024"}, {"city": "Tacloban", "status": "Churned", "email": "ohog@corp.io", "department": "Customer Support", "job_title": "Product Manager", "unit": "km", "company": "Vehement", "phone": "(536) 160-4306"}, {"city": "Trenque Lauquen", "status": "Active", "email": "jgohpp@corp.io", "department": "Marketing", "job_title": "Vice President", "unit": "mi", "company": "Vehement", "phone": "(572) 433-5543"}, {"city": "Trenque Lauquen", "status": "Trial", "email": "aijf@example.com", "department": "Customer Support", "job_title": "Senior Engineer", "unit": "mi", "company": "Initech", "phone": "(511) 500-1313"}, {"city": "Jeddah", "status": "Paused", "email": "mnmaghn@test.org", "department": "Information Technology", "job_title": "Senior Engineer", "unit": "kg", "company": "Soylent Corp", "phone": "(503) 529-7247"}, {"city": "Tacloban", "status": "Trial", "email": "mojifop@mail.com", "department": "Customer Support", "job_title": "Vice President", "unit": "kg", "company": "Vehement", "phone": "(558) 508-4233"}, {"city": "Tacloban", "status": "Trial", "email": "elec@test.org", "department": "Information Technology", "job_title": "Product Manager", "unit": "km", "company": "Wonka Co", "phone": "(500) 652-8930"}, {"city": "Trenque Lauquen", "status": "Paused", "email": "lfpkjh@mail.com", "department": "Marketing", "job_title": "Vice President", "unit": "mi", "company": "Wonka Co", "phone": "(527) 455-1841"}, {"city": "Jeddah", "status": "Churned", "email": "cgon@mail.com", "department": "Information Technology", "job_title": "Senior Engineer", "unit": "mi", "company": "Soylent Corp", "phone": "(546) 211-2942"}, {"city": "Trenque Lauquen", "status": "Trial", "email": "cdcfi@mail.com", "department": "Information Technology", "job_title": "Senior Engineer", "unit": "km", "company": "Stark Industries", "phone": "(505) 461-3830"}, {"city": "Jeddah", "status": "Trial", "email": "ghplfhf@corp.io", "department": "Customer Support", "job_title": "Product Manager", "unit": "km", "company": "Stark Industries", "phone": "(586) 843-8385"}, {"city": "Trenque Lauquen", "status": "Trial", "email": "ppabb@example.com", "department": "Information Technology", "job_title": "Product Manager", "unit": "kg", "company": "Globex", "phone": "(572) 836-7773"}, {"city": "Jeddah", "status": "Churned", "email": "mbaapga@test.org", "department": "Information Technology", "job_title": "Product Manager", "unit": "km", "company": "Globex", "phone": "(503) 208-1427"}, {"city": "Tacloban", "status": "Active", "email": "ophljgg@example.com", "department": "Marketing", "job_title": "Vice President", "unit": "kg", "company": "Vehement", "phone": "(598) 656-7923"}, {"city": "Tacloban", "status": "Paused", "email": "injelok@example.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "km", "company": "Umbrella", "phone": "(501) 382-2040"}, {"city": "Tacloban", "status": "Trial", "email": "pnnf@test.org", "department": "Customer Support", "job_title": "Vice President", "unit": "mi", "company": "Globex", "phone": "(563) 774-7025"}, {"city": "Jeddah", "status": "Paused", "email": "dkicoc@test.org", "department": "Marketing", "job_title": "Vice President", "unit": "km", "company": "Umbrella", "phone": "(541) 924-3173"}, {"city": "Trenque Lauquen", "status": "Churned", "email": "bbnpk@corp.io", "department": "Marketing", "job_title": "Senior Engineer", "unit": "kg", "company": "Vehement", "phone": "(593) 200-8872"}, {"city": "Tacloban", "status": "Paused", "email": "bpcp@test.org", "department": "Customer Support", "job_title": "Vice President", "unit": "mi", "company": "Wonka Co", "phone": "(504) 198-3235"}, {"city": "Jeddah", "status": "Churned", "email": "lhkee@example.com", "department": "Customer Support", "job_title": "Product Manager", "unit": "mi", "company": "Cyberdyne", "phone": "(577) 782-4272"}, {"city": "Trenque Lauquen", "status": "Active", "email": "hcanah@corp.io", "department": "Information Technology", "job_title": "Vice President", "unit": "km", "company": "Acme Inc", "phone": "(510) 073-8185"}, {"city": "Trenque Lauquen", "status": "Trial", "email": "ogipd@mail.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "mi", "company": "Soylent Corp", "phone": "(537) 812-8569"}, {"city": "Trenque Lauquen", "status": "Paused", "email": "lofcpi@example.com", "department": "Marketing", "job_title": "Senior Engineer", "unit": "km", "company": "Cyberdyne", "phone": "(556) 817-6799"}], "dirty_cols": ["city", "status", "email", "department", "job_title", "unit", "company", "phone"], "clean_cols": ["city", "status", "email", "department", "job_title", "unit", "company", "phone"], "plan": {"dataset_summary": "47 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TACLOBAN": "Tacloban", "Taclloban": "Tacloban", "jeddah": "Jeddah", "tresque lauquen": "Trenque Lauquen", "TRENQUE LAUQUEN": "Trenque Lauquen", "trenque eauquen": "Trenque Lauquen", "Treqnue Lauquen": "Trenque Lauquen", "Trenquc Lauquen": "Trenque Lauquen", "trenque lauuqen": "Trenque Lauquen", "tacloban": "Tacloban", "Trenquk Lauquen": "Trenque Lauquen"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"canceled": "Churned", "active": "Active", "on hold": "Paused", "free trial": "Trial", "PAUSED": "Paused", "ACTIVE": "Active", "paused": "Paused", "freetrial": "Trial", "cancelled": "Churned", "Actise": "Active", "cacneled": "Churned", "cagceled": "Churned", "TRIAL": "Trial", "paaused": "Paused", "FREE TRIAL": "Trial", "CANCELD": "Churned", "trial": "Trial", "churned": "Churned", "Trtal": "Trial"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"support": "Customer Support", "Mktg": "Marketing", "Cust Support": "Customer Support", "groxth": "Marketing", "information technology": "Information Technology", "CS": "Customer Support", "mkt": "Marketing", "Mkt": "Marketing", "Support": "Customer Support", "I.T.": "Information Technology", "INFORMATION TECHNOLOGY": "Information Technology", "informatio ntechnology": "Information Technology", "it": "Information Technology", "Suport": "Customer Support", "Supoprt": "Customer Support", "Information Technolowy": "Information Technology", "IT": "Information Technology", "MARKETING": "Marketing", "CUSTOMER SUPPORT": "Customer Support", "grlwth": "Marketing", "CUST SUPPORT": "Customer Support", "Growth": "Marketing"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"VP": "Vice President", "Snr Engineer": "Senior Engineer", "Senior Eng": "Senior Engineer", "product manager": "Product Manager", "Sr. Enggineer": "Senior Engineer", "prrod mgr": "Product Manager", "sennior eng": "Senior Engineer", "Vice Pres": "Vice President", "Sr Engineer": "Senior Engineer", "Sr Engneer": "Senior Engineer", "Sr. Engineer": "Senior Engineer", "V.P.": "Vice President", "Vice Ptes": "Vice President", "prod mgr": "Product Manager", "PROD MGR": "Product Manager", "vice pres": "Vice President", "snr engineer": "Senior Engineer", "SR. ENGINEER": "Senior Engineer", "Prod Mgr": "Product Manager", "PROD MANAGER": "Product Manager", "prod manager": "Product Manager", "vp": "Vice President", "SR EGINEER": "Senior Engineer", "Vice Prejident": "Vice President", "senior engineer": "Senior Engineer", "VICE PRES": "Vice President"}, "rationale": "Unified 26 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KG": "kg", "Km": "km", "Kiloogram": "kg", "kilometer": "km", "miles": "mi", "mile": "mi", "kilometers": "km", "KM": "km", "Mi": "mi", "killogram": "kg", "kilograsm": "kg", "milys": "mi", "kilomettrs": "km", "kilogram": "kg"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"company": "Soylent Corp   ", "unit": " Centimeter ", "phone": "(532)235-8895", "amount": "6.529,67", "country": "CL", "email": "HIKBFD@TEST.ORG   ", "is_active": "No", "col_x": ""}, {"company": "Initech", "unit": "ml ", "phone": "5570900080", "amount": "5.089,87", "country": "SR ", "email": "lgfnko@corp.io ", "is_active": "TRUE", "col_x": ""}, {"company": "Umbrella", "unit": "kilometers", "phone": "(546)879-7060", "amount": "3.551,52", "country": "SUR", "email": "  NCCN@EXAMPLE.COM", "is_active": "T", "col_x": ""}, {"company": "Globex   ", "unit": " ml ", "phone": "5544930579", "amount": "7.405,01", "country": "montserrat", "email": " digd@example.com ", "is_active": "FALSE", "col_x": ""}, {"company": "Vehement", "unit": " km", "phone": "5263518422", "amount": "8.277,18", "country": "Republic of Chile", "email": "plfjipb@mail.com  ", "is_active": "TRUE", "col_x": ""}, {"company": "Cyberdyne ", "unit": " km ", "phone": "5346719545", "amount": "6.535,29", "country": "SR  ", "email": "IDNLNO@EXAMPLE.COM ", "is_active": "Y", "col_x": ""}, {"company": "Soylent Corp", "unit": "kiwometers", "phone": "568-120-4359", "amount": "TBD", "country": "MSR", "email": "   anehco@example.com", "is_active": "F", "col_x": ""}, {"company": "Acme Inc ", "unit": "  km", "phone": "5701476756", "amount": "6.789,04", "country": "cl", "email": "HFHIC@MAIL.COM", "is_active": "true", "col_x": ""}, {"company": "Umbrella", "unit": "centmieter", "phone": "(518)953-1909", "amount": "$2,333.30", "country": "Sur", "email": "AJFPP@EXAMPLE.COM", "is_active": "true", "col_x": ""}, {"company": " Soylent Corp", "unit": "milliliter", "phone": "5659146826", "amount": "7.064,02", "country": "Republic Of Chile", "email": "gimf@corp.io", "is_active": "false", "col_x": ""}, {"company": "   Cyberdyne", "unit": "kilometer", "phone": "5758651030", "amount": "2.700,84", "country": "SUR", "email": "   JKMAJ@CORP.IO", "is_active": "FALSE", "col_x": ""}, {"company": " Initech", "unit": "km", "phone": "5246019309", "amount": "4.310,52", "country": "Suriname", "email": "kbdchf@mail.com", "is_active": "N", "col_x": ""}, {"company": " Stark Industries", "unit": "mls", "phone": "570.492.3459", "amount": "$5,298.51", "country": "ms", "email": "nnhmblm@mail.com   ", "is_active": "F", "col_x": ""}, {"company": "  Hooli", "unit": "kilometer  ", "phone": "(584)018-5855", "amount": "$8,166.18", "country": " suriname", "email": "  jmflgb@test.org", "is_active": "false", "col_x": ""}, {"company": "Wonka Co", "unit": "kilometers", "phone": "580.444.1051", "amount": "3.429,96", "country": " CL ", "email": "obeep@corp.io  ", "is_active": "FALSE", "col_x": ""}, {"company": "   Umbrella", "unit": "cm", "phone": "555.875.1971", "amount": "?", "country": "CL", "email": "  PNBMP@CORP.IO", "is_active": "false", "col_x": ""}, {"company": "Hooli", "unit": "centimeters", "phone": "(546)853-1288", "amount": "2.198,22", "country": "sr", "email": "   cfiic@example.com", "is_active": "N", "col_x": ""}, {"company": "Globex", "unit": "mls", "phone": "(501)072-4575", "amount": "2.130,76", "country": "   cl", "email": "ohem@corp.io", "is_active": "0", "col_x": ""}, {"company": "Soylent Corp ", "unit": "KM", "phone": "561-327-1800", "amount": "None", "country": "MSR", "email": " HDKCK@EXAMPLE.COM ", "is_active": "1", "col_x": ""}, {"company": "  Globex", "unit": "kilometer", "phone": "579-447-7647", "amount": "2.824,07", "country": "MS", "email": "   BJLOD@EXAMPLE.COM", "is_active": "Y", "col_x": ""}, {"company": "Umbrella", "unit": "Km", "phone": "595-745-7710", "amount": "4.287,49", "country": "   MSR", "email": "flemogp@test.org", "is_active": "No", "col_x": ""}, {"company": "   Hooli", "unit": "cm", "phone": "544.529.2873", "amount": "$4,330.27", "country": "Montserrat  ", "email": "gkemadj@example.com", "is_active": "T", "col_x": ""}, {"company": "Stark Industries  ", "unit": "km", "phone": "5979926335", "amount": "$4,098.93", "country": "Republic Of Surinme", "email": "DKNPEMP@MAIL.COM", "is_active": "T", "col_x": ""}, {"company": "Umbrella", "unit": "centimeters", "phone": "588-956-4947", "amount": "$1,934.82", "country": "Republic of Chile", "email": "dlcpdgb@mail.com", "is_active": "N", "col_x": ""}, {"company": "Globex", "unit": "kilometer", "phone": "5443133787", "amount": "$2,401.37", "country": " MSR", "email": "omipbng@test.org", "is_active": "N", "col_x": ""}, {"company": " Cyberdyne ", "unit": "mls", "phone": "532.649.7907", "amount": "4.948,83", "country": " chile ", "email": "olbhdde@corp.io", "is_active": "F", "col_x": ""}, {"company": "Wonka  Co", "unit": "   millliter", "phone": "573-232-2321", "amount": "TBD", "country": "MSR", "email": "MBIF@EXAMPLE.COM", "is_active": "Y", "col_x": ""}, {"company": "   Wonka Co", "unit": "centimeter", "phone": "567-195-1486", "amount": "8.744,96", "country": "Msr", "email": "dcbe@mail.com  ", "is_active": "TRUE", "col_x": ""}, {"company": "Wonka Co", "unit": "kliometers", "phone": "511-935-0679", "amount": "$8,217.26", "country": "Montserrat", "email": "ncoofi@corp.io", "is_active": "0", "col_x": ""}, {"company": "Vehement", "unit": "milliliter", "phone": "5390834294", "amount": "$581.28", "country": " ms", "email": "ache@corp.io", "is_active": "No", "col_x": ""}, {"company": "Umbrella", "unit": "  Kilometer", "phone": "524-597-6991", "amount": "$7,860.25", "country": "republic of surinamme", "email": "phkjeo@example.com", "is_active": "T", "col_x": ""}, {"company": "Soylent Corp  ", "unit": "KM", "phone": "5772701911", "amount": "2.366,40", "country": "Montserrat", "email": "KHADEE@CORP.IO", "is_active": "0", "col_x": ""}, {"company": "Stark Industries  ", "unit": "   km", "phone": "502.773.9243", "amount": "-", "country": "sr", "email": "  LIOP@EXAMPLE.COM", "is_active": "0", "col_x": ""}, {"company": "  Vehement", "unit": "centimeters", "phone": "592.043.1492", "amount": "8.457,62", "country": "msr", "email": " maei@test.org ", "is_active": "N", "col_x": ""}, {"company": "Stark Industries", "unit": "km", "phone": "502-957-1420", "amount": "678,32", "country": "REPUBLIC OF CHILE", "email": " DOAO@MAIL.COM", "is_active": "FALSE", "col_x": ""}, {"company": "Umbrella", "unit": "mls", "phone": "(553)285-8625", "amount": "?", "country": "MSR", "email": "KAFPN@EXAMPLE.COM", "is_active": "false", "col_x": ""}, {"company": "Initech", "unit": "mls", "phone": "581.960.1536", "amount": "$5,509.91", "country": "Republic of Chile", "email": "giacj@example.com", "is_active": "true", "col_x": ""}, {"company": " Globex ", "unit": "ML", "phone": "508-901-5850", "amount": "4.417,15", "country": "REPUBLIC OF CHILE", "email": "  pemegh@mail.com", "is_active": "F", "col_x": ""}, {"company": "   Hooli", "unit": "kilometer", "phone": "594-612-4101", "amount": "--", "country": " Republic of Chile", "email": "   EEBIJ@EXAMPLE.COM", "is_active": "Y", "col_x": ""}, {"company": "Stark Industries  ", "unit": "ml  ", "phone": "(537)199-1903", "amount": "--", "country": " suriname ", "email": " FADBPC@EXAMPLE.COM ", "is_active": "Y", "col_x": ""}, {"company": " Hooli", "unit": "kilometer", "phone": "5318123321", "amount": "2.273,12", "country": "   Republic of Chile", "email": "COPI@CORP.IO", "is_active": "true", "col_x": ""}, {"company": "   Stark Industries", "unit": "CM", "phone": "(501)988-0927", "amount": "$5,027.58", "country": "sr ", "email": "BOPA@TEST.ORG", "is_active": "T", "col_x": ""}, {"company": " Hooli ", "unit": "Mls", "phone": "513-491-7894", "amount": "-", "country": "Ms", "email": "APIB@EXAMPLE.COM ", "is_active": "No", "col_x": ""}, {"company": " Cyberdyne ", "unit": "mL", "phone": "519.079.7774", "amount": "5.255,85", "country": "Suriname", "email": "pjcndmi@test.org", "is_active": "No", "col_x": ""}, {"company": "   Umbrella", "unit": "KM", "phone": "(568)425-0635", "amount": "1.400,33", "country": "MS", "email": "  ECBNBC@CORP.IO", "is_active": "1", "col_x": ""}, {"company": "Umbrella", "unit": " CENTIMETERS", "phone": "(523)173-4224", "amount": "$406.03", "country": "  CHL", "email": "plajli@test.org", "is_active": "N", "col_x": ""}, {"company": "Wonka Co", "unit": "Cm", "phone": "510-568-2765", "amount": "7.984,48", "country": " MS", "email": "dipm@example.com", "is_active": "0", "col_x": ""}, {"company": "Wonka Co", "unit": " cm ", "phone": "540.578.5494", "amount": "1.168,97", "country": "suriname", "email": " IKLOBG@EXAMPLE.COM ", "is_active": "true", "col_x": ""}, {"company": "Soylent Corp", "unit": "centimeters", "phone": "512-253-0492", "amount": "6.603,04", "country": "MSR ", "email": "icam@corp.io", "is_active": "false", "col_x": ""}, {"company": " Acme Inc", "unit": "   Cm", "phone": "5023598292", "amount": "?", "country": "chile", "email": "AEINK@MAIL.COM", "is_active": "1", "col_x": ""}, {"company": "   Soylent Corp", "unit": "centimeters", "phone": "5377556999", "amount": "2.699,94", "country": "sur", "email": " DDLMC@TEST.ORG", "is_active": "Y", "col_x": ""}, {"company": "Acme Inc", "unit": "mls", "phone": "569-633-4773", "amount": "$6,250.06", "country": "CL", "email": "cbhk@example.com", "is_active": "No", "col_x": ""}, {"company": "  Stark Industries", "unit": "cm", "phone": "(565)779-0632", "amount": "6.865,73", "country": "  MSR", "email": "efjcfeg@example.com", "is_active": "No", "col_x": ""}, {"company": " Globex ", "unit": "cm", "phone": "5951641139", "amount": "na", "country": "chlie  ", "email": "igifb@corp.io", "is_active": "N", "col_x": ""}, {"company": " Wonka Co", "unit": "Centimeters", "phone": "5313105098", "amount": "#N/A", "country": "Montserrat", "email": "ndjco@test.org ", "is_active": "Y", "col_x": ""}, {"company": "   Vehement", "unit": " centimeters", "phone": "5814389193", "amount": "3.852,21", "country": "Suriname", "email": " DLLD@EXAMPLE.COM ", "is_active": "0", "col_x": ""}, {"company": "Hooli", "unit": "KM", "phone": "595-945-4188", "amount": "$4,119.39", "country": " SR ", "email": "CIHEGJP@CORP.IO", "is_active": "0", "col_x": ""}, {"company": "Soylent Corp  ", "unit": "milliliter", "phone": "598-128-9085", "amount": "$4,716.20", "country": "sr", "email": " pceed@example.com ", "is_active": "TRUE", "col_x": ""}, {"company": "Umbrella", "unit": "Km", "phone": "595-745-7710", "amount": "4.287,49", "country": "   MSR", "email": "flemogp@test.org", "is_active": "No", "col_x": ""}], "clean": [{"company": "Soylent Corp", "unit": "cm", "phone": "(532) 235-8895", "amount": 6529.67, "country": "Chile", "email": "hikbfd@test.org", "is_active": false}, {"company": "Initech", "unit": "mL", "phone": "(557) 090-0080", "amount": 5089.87, "country": "Suriname", "email": "lgfnko@corp.io", "is_active": true}, {"company": "Umbrella", "unit": "km", "phone": "(546) 879-7060", "amount": 3551.52, "country": "Suriname", "email": "nccn@example.com", "is_active": true}, {"company": "Globex", "unit": "mL", "phone": "(554) 493-0579", "amount": 7405.01, "country": "Montserrat", "email": "digd@example.com", "is_active": false}, {"company": "Vehement", "unit": "km", "phone": "(526) 351-8422", "amount": 8277.18, "country": "Chile", "email": "plfjipb@mail.com", "is_active": true}, {"company": "Cyberdyne", "unit": "km", "phone": "(534) 671-9545", "amount": 6535.29, "country": "Suriname", "email": "idnlno@example.com", "is_active": true}, {"company": "Soylent Corp", "unit": "km", "phone": "(568) 120-4359", "amount": NaN, "country": "Montserrat", "email": "anehco@example.com", "is_active": false}, {"company": "Acme Inc", "unit": "km", "phone": "(570) 147-6756", "amount": 6789.04, "country": "Chile", "email": "hfhic@mail.com", "is_active": true}, {"company": "Umbrella", "unit": "cm", "phone": "(518) 953-1909", "amount": 2333.3, "country": "Suriname", "email": "ajfpp@example.com", "is_active": true}, {"company": "Soylent Corp", "unit": "mL", "phone": "(565) 914-6826", "amount": 7064.02, "country": "Chile", "email": "gimf@corp.io", "is_active": false}, {"company": "Cyberdyne", "unit": "km", "phone": "(575) 865-1030", "amount": 2700.84, "country": "Suriname", "email": "jkmaj@corp.io", "is_active": false}, {"company": "Initech", "unit": "km", "phone": "(524) 601-9309", "amount": 4310.52, "country": "Suriname", "email": "kbdchf@mail.com", "is_active": false}, {"company": "Stark Industries", "unit": "mL", "phone": "(570) 492-3459", "amount": 5298.51, "country": "Montserrat", "email": "nnhmblm@mail.com", "is_active": false}, {"company": "Hooli", "unit": "km", "phone": "(584) 018-5855", "amount": 8166.18, "country": "Suriname", "email": "jmflgb@test.org", "is_active": false}, {"company": "Wonka Co", "unit": "km", "phone": "(580) 444-1051", "amount": 3429.96, "country": "Chile", "email": "obeep@corp.io", "is_active": false}, {"company": "Umbrella", "unit": "cm", "phone": "(555) 875-1971", "amount": NaN, "country": "Chile", "email": "pnbmp@corp.io", "is_active": false}, {"company": "Hooli", "unit": "cm", "phone": "(546) 853-1288", "amount": 2198.22, "country": "Suriname", "email": "cfiic@example.com", "is_active": false}, {"company": "Globex", "unit": "mL", "phone": "(501) 072-4575", "amount": 2130.76, "country": "Chile", "email": "ohem@corp.io", "is_active": false}, {"company": "Soylent Corp", "unit": "km", "phone": "(561) 327-1800", "amount": NaN, "country": "Montserrat", "email": "hdkck@example.com", "is_active": true}, {"company": "Globex", "unit": "km", "phone": "(579) 447-7647", "amount": 2824.07, "country": "Montserrat", "email": "bjlod@example.com", "is_active": true}, {"company": "Umbrella", "unit": "km", "phone": "(595) 745-7710", "amount": 4287.49, "country": "Montserrat", "email": "flemogp@test.org", "is_active": false}, {"company": "Hooli", "unit": "cm", "phone": "(544) 529-2873", "amount": 4330.27, "country": "Montserrat", "email": "gkemadj@example.com", "is_active": true}, {"company": "Stark Industries", "unit": "km", "phone": "(597) 992-6335", "amount": 4098.93, "country": "Suriname", "email": "dknpemp@mail.com", "is_active": true}, {"company": "Umbrella", "unit": "cm", "phone": "(588) 956-4947", "amount": 1934.82, "country": "Chile", "email": "dlcpdgb@mail.com", "is_active": false}, {"company": "Globex", "unit": "km", "phone": "(544) 313-3787", "amount": 2401.37, "country": "Montserrat", "email": "omipbng@test.org", "is_active": false}, {"company": "Cyberdyne", "unit": "mL", "phone": "(532) 649-7907", "amount": 4948.83, "country": "Chile", "email": "olbhdde@corp.io", "is_active": false}, {"company": "Wonka Co", "unit": "mL", "phone": "(573) 232-2321", "amount": NaN, "country": "Montserrat", "email": "mbif@example.com", "is_active": true}, {"company": "Wonka Co", "unit": "cm", "phone": "(567) 195-1486", "amount": 8744.96, "country": "Montserrat", "email": "dcbe@mail.com", "is_active": true}, {"company": "Wonka Co", "unit": "km", "phone": "(511) 935-0679", "amount": 8217.26, "country": "Montserrat", "email": "ncoofi@corp.io", "is_active": false}, {"company": "Vehement", "unit": "mL", "phone": "(539) 083-4294", "amount": 581.28, "country": "Montserrat", "email": "ache@corp.io", "is_active": false}, {"company": "Umbrella", "unit": "km", "phone": "(524) 597-6991", "amount": 7860.25, "country": "Suriname", "email": "phkjeo@example.com", "is_active": true}, {"company": "Soylent Corp", "unit": "km", "phone": "(577) 270-1911", "amount": 2366.4, "country": "Montserrat", "email": "khadee@corp.io", "is_active": false}, {"company": "Stark Industries", "unit": "km", "phone": "(502) 773-9243", "amount": NaN, "country": "Suriname", "email": "liop@example.com", "is_active": false}, {"company": "Vehement", "unit": "cm", "phone": "(592) 043-1492", "amount": 8457.62, "country": "Montserrat", "email": "maei@test.org", "is_active": false}, {"company": "Stark Industries", "unit": "km", "phone": "(502) 957-1420", "amount": 678.32, "country": "Chile", "email": "doao@mail.com", "is_active": false}, {"company": "Umbrella", "unit": "mL", "phone": "(553) 285-8625", "amount": NaN, "country": "Montserrat", "email": "kafpn@example.com", "is_active": false}, {"company": "Initech", "unit": "mL", "phone": "(581) 960-1536", "amount": 5509.91, "country": "Chile", "email": "giacj@example.com", "is_active": true}, {"company": "Globex", "unit": "mL", "phone": "(508) 901-5850", "amount": 4417.15, "country": "Chile", "email": "pemegh@mail.com", "is_active": false}, {"company": "Hooli", "unit": "km", "phone": "(594) 612-4101", "amount": NaN, "country": "Chile", "email": "eebij@example.com", "is_active": true}, {"company": "Stark Industries", "unit": "mL", "phone": "(537) 199-1903", "amount": NaN, "country": "Suriname", "email": "fadbpc@example.com", "is_active": true}, {"company": "Hooli", "unit": "km", "phone": "(531) 812-3321", "amount": 2273.12, "country": "Chile", "email": "copi@corp.io", "is_active": true}, {"company": "Stark Industries", "unit": "cm", "phone": "(501) 988-0927", "amount": 5027.58, "country": "Suriname", "email": "bopa@test.org", "is_active": true}, {"company": "Hooli", "unit": "mL", "phone": "(513) 491-7894", "amount": NaN, "country": "Montserrat", "email": "apib@example.com", "is_active": false}, {"company": "Cyberdyne", "unit": "mL", "phone": "(519) 079-7774", "amount": 5255.85, "country": "Suriname", "email": "pjcndmi@test.org", "is_active": false}, {"company": "Umbrella", "unit": "km", "phone": "(568) 425-0635", "amount": 1400.33, "country": "Montserrat", "email": "ecbnbc@corp.io", "is_active": true}, {"company": "Umbrella", "unit": "cm", "phone": "(523) 173-4224", "amount": 406.03, "country": "Chile", "email": "plajli@test.org", "is_active": false}, {"company": "Wonka Co", "unit": "cm", "phone": "(510) 568-2765", "amount": 7984.48, "country": "Montserrat", "email": "dipm@example.com", "is_active": false}, {"company": "Wonka Co", "unit": "cm", "phone": "(540) 578-5494", "amount": 1168.97, "country": "Suriname", "email": "iklobg@example.com", "is_active": true}, {"company": "Soylent Corp", "unit": "cm", "phone": "(512) 253-0492", "amount": 6603.04, "country": "Montserrat", "email": "icam@corp.io", "is_active": false}, {"company": "Acme Inc", "unit": "cm", "phone": "(502) 359-8292", "amount": NaN, "country": "Chile", "email": "aeink@mail.com", "is_active": true}, {"company": "Soylent Corp", "unit": "cm", "phone": "(537) 755-6999", "amount": 2699.94, "country": "Suriname", "email": "ddlmc@test.org", "is_active": true}, {"company": "Acme Inc", "unit": "mL", "phone": "(569) 633-4773", "amount": 6250.06, "country": "Chile", "email": "cbhk@example.com", "is_active": false}, {"company": "Stark Industries", "unit": "cm", "phone": "(565) 779-0632", "amount": 6865.73, "country": "Montserrat", "email": "efjcfeg@example.com", "is_active": false}, {"company": "Globex", "unit": "cm", "phone": "(595) 164-1139", "amount": NaN, "country": "Chile", "email": "igifb@corp.io", "is_active": false}, {"company": "Wonka Co", "unit": "cm", "phone": "(531) 310-5098", "amount": NaN, "country": "Montserrat", "email": "ndjco@test.org", "is_active": true}, {"company": "Vehement", "unit": "cm", "phone": "(581) 438-9193", "amount": 3852.21, "country": "Suriname", "email": "dlld@example.com", "is_active": false}, {"company": "Hooli", "unit": "km", "phone": "(595) 945-4188", "amount": 4119.39, "country": "Suriname", "email": "cihegjp@corp.io", "is_active": false}, {"company": "Soylent Corp", "unit": "mL", "phone": "(598) 128-9085", "amount": 4716.2, "country": "Suriname", "email": "pceed@example.com", "is_active": true}], "dirty_cols": ["company", "unit", "phone", "amount", "country", "email", "is_active", "col_x"], "clean_cols": ["company", "unit", "phone", "amount", "country", "email", "is_active"], "plan": {"dataset_summary": "59 rows × 8 columns. 7 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Centimeter": "cm", "ml": "mL", "kilometers": "km", "kiwometers": "km", "centmieter": "cm", "milliliter": "mL", "kilometer": "km", "mls": "mL", "centimeters": "cm", "KM": "km", "Km": "km", "millliter": "mL", "centimeter": "cm", "kliometers": "km", "Kilometer": "km", "ML": "mL", "CM": "cm", "Mls": "mL", "CENTIMETERS": "cm", "Cm": "cm", "Centimeters": "cm"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CL": "Chile", "SR": "Suriname", "SUR": "Suriname", "montserrat": "Montserrat", "Republic of Chile": "Chile", "MSR": "Montserrat", "cl": "Chile", "Sur": "Suriname", "Republic Of Chile": "Chile", "ms": "Montserrat", "suriname": "Suriname", "sr": "Suriname", "MS": "Montserrat", "Republic Of Surinme": "Suriname", "chile": "Chile", "Msr": "Montserrat", "republic of surinamme": "Suriname", "msr": "Montserrat", "REPUBLIC OF CHILE": "Chile", "Ms": "Montserrat", "CHL": "Chile", "sur": "Suriname", "chlie": "Chile"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"city": "Bloemhof", "country": "KR", "unit": "mile", "name": "Alice Ali", "company": "   Globex"}, {"city": "Tadarka", "country": "Republic of Azerbaijan   ", "unit": "centimeters", "name": "Karl Brandt", "company": "Stark Industries"}, {"city": "Piendamo", "country": "KOREA REPUBLIC OF", "unit": " kilometers", "name": "Judy Diaz", "company": "Initech"}, {"city": "Tatarka", "country": "HMD", "unit": "Miwe", "name": "Eve Ali", "company": "Acme Inc"}, {"city": "piendamo", "country": "republic of azerbaijan  ", "unit": " miles", "name": "Carol Khan", "company": "Stark Industries   "}, {"city": " Bloemhof", "country": "   UM", "unit": "cm", "name": "Heidi Park  ", "company": " Initech"}, {"city": " BLOEMHOF", "country": "Azerbaijan", "unit": "miles", "name": "Judy Cruz", "company": "Cyberdyne"}, {"city": "bloemhof", "country": "republic  of azerbaijan", "unit": "L", "name": "  Sara Lee", "company": "  Initech"}, {"city": "piendamo", "country": "United States Minor Outlying Islands  ", "unit": "Km", "name": "Frank Novak", "company": "Stark Industries  "}, {"city": "tatarka", "country": "Heard Island and McDonald Islands", "unit": "mile", "name": "Judy Moore", "company": "   Vehement"}, {"city": "bloemhof", "country": "KOR", "unit": "KM", "name": "  David Diaz", "company": " Acme Inc"}, {"city": "Piendamo", "country": "korea, republic of", "unit": "Centimeters", "name": "  Heidi Brandt", "company": "  Cyberdyne"}, {"city": "piendamo", "country": "united states minor outlying islands", "unit": "litre", "name": "Grace Reyes", "company": "Umbrella"}, {"city": "Picndamo", "country": "AZ ", "unit": "centimeter", "name": "Alice  Reyes", "company": "Acme Inc"}, {"city": " Tatarka ", "country": "hmd", "unit": " L ", "name": "Heidi Novak", "company": " Umbrella"}], "clean": [{"city": "Bloemhof", "country": "Korea, Republic of", "unit": "mi", "name": "Alice Ali", "company": "Globex"}, {"city": "Tatarka", "country": "Azerbaijan", "unit": "cm", "name": "Karl Brandt", "company": "Stark Industries"}, {"city": "Piendamo", "country": "Korea, Republic of", "unit": "km", "name": "Judy Diaz", "company": "Initech"}, {"city": "Tatarka", "country": "Heard Island and McDonald Islands", "unit": "mi", "name": "Eve Ali", "company": "Acme Inc"}, {"city": "Piendamo", "country": "Azerbaijan", "unit": "mi", "name": "Carol Khan", "company": "Stark Industries"}, {"city": "Bloemhof", "country": "United States Minor Outlying Islands", "unit": "cm", "name": "Heidi Park", "company": "Initech"}, {"city": "Bloemhof", "country": "Azerbaijan", "unit": "mi", "name": "Judy Cruz", "company": "Cyberdyne"}, {"city": "Bloemhof", "country": "Azerbaijan", "unit": "L", "name": "Sara Lee", "company": "Initech"}, {"city": "Piendamo", "country": "United States Minor Outlying Islands", "unit": "km", "name": "Frank Novak", "company": "Stark Industries"}, {"city": "Tatarka", "country": "Heard Island and McDonald Islands", "unit": "mi", "name": "Judy Moore", "company": "Vehement"}, {"city": "Bloemhof", "country": "Korea, Republic of", "unit": "km", "name": "David Diaz", "company": "Acme Inc"}, {"city": "Piendamo", "country": "Korea, Republic of", "unit": "cm", "name": "Heidi Brandt", "company": "Cyberdyne"}, {"city": "Piendamo", "country": "United States Minor Outlying Islands", "unit": "L", "name": "Grace Reyes", "company": "Umbrella"}, {"city": "Piendamo", "country": "Azerbaijan", "unit": "cm", "name": "Alice Reyes", "company": "Acme Inc"}, {"city": "Tatarka", "country": "Heard Island and McDonald Islands", "unit": "L", "name": "Heidi Novak", "company": "Umbrella"}], "dirty_cols": ["city", "country", "unit", "name", "company"], "clean_cols": ["city", "country", "unit", "name", "company"], "plan": {"dataset_summary": "15 rows × 5 columns. 5 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Tadarka": "Tatarka", "piendamo": "Piendamo", "BLOEMHOF": "Bloemhof", "bloemhof": "Bloemhof", "tatarka": "Tatarka", "Picndamo": "Piendamo"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KR": "Korea, Republic of", "Republic of Azerbaijan": "Azerbaijan", "KOREA REPUBLIC OF": "Korea, Republic of", "HMD": "Heard Island and McDonald Islands", "republic of azerbaijan": "Azerbaijan", "UM": "United States Minor Outlying Islands", "KOR": "Korea, Republic of", "korea, republic of": "Korea, Republic of", "united states minor outlying islands": "United States Minor Outlying Islands", "AZ": "Azerbaijan", "hmd": "Heard Island and McDonald Islands"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mile": "mi", "centimeters": "cm", "kilometers": "km", "Miwe": "mi", "miles": "mi", "Km": "km", "KM": "km", "Centimeters": "cm", "litre": "L", "centimeter": "cm"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"company": "Soylent Corp", "state": " CA ", "name": " Alice Smith", "currency": "Forint", "is_active": "0", "job_title": "pm", "city": "acacias"}, {"company": " Globex ", "state": "Minnesota", "name": "  Bob Diaz", "currency": "gibraltar  pound", "is_active": "0", "job_title": "Prod Manager", "city": "Acacias"}, {"company": " Cyberdyne ", "state": "California", "name": "Carol  Brandt", "currency": "Gibraltar Pound", "is_active": "false", "job_title": "CTO", "city": "ACACIAS"}, {"company": "Vehement", "state": "MN", "name": "Grace Moore   ", "currency": "Gibraltar Pound", "is_active": "0", "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Wonka Co ", "state": "mn ", "name": "   Lena Reyes", "currency": "Lari", "is_active": "F", "job_title": "chief technology officer", "city": "PUERTO DEL ROSARIO"}, {"company": "Stark Industries", "state": "MINNESOTA", "name": "Ivan Adams", "currency": "   HUF", "is_active": "TRUE", "job_title": "PM", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "Minnesota", "name": "Bob Brandt", "currency": "GIP", "is_active": "Yes", "job_title": "Chief Technology Officer", "city": "   Puerto del Rosario"}, {"company": "  Cyberdyne", "state": " MN ", "name": "Eve Moore", "currency": "Gibraltar Pound", "is_active": "Y", "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Globex", "state": "Minnesota", "name": "Ivan Cruz ", "currency": "LARI", "is_active": "F", "job_title": "CTO", "city": "Puerto del  Rosario"}, {"company": "Soylent Corp", "state": "Minnesota   ", "name": "Judy Adams", "currency": "lari", "is_active": "F", "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Globex   ", "state": "Minnesota", "name": "Ivan  Wong", "currency": " gel ", "is_active": "F", "job_title": "Product Mangaer", "city": "ACACIAS"}, {"company": "Wonka Co ", "state": "Minnesota", "name": "Judy  Petrov", "currency": "huf", "is_active": "true", "job_title": "C.T.O.", "city": "Puerto  del Rosario"}, {"company": "Cyberdyne", "state": "California", "name": "Ivan Petrov", "currency": "lari", "is_active": "Y", "job_title": "CTO", "city": "VILLASIS"}, {"company": "Umbrella   ", "state": "Minnelota", "name": "Ivan Petrov", "currency": " GIP", "is_active": "T", "job_title": "Prod Mgr", "city": "villasis"}, {"company": " Vehement ", "state": "Ca", "name": "Ivan Smith ", "currency": "Gibraltar Pound", "is_active": "TRUE", "job_title": "Prod Manager", "city": "puerto del rosario"}, {"company": "Cyberdyne  ", "state": "MN", "name": " Omar Smith", "currency": "LARI", "is_active": "false", "job_title": "CHIEF TECHNOLOGY OFFICER", "city": "Accias"}, {"company": "  Initech", "state": "MN", "name": "Alice  Adams", "currency": "GEL", "is_active": "1", "job_title": "Chief Technology Officer", "city": "villasis"}, {"company": "Umbrella", "state": "MN", "name": "Sara Cruz", "currency": "  Gibraltar Pound", "is_active": "No", "job_title": "Chif Technology Officer", "city": "Villasis "}, {"company": "Hooli", "state": "CA", "name": " Sara Smith", "currency": "GIP  ", "is_active": "TRUE", "job_title": " C.T.O.", "city": "puert del rosario"}, {"company": "Cyberdyne", "state": "mn", "name": "Eve Fischer", "currency": "Gel", "is_active": "No", "job_title": "Chief  Technology Officer", "city": "villasis"}, {"company": " Globex ", "state": " California ", "name": "   Omar Moore", "currency": "gel", "is_active": "T", "job_title": "prod manager", "city": "acacias"}, {"company": " Hooli ", "state": "CALIFORNIA", "name": "Bob Petrov", "currency": "HUF", "is_active": "FALSE", "job_title": " C.T.O.", "city": "Puerto Del Rosario   "}, {"company": " Soylent Corp", "state": "California", "name": "David Smith", "currency": " GEL ", "is_active": "F", "job_title": "pm", "city": "  Puerto Del Rosario"}, {"company": "  Hooli", "state": "CA", "name": "Lena Khan", "currency": "Gibraltar Pound", "is_active": "false", "job_title": "Prod Mgr", "city": "Puerto del Rosario"}, {"company": "Umbrella", "state": "MN  ", "name": "Carol Ali", "currency": "HUF", "is_active": "FALSE", "job_title": "prod mrnager", "city": "Accias"}, {"company": "Acme Inc", "state": "  CALIFORNIA", "name": "Eve  Johnson", "currency": "Gel", "is_active": "T", "job_title": "   pm", "city": "ACACIAS"}, {"company": "  Stark Industries", "state": "MN", "name": "Priya Smith", "currency": "GEL", "is_active": "F", "job_title": " Chief Technology Officer", "city": "   puerto del rosario"}, {"company": "Umbrella", "state": "california", "name": "   Priya Adams", "currency": "gel", "is_active": "true", "job_title": "Prod Manager ", "city": "villasis"}, {"company": "Soylent Corp  ", "state": "CA", "name": "Mona  Smith", "currency": " FORINT ", "is_active": "N", "job_title": "PM", "city": "ACACIAS"}, {"company": " Acme Inc", "state": " MN ", "name": "Sara Wong   ", "currency": "GIBRALTAR POUND", "is_active": "Y", "job_title": "PM", "city": "Villasis"}, {"company": "Cyberdyne", "state": "California", "name": "Omar Cruz   ", "currency": "Forint", "is_active": "TRUE", "job_title": "Prod Mgr", "city": "villasis"}, {"company": "Vehement", "state": "MN", "name": "Carol Wong", "currency": "Gibraltar Pound", "is_active": "FALSE", "job_title": " chief ttechnology officer", "city": "ACACSAS"}, {"company": "Hooli", "state": "Minnesota", "name": "David  Park", "currency": "Gel", "is_active": "1", "job_title": "Prod  Mgr", "city": "VILLASIS"}, {"company": " Acme Inc", "state": "mn", "name": "Mona Brandt", "currency": "GEL", "is_active": "FALSE", "job_title": "CTO", "city": "Acacias"}, {"company": "Initech", "state": "California", "name": "Grace  Khan", "currency": "Forint", "is_active": "F", "job_title": "Prod Manager", "city": "Puerto del Rosario"}, {"company": "Cyberdyne   ", "state": "CA", "name": "David Moore  ", "currency": "gibraltar pound", "is_active": "T", "job_title": "C.T.O.", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "ca", "name": "   Lena Khan", "currency": "GIBRALTAR POUND", "is_active": "0", "job_title": " cto ", "city": "Puerto del Rosario"}, {"company": " Stark Industries", "state": " califrnia", "name": "   Eve Wong", "currency": "Huf", "is_active": "N", "job_title": "PORD  MGR", "city": "VILLASIS "}, {"company": "Hooli ", "state": "MN", "name": " Bob Fischer", "currency": " GIP", "is_active": "N", "job_title": "  PRODUCT MANAGER", "city": "VILLASIS   "}, {"company": " Acme Inc", "state": "MN", "name": "Judy Petrov", "currency": "Gip", "is_active": "T", "job_title": "CHIEF TECHNOLOGY OFFICER", "city": "Acacias"}, {"company": "Stark  Industries", "state": "Minnesoat", "name": "Omar Fischer", "currency": "GEL", "is_active": "Yes", "job_title": "Product Manager", "city": "PUERTO DEL ROSARIO"}, {"company": "Stark Industries", "state": "mn", "name": "  Priya Park", "currency": "lari", "is_active": "true", "job_title": "Product Manager", "city": "Puerto del Rosairo"}, {"company": "Cyberdyne", "state": " Ca ", "name": "Grace Brandt", "currency": "Lari", "is_active": "Yes", "job_title": "PROD MANAGER", "city": "VILLAASIS"}, {"company": "Acme  Inc", "state": "  CA", "name": "Priya  Lee", "currency": "gel", "is_active": "0", "job_title": "CTO  ", "city": "villasis"}, {"company": "Stark  Industries", "state": " CA ", "name": "Alice Moore", "currency": "HUF", "is_active": "Y", "job_title": "CTO", "city": "Aaccias"}, {"company": " Acme Inc", "state": "ca", "name": "Sara Smith ", "currency": "GIBRALTAR POUND", "is_active": "N", "job_title": "PM", "city": "  Puerto Del Rosario"}, {"company": "Soylent Corp ", "state": "MN", "name": "Judy  Fischer", "currency": "lari", "is_active": "TRUE", "job_title": " PM ", "city": "puerto del rosario"}, {"company": "Wonka Co", "state": "   MN", "name": "Priya Khan", "currency": "HUF", "is_active": "true", "job_title": "Prod Mgr", "city": "Puerto del Rosario"}, {"company": "  Globex", "state": "ca", "name": "Karl  Lee", "currency": "Forint", "is_active": "true", "job_title": "Prod Manager ", "city": "PUERTO DEL ROSARIO"}, {"company": "Acme Inc", "state": " Minnesoota", "name": "Lena  Johnson", "currency": "Lari", "is_active": "Yes", "job_title": "chief technology officer   ", "city": "puerto del rosario "}, {"company": " Soylent Corp", "state": "California", "name": "David Smith", "currency": " GEL ", "is_active": "F", "job_title": "pm", "city": "  Puerto Del Rosario"}], "clean": [{"company": "Soylent Corp", "state": "California", "name": "Alice Smith", "currency": "HUF", "is_active": false, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Globex", "state": "Minnesota", "name": "Bob Diaz", "currency": "GIP", "is_active": false, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Cyberdyne", "state": "California", "name": "Carol Brandt", "currency": "GIP", "is_active": false, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Vehement", "state": "Minnesota", "name": "Grace Moore", "currency": "GIP", "is_active": false, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Wonka Co", "state": "Minnesota", "name": "Lena Reyes", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Stark Industries", "state": "Minnesota", "name": "Ivan Adams", "currency": "HUF", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "Minnesota", "name": "Bob Brandt", "currency": "GIP", "is_active": true, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Cyberdyne", "state": "Minnesota", "name": "Eve Moore", "currency": "GIP", "is_active": true, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Globex", "state": "Minnesota", "name": "Ivan Cruz", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "Minnesota", "name": "Judy Adams", "currency": "GEL", "is_active": false, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Globex", "state": "Minnesota", "name": "Ivan Wong", "currency": "GEL", "is_active": false, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Wonka Co", "state": "Minnesota", "name": "Judy Petrov", "currency": "HUF", "is_active": true, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Cyberdyne", "state": "California", "name": "Ivan Petrov", "currency": "GEL", "is_active": true, "job_title": "Chief Technology Officer", "city": "Villasis"}, {"company": "Umbrella", "state": "Minnesota", "name": "Ivan Petrov", "currency": "GIP", "is_active": true, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Vehement", "state": "California", "name": "Ivan Smith", "currency": "GIP", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Cyberdyne", "state": "Minnesota", "name": "Omar Smith", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Initech", "state": "Minnesota", "name": "Alice Adams", "currency": "GEL", "is_active": true, "job_title": "Chief Technology Officer", "city": "Villasis"}, {"company": "Umbrella", "state": "Minnesota", "name": "Sara Cruz", "currency": "GIP", "is_active": false, "job_title": "Chief Technology Officer", "city": "Villasis"}, {"company": "Hooli", "state": "California", "name": "Sara Smith", "currency": "GIP", "is_active": true, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Cyberdyne", "state": "Minnesota", "name": "Eve Fischer", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Villasis"}, {"company": "Globex", "state": "California", "name": "Omar Moore", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Hooli", "state": "California", "name": "Bob Petrov", "currency": "HUF", "is_active": false, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "California", "name": "David Smith", "currency": "GEL", "is_active": false, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Hooli", "state": "California", "name": "Lena Khan", "currency": "GIP", "is_active": false, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Umbrella", "state": "Minnesota", "name": "Carol Ali", "currency": "HUF", "is_active": false, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Acme Inc", "state": "California", "name": "Eve Johnson", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Stark Industries", "state": "Minnesota", "name": "Priya Smith", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Umbrella", "state": "California", "name": "Priya Adams", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Soylent Corp", "state": "California", "name": "Mona Smith", "currency": "HUF", "is_active": false, "job_title": "Product Manager", "city": "Acacias"}, {"company": "Acme Inc", "state": "Minnesota", "name": "Sara Wong", "currency": "GIP", "is_active": true, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Cyberdyne", "state": "California", "name": "Omar Cruz", "currency": "HUF", "is_active": true, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Vehement", "state": "Minnesota", "name": "Carol Wong", "currency": "GIP", "is_active": false, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Hooli", "state": "Minnesota", "name": "David Park", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Acme Inc", "state": "Minnesota", "name": "Mona Brandt", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Initech", "state": "California", "name": "Grace Khan", "currency": "HUF", "is_active": false, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Cyberdyne", "state": "California", "name": "David Moore", "currency": "GIP", "is_active": true, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "California", "name": "Lena Khan", "currency": "GIP", "is_active": false, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}, {"company": "Stark Industries", "state": "California", "name": "Eve Wong", "currency": "HUF", "is_active": false, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Hooli", "state": "Minnesota", "name": "Bob Fischer", "currency": "GIP", "is_active": false, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Acme Inc", "state": "Minnesota", "name": "Judy Petrov", "currency": "GIP", "is_active": true, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Stark Industries", "state": "Minnesota", "name": "Omar Fischer", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Stark Industries", "state": "Minnesota", "name": "Priya Park", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Cyberdyne", "state": "California", "name": "Grace Brandt", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Villasis"}, {"company": "Acme Inc", "state": "California", "name": "Priya Lee", "currency": "GEL", "is_active": false, "job_title": "Chief Technology Officer", "city": "Villasis"}, {"company": "Stark Industries", "state": "California", "name": "Alice Moore", "currency": "HUF", "is_active": true, "job_title": "Chief Technology Officer", "city": "Acacias"}, {"company": "Acme Inc", "state": "California", "name": "Sara Smith", "currency": "GIP", "is_active": false, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Soylent Corp", "state": "Minnesota", "name": "Judy Fischer", "currency": "GEL", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Wonka Co", "state": "Minnesota", "name": "Priya Khan", "currency": "HUF", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Globex", "state": "California", "name": "Karl Lee", "currency": "HUF", "is_active": true, "job_title": "Product Manager", "city": "Puerto del Rosario"}, {"company": "Acme Inc", "state": "Minnesota", "name": "Lena Johnson", "currency": "GEL", "is_active": true, "job_title": "Chief Technology Officer", "city": "Puerto del Rosario"}], "dirty_cols": ["company", "state", "name", "currency", "is_active", "job_title", "city"], "clean_cols": ["company", "state", "name", "currency", "is_active", "job_title", "city"], "plan": {"dataset_summary": "51 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CA": "California", "MN": "Minnesota", "mn": "Minnesota", "MINNESOTA": "Minnesota", "Minnelota": "Minnesota", "Ca": "California", "CALIFORNIA": "California", "california": "California", "ca": "California", "califrnia": "California", "Minnesoat": "Minnesota", "Minnesoota": "Minnesota"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Forint": "HUF", "gibraltar pound": "GIP", "Gibraltar Pound": "GIP", "Lari": "GEL", "LARI": "GEL", "lari": "GEL", "gel": "GEL", "huf": "HUF", "Gel": "GEL", "FORINT": "HUF", "GIBRALTAR POUND": "GIP", "Huf": "HUF", "Gip": "GIP"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pm": "Product Manager", "Prod Manager": "Product Manager", "CTO": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "PM": "Product Manager", "Product Mangaer": "Product Manager", "C.T.O.": "Chief Technology Officer", "Prod Mgr": "Product Manager", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "Chif Technology Officer": "Chief Technology Officer", "prod manager": "Product Manager", "prod mrnager": "Product Manager", "chief ttechnology officer": "Chief Technology Officer", "cto": "Chief Technology Officer", "PORD MGR": "Product Manager", "PRODUCT MANAGER": "Product Manager", "PROD MANAGER": "Product Manager"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"acacias": "Acacias", "ACACIAS": "Acacias", "PUERTO DEL ROSARIO": "Puerto del Rosario", "VILLASIS": "Villasis", "villasis": "Villasis", "puerto del rosario": "Puerto del Rosario", "Accias": "Acacias", "puert del rosario": "Puerto del Rosario", "Puerto Del Rosario": "Puerto del Rosario", "ACACSAS": "Acacias", "Puerto del Rosairo": "Puerto del Rosario", "VILLAASIS": "Villasis", "Aaccias": "Acacias"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "531-675-6167", "job_title": "pm", "currency": "Cuan Peso", "department": "IT", "company": "Stark Industries", "city": "yicheng"}, {"phone": "5031808705", "job_title": "Engineering Manager", "currency": "BYN", "department": "I..T", "company": "Soylent  Corp", "city": "Rendsburg"}, {"phone": "548-914-6597", "job_title": "C.E.O.", "currency": "cup", "department": "Fin", "company": "  Soylent Corp", "city": "Yicheng"}, {"phone": "542-574-1631", "job_title": "CHIEF EXECUTIVE OFFICER", "currency": "Belarusian Ruble", "department": "Fin", "company": "Hooli", "city": "rendsburg   "}, {"phone": "(597)475-2976", "job_title": "ENG MGR", "currency": "CUP", "department": " finance", "company": "Acme Inc   ", "city": "CHANGYI   "}, {"phone": "596-162-5342", "job_title": "Chief Executive Officer", "currency": "CUP", "department": "it", "company": "Cyberdyne  ", "city": " yicheng "}, {"phone": "575-794-6313", "job_title": "Engineering Mgr", "currency": "CUBANN PESO", "department": "It  ", "company": "Soylent  Corp", "city": "Rewdsburg"}, {"phone": "(571)089-3710", "job_title": "PROD MGR", "currency": "belarusian ruble", "department": "IT", "company": "Wonka Co", "city": "YICHENG"}, {"phone": "576-074-1104", "job_title": "ENG MANAGER", "currency": "Cuban Peso", "department": "it", "company": "Vehement", "city": "yicheng"}, {"phone": "(534)419-1478", "job_title": "   Prod Mgr", "currency": "Byn", "department": "Information  Technology", "company": "Cyberdyne", "city": "CHANGYI"}, {"phone": "551.948.0048", "job_title": "Product Manager", "currency": "  CUP", "department": "INFORMATION TECHNOLOGY", "company": "Hooli", "city": "rendsburg"}, {"phone": "5488317444", "job_title": " C.E.O. ", "currency": "CUBAN EPSO", "department": "Information Technology", "company": "Acme Inc", "city": "rendsburg"}, {"phone": "5341766126", "job_title": "Admin Asst", "currency": "Belarusian Ruble  ", "department": "Finance", "company": "Wonka Co", "city": "Rendsburg"}, {"phone": "567.835.5302", "job_title": "Admin Asst", "currency": "BYN", "department": "Finance", "company": " Vehement ", "city": "changyi"}, {"phone": "560.667.4145", "job_title": "Eng  Mgr", "currency": "Cuban Peso", "department": "Information Technology", "company": "Cyberdyne", "city": "Yicheng"}], "clean": [{"phone": "(531) 675-6167", "job_title": "Product Manager", "currency": "CUP", "department": "Information Technology", "company": "Stark Industries", "city": "Yicheng"}, {"phone": "(503) 180-8705", "job_title": "Engineering Manager", "currency": "BYN", "department": "Information Technology", "company": "Soylent Corp", "city": "Rendsburg"}, {"phone": "(548) 914-6597", "job_title": "Chief Executive Officer", "currency": "CUP", "department": "Finance", "company": "Soylent Corp", "city": "Yicheng"}, {"phone": "(542) 574-1631", "job_title": "Chief Executive Officer", "currency": "BYN", "department": "Finance", "company": "Hooli", "city": "Rendsburg"}, {"phone": "(597) 475-2976", "job_title": "Engineering Manager", "currency": "CUP", "department": "Finance", "company": "Acme Inc", "city": "Changyi"}, {"phone": "(596) 162-5342", "job_title": "Chief Executive Officer", "currency": "CUP", "department": "Information Technology", "company": "Cyberdyne", "city": "Yicheng"}, {"phone": "(575) 794-6313", "job_title": "Engineering Manager", "currency": "CUP", "department": "Information Technology", "company": "Soylent Corp", "city": "Rendsburg"}, {"phone": "(571) 089-3710", "job_title": "Product Manager", "currency": "BYN", "department": "Information Technology", "company": "Wonka Co", "city": "Yicheng"}, {"phone": "(576) 074-1104", "job_title": "Engineering Manager", "currency": "CUP", "department": "Information Technology", "company": "Vehement", "city": "Yicheng"}, {"phone": "(534) 419-1478", "job_title": "Product Manager", "currency": "BYN", "department": "Information Technology", "company": "Cyberdyne", "city": "Changyi"}, {"phone": "(551) 948-0048", "job_title": "Product Manager", "currency": "CUP", "department": "Information Technology", "company": "Hooli", "city": "Rendsburg"}, {"phone": "(548) 831-7444", "job_title": "Chief Executive Officer", "currency": "CUP", "department": "Information Technology", "company": "Acme Inc", "city": "Rendsburg"}, {"phone": "(534) 176-6126", "job_title": "Administrative Assistant", "currency": "BYN", "department": "Finance", "company": "Wonka Co", "city": "Rendsburg"}, {"phone": "(567) 835-5302", "job_title": "Administrative Assistant", "currency": "BYN", "department": "Finance", "company": "Vehement", "city": "Changyi"}, {"phone": "(560) 667-4145", "job_title": "Engineering Manager", "currency": "CUP", "department": "Information Technology", "company": "Cyberdyne", "city": "Yicheng"}], "dirty_cols": ["phone", "job_title", "currency", "department", "company", "city"], "clean_cols": ["phone", "job_title", "currency", "department", "company", "city"], "plan": {"dataset_summary": "15 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pm": "Product Manager", "C.E.O.": "Chief Executive Officer", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "ENG MGR": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "PROD MGR": "Product Manager", "ENG MANAGER": "Engineering Manager", "Prod Mgr": "Product Manager", "Admin Asst": "Administrative Assistant", "Eng Mgr": "Engineering Manager"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cuan Peso": "CUP", "cup": "CUP", "Belarusian Ruble": "BYN", "CUBANN PESO": "CUP", "belarusian ruble": "BYN", "Cuban Peso": "CUP", "Byn": "BYN", "CUBAN EPSO": "CUP"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "I..T": "Information Technology", "Fin": "Finance", "finance": "Finance", "it": "Information Technology", "It": "Information Technology", "INFORMATION TECHNOLOGY": "Information Technology"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"yicheng": "Yicheng", "rendsburg": "Rendsburg", "CHANGYI": "Changyi", "Rewdsburg": "Rendsburg", "YICHENG": "Yicheng", "changyi": "Changyi"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "CRESPO", "department": "Legal", "state": "az", "signup_date": "2023-08-18"}, {"city": " waldkirch ", "department": "LEGAL & COMPLIANCE", "state": "Nevada", "signup_date": "45059"}, {"city": "Crespo", "department": "Legal & Compliance", "state": "Nevaya   ", "signup_date": "45022"}, {"city": "  middelkerke", "department": "finance", "state": "Arizona", "signup_date": "25 Jul 2023"}, {"city": "WADLKIRCH", "department": "LEGAL", "state": "az", "signup_date": "2023-11-14"}, {"city": "MIDDELKERKE", "department": "Legal & Compliance", "state": "NV", "signup_date": "1/19/2023"}, {"city": "Waldkirch", "department": "Fin", "state": "AZ", "signup_date": "4/25/2023"}, {"city": "WALDKIRCH", "department": "legal", "state": "az", "signup_date": "2023-01-04"}, {"city": " middelkerke ", "department": "legal  ", "state": " AZ ", "signup_date": "4/12/2023"}, {"city": "middelkerke", "department": "Finance", "state": "NV", "signup_date": "2023-05-22"}, {"city": "WALDKIRCH", "department": "  Legal", "state": "nv", "signup_date": "1/27/2023"}, {"city": "Waldkirch", "department": " LEGAL", "state": "nv", "signup_date": "45204"}], "clean": [{"city": "Crespo", "department": "Legal", "state": "Arizona", "signup_date": "2023-08-18"}, {"city": "Waldkirch", "department": "Legal", "state": "Nevada", "signup_date": "2023-05-13"}, {"city": "Crespo", "department": "Legal", "state": "Nevada", "signup_date": "2023-04-06"}, {"city": "Middelkerke", "department": "Finance", "state": "Arizona", "signup_date": "2023-07-25"}, {"city": "Waldkirch", "department": "Legal", "state": "Arizona", "signup_date": "2023-11-14"}, {"city": "Middelkerke", "department": "Legal", "state": "Nevada", "signup_date": "2023-01-19"}, {"city": "Waldkirch", "department": "Finance", "state": "Arizona", "signup_date": "2023-04-25"}, {"city": "Waldkirch", "department": "Legal", "state": "Arizona", "signup_date": "2023-01-04"}, {"city": "Middelkerke", "department": "Legal", "state": "Arizona", "signup_date": "2023-04-12"}, {"city": "Middelkerke", "department": "Finance", "state": "Nevada", "signup_date": "2023-05-22"}, {"city": "Waldkirch", "department": "Legal", "state": "Nevada", "signup_date": "2023-01-27"}, {"city": "Waldkirch", "department": "Legal", "state": "Nevada", "signup_date": "2023-10-05"}], "dirty_cols": ["city", "department", "state", "signup_date"], "clean_cols": ["city", "department", "state", "signup_date"], "plan": {"dataset_summary": "12 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CRESPO": "Crespo", "waldkirch": "Waldkirch", "middelkerke": "Middelkerke", "WADLKIRCH": "Waldkirch", "MIDDELKERKE": "Middelkerke", "WALDKIRCH": "Waldkirch"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LEGAL & COMPLIANCE": "Legal", "Legal & Compliance": "Legal", "finance": "Finance", "LEGAL": "Legal", "Fin": "Finance", "legal": "Legal"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"az": "Arizona", "Nevaya": "Nevada", "NV": "Nevada", "AZ": "Arizona", "nv": "Nevada"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"job_title": "Admin Asst", "phone": "5927460377", "department": " marketing ", "state": "Maryland"}, {"job_title": "c.t.o.", "phone": "568.339.7815", "department": "PEOPLE OPS", "state": "MI"}, {"job_title": "chief technology officer", "phone": "581.222.1490", "department": "human resources", "state": "   MD"}, {"job_title": "ADMINA SST", "phone": "(555)650-7824", "department": "Marketing", "state": "MD"}, {"job_title": "Admin Asst", "phone": "5656815950", "department": "growth", "state": "michigan"}, {"job_title": "C.T.O.", "phone": "(561)522-3323", "department": "peple", "state": "AZ"}, {"job_title": "senior engineer", "phone": "5729069975", "department": "Marketing   ", "state": "GA"}, {"job_title": "amdin", "phone": "523-639-5269", "department": "mkt", "state": "Maryland"}, {"job_title": "SR ENGINEER", "phone": "567-014-7522", "department": "Peopel  ", "state": "Maryland"}, {"job_title": "Administrative Assistant", "phone": "502.830.8423", "department": "Hr", "state": "Az"}, {"job_title": " CTO ", "phone": "(564)247-9543", "department": "   People Ops", "state": "georgia"}, {"job_title": "C.T.O.", "phone": "(541)584-6299", "department": "eng", "state": " Georgia "}, {"job_title": "C.T.O.", "phone": "(550)372-2604", "department": "Mkt", "state": "Michigan"}, {"job_title": "  admin assistant", "phone": "(599)672-9877", "department": "Peopfe Ops", "state": "Michigan"}, {"job_title": "Administrative Assistant", "phone": "521-243-6726", "department": "Marketing", "state": "Maryland "}, {"job_title": "CTO", "phone": "517-724-5833", "department": "HR", "state": "georgia"}, {"job_title": "admin  asst", "phone": "5633979440", "department": "Finance", "state": "   ARIZONA"}, {"job_title": "Admin Asst", "phone": "(594)304-0055", "department": "R&D", "state": "Georgia"}, {"job_title": "Addmin", "phone": "5929162828", "department": "PEOPLE   ", "state": " Az"}], "clean": [{"job_title": "Administrative Assistant", "phone": "(592) 746-0377", "department": "Marketing", "state": "Maryland"}, {"job_title": "Chief Technology Officer", "phone": "(568) 339-7815", "department": "Human Resources", "state": "Michigan"}, {"job_title": "Chief Technology Officer", "phone": "(581) 222-1490", "department": "Human Resources", "state": "Maryland"}, {"job_title": "Administrative Assistant", "phone": "(555) 650-7824", "department": "Marketing", "state": "Maryland"}, {"job_title": "Administrative Assistant", "phone": "(565) 681-5950", "department": "Marketing", "state": "Michigan"}, {"job_title": "Chief Technology Officer", "phone": "(561) 522-3323", "department": "Human Resources", "state": "Arizona"}, {"job_title": "Senior Engineer", "phone": "(572) 906-9975", "department": "Marketing", "state": "Georgia"}, {"job_title": "Administrative Assistant", "phone": "(523) 639-5269", "department": "Marketing", "state": "Maryland"}, {"job_title": "Senior Engineer", "phone": "(567) 014-7522", "department": "Human Resources", "state": "Maryland"}, {"job_title": "Administrative Assistant", "phone": "(502) 830-8423", "department": "Human Resources", "state": "Arizona"}, {"job_title": "Chief Technology Officer", "phone": "(564) 247-9543", "department": "Human Resources", "state": "Georgia"}, {"job_title": "Chief Technology Officer", "phone": "(541) 584-6299", "department": "Engineering", "state": "Georgia"}, {"job_title": "Chief Technology Officer", "phone": "(550) 372-2604", "department": "Marketing", "state": "Michigan"}, {"job_title": "Administrative Assistant", "phone": "(599) 672-9877", "department": "Human Resources", "state": "Michigan"}, {"job_title": "Administrative Assistant", "phone": "(521) 243-6726", "department": "Marketing", "state": "Maryland"}, {"job_title": "Chief Technology Officer", "phone": "(517) 724-5833", "department": "Human Resources", "state": "Georgia"}, {"job_title": "Administrative Assistant", "phone": "(563) 397-9440", "department": "Finance", "state": "Arizona"}, {"job_title": "Administrative Assistant", "phone": "(594) 304-0055", "department": "Engineering", "state": "Georgia"}, {"job_title": "Administrative Assistant", "phone": "(592) 916-2828", "department": "Human Resources", "state": "Arizona"}], "dirty_cols": ["job_title", "phone", "department", "state"], "clean_cols": ["job_title", "phone", "department", "state"], "plan": {"dataset_summary": "19 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Admin Asst": "Administrative Assistant", "c.t.o.": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "ADMINA SST": "Administrative Assistant", "C.T.O.": "Chief Technology Officer", "senior engineer": "Senior Engineer", "amdin": "Administrative Assistant", "SR ENGINEER": "Senior Engineer", "CTO": "Chief Technology Officer", "admin assistant": "Administrative Assistant", "admin asst": "Administrative Assistant", "Addmin": "Administrative Assistant"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"marketing": "Marketing", "PEOPLE OPS": "Human Resources", "human resources": "Human Resources", "growth": "Marketing", "peple": "Human Resources", "mkt": "Marketing", "Peopel": "Human Resources", "Hr": "Human Resources", "People Ops": "Human Resources", "eng": "Engineering", "Mkt": "Marketing", "Peopfe Ops": "Human Resources", "HR": "Human Resources", "R&D": "Engineering", "PEOPLE": "Human Resources"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MI": "Michigan", "MD": "Maryland", "michigan": "Michigan", "AZ": "Arizona", "GA": "Georgia", "Az": "Arizona", "georgia": "Georgia", "ARIZONA": "Arizona"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "8.929,87", "country": "LIECHTENSTEIN", "city": "Nola", "extra": ""}, {"amount": "?", "country": "Mk", "city": "New Orleans", "extra": ""}, {"amount": "$3,275.43", "country": "Notrh Macedonia", "city": " Punwgol ", "extra": ""}, {"amount": "4.667,16", "country": "PAN  ", "city": "maragogi", "extra": ""}, {"amount": "4.704,83", "country": "Lie", "city": "NOLA", "extra": ""}, {"amount": "8.923,38", "country": "MKD", "city": " Maragogi ", "extra": ""}, {"amount": "$6,450.25", "country": "  North Macedonia", "city": "Maraggogi", "extra": ""}, {"amount": "3.269,95", "country": "Mkd", "city": "New Orleans", "extra": ""}, {"amount": "937,81", "country": "   RS", "city": " Solingen", "extra": ""}, {"amount": "270,13", "country": "Republic  Of North Macedonia", "city": "MARAGOGI   ", "extra": ""}, {"amount": "$2,188.83", "country": "Principality Of  Liechtenstein", "city": "  MARAGOGI", "extra": ""}, {"amount": "8.890,87", "country": "rs", "city": "New Orleans   ", "extra": ""}, {"amount": "1.433,75", "country": "panama", "city": " Solingen", "extra": ""}, {"amount": "270,13", "country": "Republic  Of North Macedonia", "city": "MARAGOGI   ", "extra": ""}, {"amount": "", "country": "", "city": "", "extra": ""}], "clean": [{"amount": 8929.87, "country": "Liechtenstein", "city": "New Orleans"}, {"amount": NaN, "country": "North Macedonia", "city": "New Orleans"}, {"amount": 3275.43, "country": "North Macedonia", "city": "Punggol"}, {"amount": 4667.16, "country": "Panama", "city": "Maragogi"}, {"amount": 4704.83, "country": "Liechtenstein", "city": "New Orleans"}, {"amount": 8923.38, "country": "North Macedonia", "city": "Maragogi"}, {"amount": 6450.25, "country": "North Macedonia", "city": "Maragogi"}, {"amount": 3269.95, "country": "North Macedonia", "city": "New Orleans"}, {"amount": 937.81, "country": "Serbia", "city": "Solingen"}, {"amount": 270.13, "country": "North Macedonia", "city": "Maragogi"}, {"amount": 2188.83, "country": "Liechtenstein", "city": "Maragogi"}, {"amount": 8890.87, "country": "Serbia", "city": "New Orleans"}, {"amount": 1433.75, "country": "Panama", "city": "Solingen"}], "dirty_cols": ["amount", "country", "city", "extra"], "clean_cols": ["amount", "country", "city"], "plan": {"dataset_summary": "15 rows × 4 columns. 3 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LIECHTENSTEIN": "Liechtenstein", "Mk": "North Macedonia", "Notrh Macedonia": "North Macedonia", "PAN": "Panama", "Lie": "Liechtenstein", "MKD": "North Macedonia", "Mkd": "North Macedonia", "RS": "Serbia", "Republic Of North Macedonia": "North Macedonia", "Principality Of Liechtenstein": "Liechtenstein", "rs": "Serbia", "panama": "Panama"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Nola": "New Orleans", "Punwgol": "Punggol", "maragogi": "Maragogi", "NOLA": "New Orleans", "Maraggogi": "Maragogi", "MARAGOGI": "Maragogi"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "NR", "status": "PAID", "industry": "FinServ", "city": "Taquari", "name": "   Grace Park", "job_title": "SALES REP", "company": " Initech "}, {"country": "Nauru", "status": "refunded", "industry": " ENERGY ", "city": "Taquari", "name": "  Lena Novak", "job_title": "prod manager", "company": "Umbrella"}, {"country": "nr", "status": "pakd", "industry": "FINSREV", "city": "Kasugnu", "name": "Eve Fischer", "job_title": "Rep", "company": " Umbrella "}, {"country": "Republic Of Malta", "status": "laet", "industry": "Financial Services", "city": "Taquari", "name": "Priya Smith   ", "job_title": "prdduct manager", "company": " Hooli"}, {"country": "UKR", "status": " pending ", "industry": "energy", "city": "  Kasungu", "name": "   Grace Diaz", "job_title": "PM", "company": " Vehement "}, {"country": "Mlt", "status": "due", "industry": "Financial Services ", "city": "Rio Bonito  ", "name": "   Frank Cruz", "job_title": "Sales Representative", "company": "Stark Industries"}, {"country": "ukr", "status": "Refunded", "industry": "finsrv", "city": "Rio Bonito", "name": "Ivan Park  ", "job_title": "PROD MANAGER", "company": "Initech"}, {"country": "FEDERAL REPUBLIC OF SOMALIA", "status": "late  ", "industry": "utilities", "city": "Santa Teresa", "name": "Ivan Adams", "job_title": "sales rep", "company": "Acme Inc  "}, {"country": "pk", "status": "unpaid", "industry": "energy", "city": "santa teresa", "name": "Ivan Diaz", "job_title": "Prod Mgr", "company": "Cyberdyne"}, {"country": " NR", "status": "REFUNDED", "industry": "Banking", "city": "taquari   ", "name": "Judy Petrov", "job_title": "sales rep", "company": "Vehement"}, {"country": "   PK", "status": "DUE", "industry": "finserv", "city": "rio bonito", "name": "Lena Johnson", "job_title": "Sales Rep", "company": " Stark Industries"}, {"country": "NRU", "status": "overdue", "industry": "Fintech", "city": "snata  teresa", "name": "Priya Fischer", "job_title": "  sales representative", "company": "Wonka Co"}, {"country": "republic of nauru", "status": "Refunded", "industry": "energy", "city": "Santa Teresa", "name": "Bob Brandt", "job_title": "Prod Manager", "company": "Cyberdyne"}, {"country": "  MLT", "status": "  OVERDUE", "industry": "banking  ", "city": "Kasungu", "name": "David Cruz", "job_title": "Rep", "company": "Vehement   "}, {"country": "ua", "status": "refund", "industry": "Energy", "city": "Tqauari", "name": "Lena  Reyes", "job_title": "PM", "company": "Wonka Co "}, {"country": "MT", "status": "PAID", "industry": "Utilities", "city": "   Rio Bonito", "name": "Alice Adams", "job_title": "product manager", "company": "Soylent Corp"}, {"country": "MT", "status": " late ", "industry": "utilities", "city": "SANTA TERESA", "name": "Grace  Cruz", "job_title": "sales rep", "company": "Cyberdyne"}, {"country": "  ua", "status": "due", "industry": "FinServ", "city": "Taquai", "name": "Judy Brandt  ", "job_title": "Sales Representative", "company": "   Soylent Corp"}, {"country": "   Federal Republic of Somalia", "status": "Overdue", "industry": "Utilities", "city": "  Kasungu", "name": " Frank Brandt", "job_title": "Rep", "company": " Acme Inc"}, {"country": "somalia", "status": "Pending", "industry": "   Oil & Gas", "city": "TAQQUARI", "name": "Ivan Reyes", "job_title": " PRDO MANAGER", "company": "Vehement  "}, {"country": "malta", "status": "Refunded", "industry": "fnitech", "city": "Taquari   ", "name": "Karl Lee", "job_title": "Prod Manager", "company": "Hooli"}, {"country": "Nauru", "status": "   Refuned", "industry": "energy", "city": "kasvngu", "name": "Judy Novak", "job_title": "Sales Representative", "company": "Initech"}, {"country": "ukr", "status": " PAID ", "industry": "Fintech", "city": "Santa Tersa", "name": "Ivan Ali", "job_title": " prod manager", "company": "Hooli"}, {"country": "FEDERAL REPUBLIC OF SOMALIA", "status": "unaid", "industry": "Finance", "city": "Taquari", "name": "Mona Adams", "job_title": "PM", "company": "Cyberdyne"}, {"country": "ukraine", "status": "Pending", "industry": "FINANCE", "city": "Kasungu  ", "name": "   Omar Brandt", "job_title": "Prxd Mgr", "company": "  Umbrella"}, {"country": "Republic  of Malta", "status": "Refunded", "industry": "Financial Services", "city": "santa teresa", "name": " Frank Novak", "job_title": " Prod Manager", "company": " Globex"}, {"country": "UA", "status": " refunded ", "industry": "Finance", "city": "rio bonito", "name": " Ivan Adams", "job_title": "rep", "company": "Hooli"}, {"country": "nauru", "status": "Pending ", "industry": "energy", "city": "Santa Teresa", "name": "Frank Smith", "job_title": "Sales Rep", "company": "   Globex"}, {"country": "PAK", "status": "PAID   ", "industry": "energy", "city": " Kasungu", "name": " Omar Brandt", "job_title": "SALES REP", "company": "Stark Industries"}, {"country": "UKR", "status": "pending", "industry": "Banking", "city": "santa teresa", "name": "Alice  Khan", "job_title": "Product Manager", "company": "Soylent Corp   "}, {"country": " pk ", "status": "Paid", "industry": "BANKING", "city": "taquari", "name": "Grace Petrov", "job_title": "Product anager   ", "company": "Acme  Inc"}, {"country": "Republic of Nauru", "status": " late ", "industry": "finsserv", "city": "santa teresa", "name": "Carol Park", "job_title": "Prod aMnager", "company": "Initech  "}, {"country": "somalia", "status": "Overdue", "industry": "FinServ", "city": "rio konito", "name": "Bob  Park", "job_title": "Product Manager", "company": "Initech   "}, {"country": "Somalta", "status": "REFUNDED", "industry": "Finance", "city": " kasungu ", "name": "Lena Johnson", "job_title": "Sales Reprzsentative", "company": "Acme Inc"}, {"country": "mt", "status": " Paid", "industry": "Energy", "city": "TAQUARI", "name": "David  Khan", "job_title": "Prod Mgr", "company": "Hooli"}, {"country": " mlt ", "status": "Overdue", "industry": " Banking", "city": "kasungu", "name": "Heidi Brandt", "job_title": "Pm", "company": "Stark Industries"}, {"country": "Malta", "status": "PAID", "industry": "banking", "city": "Taquari", "name": "Frank Novak   ", "job_title": "Prod Manager", "company": "Initech"}, {"country": "   mt", "status": "Refund", "industry": "fintech", "city": "Taquari", "name": "   David Ali", "job_title": "Sales Rep", "company": " Globex"}, {"country": "UA", "status": "overde", "industry": "Banking", "city": "SANTA  TERESA", "name": "Eve Diaz   ", "job_title": "SALES REP", "company": " Cyberdyne "}, {"country": "pakistan", "status": "OVERDUE", "industry": "Energy", "city": "santa teresa", "name": "Lena Petrov", "job_title": "prod mgr", "company": "Globex"}, {"country": " pak ", "status": "paiid  ", "industry": "Fintech", "city": "rio bonito", "name": "Priya Novak", "job_title": "rep", "company": "Acme Inc   "}, {"country": "somalia ", "status": "pending", "industry": " Energy", "city": " Taquari ", "name": "Heidi Wong", "job_title": "   Sales Representative", "company": "Cyberdyne"}, {"country": "  PK", "status": "Late", "industry": "Financial  Services", "city": "Taquari", "name": " Ivan Park", "job_title": " PROD MANAGER", "company": "Stark Industries"}, {"country": "MLT  ", "status": "Paid   ", "industry": "ENERGY", "city": "santa teresa", "name": "Lena Petrov", "job_title": "   Prod Manager", "company": "Hooli"}, {"country": "so", "status": " refunded ", "industry": "energy", "city": "Kasungu", "name": "   Eve Smith", "job_title": "prod mgr", "company": "  Wonka Co"}, {"country": "SO", "status": "paid", "industry": "Banking", "city": "Kasungu", "name": "Frank Novak", "job_title": "REP   ", "company": "Soylent Corp  "}, {"country": "SOM", "status": "piad ", "industry": "Oil & Gas   ", "city": "ksaungu", "name": "Sara Ali", "job_title": "PM", "company": "Soylent Corp"}, {"country": "Ukr", "status": "PAID", "industry": "finance", "city": "   SANTA TERESA", "name": "Lena Moore", "job_title": "SALES REP", "company": " Vehement "}, {"country": "ua", "status": "overdue", "industry": "   fiannce", "city": "Rio Bonito", "name": "Bob Cruz", "job_title": "sales rep", "company": " Globex "}, {"country": "Ua", "status": "pending", "industry": "Oil & Gas", "city": "taquari", "name": "Karl Moore   ", "job_title": "Rep", "company": "Acme Inc"}, {"country": " SOM ", "status": "refunded", "industry": " utilities", "city": "Kasungu", "name": " Carol Novak", "job_title": "Sales Rep", "company": "Acme Inc"}, {"country": "   ukr", "status": "Overdue", "industry": "energy", "city": "Rio Bonito", "name": "Eve Ali ", "job_title": "SALES REP", "company": "Soylent Corp"}, {"country": " Somalia ", "status": "paid", "industry": "   fintech", "city": "santa teresa", "name": "Judy Ali", "job_title": "Prod Manager", "company": "Acme Inc"}, {"country": "ukr", "status": "PAID", "industry": "Utilities", "city": "Santa Teresa", "name": "Grace Fischer", "job_title": "  sales rep", "company": "   Stark Industries"}, {"country": "", "status": "", "industry": "", "city": "", "name": "", "job_title": "", "company": ""}], "clean": [{"country": "Nauru", "status": "Paid", "industry": "Financial Services", "city": "Taquari", "name": "Grace Park", "job_title": "Sales Representative", "company": "Initech"}, {"country": "Nauru", "status": "Refunded", "industry": "Energy", "city": "Taquari", "name": "Lena Novak", "job_title": "Product Manager", "company": "Umbrella"}, {"country": "Nauru", "status": "Paid", "industry": "Financial Services", "city": "Kasungu", "name": "Eve Fischer", "job_title": "Sales Representative", "company": "Umbrella"}, {"country": "Malta", "status": "Overdue", "industry": "Financial Services", "city": "Taquari", "name": "Priya Smith", "job_title": "Product Manager", "company": "Hooli"}, {"country": "Ukraine", "status": "Pending", "industry": "Energy", "city": "Kasungu", "name": "Grace Diaz", "job_title": "Product Manager", "company": "Vehement"}, {"country": "Malta", "status": "Pending", "industry": "Financial Services", "city": "Rio Bonito", "name": "Frank Cruz", "job_title": "Sales Representative", "company": "Stark Industries"}, {"country": "Ukraine", "status": "Refunded", "industry": "Financial Services", "city": "Rio Bonito", "name": "Ivan Park", "job_title": "Product Manager", "company": "Initech"}, {"country": "Somalia", "status": "Overdue", "industry": "Energy", "city": "Santa Teresa", "name": "Ivan Adams", "job_title": "Sales Representative", "company": "Acme Inc"}, {"country": "Pakistan", "status": "Pending", "industry": "Energy", "city": "Santa Teresa", "name": "Ivan Diaz", "job_title": "Product Manager", "company": "Cyberdyne"}, {"country": "Nauru", "status": "Refunded", "industry": "Financial Services", "city": "Taquari", "name": "Judy Petrov", "job_title": "Sales Representative", "company": "Vehement"}, {"country": "Pakistan", "status": "Pending", "industry": "Financial Services", "city": "Rio Bonito", "name": "Lena Johnson", "job_title": "Sales Representative", "company": "Stark Industries"}, {"country": "Nauru", "status": "Overdue", "industry": "Financial Services", "city": "Santa Teresa", "name": "Priya Fischer", "job_title": "Sales Representative", "company": "Wonka Co"}, {"country": "Nauru", "status": "Refunded", "industry": "Energy", "city": "Santa Teresa", "name": "Bob Brandt", "job_title": "Product Manager", "company": "Cyberdyne"}, {"country": "Malta", "status": "Overdue", "industry": "Financial Services", "city": "Kasungu", "name": "David Cruz", "job_title": "Sales Representative", "company": "Vehement"}, {"country": "Ukraine", "status": "Refunded", "industry": "Energy", "city": "Taquari", "name": "Lena Reyes", "job_title": "Product Manager", "company": "Wonka Co"}, {"country": "Malta", "status": "Paid", "industry": "Energy", "city": "Rio Bonito", "name": "Alice Adams", "job_title": "Product Manager", "company": "Soylent Corp"}, {"country": "Malta", "status": "Overdue", "industry": "Energy", "city": "Santa Teresa", "name": "Grace Cruz", "job_title": "Sales Representative", "company": "Cyberdyne"}, {"country": "Ukraine", "status": "Pending", "industry": "Financial Services", "city": "Taquari", "name": "Judy Brandt", "job_title": "Sales Representative", "company": "Soylent Corp"}, {"country": "Somalia", "status": "Overdue", "industry": "Energy", "city": "Kasungu", "name": "Frank Brandt", "job_title": "Sales Representative", "company": "Acme Inc"}, {"country": "Somalia", "status": "Pending", "industry": "Energy", "city": "Taquari", "name": "Ivan Reyes", "job_title": "Product Manager", "company": "Vehement"}, {"country": "Malta", "status": "Refunded", "industry": "Financial Services", "city": "Taquari", "name": "Karl Lee", "job_title": "Product Manager", "company": "Hooli"}, {"country": "Nauru", "status": "Refunded", "industry": "Energy", "city": "Kasungu", "name": "Judy Novak", "job_title": "Sales Representative", "company": "Initech"}, {"country": "Ukraine", "status": "Paid", "industry": "Financial Services", "city": "Santa Teresa", "name": "Ivan Ali", "job_title": "Product Manager", "company": "Hooli"}, {"country": "Somalia", "status": "Pending", "industry": "Financial Services", "city": "Taquari", "name": "Mona Adams", "job_title": "Product Manager", "company": "Cyberdyne"}, {"country": "Ukraine", "status": "Pending", "industry": "Financial Services", "city": "Kasungu", "name": "Omar Brandt", "job_title": "Product Manager", "company": "Umbrella"}, {"country": "Malta", "status": "Refunded", "industry": "Financial Services", "city": "Santa Teresa", "name": "Frank Novak", "job_title": "Product Manager", "company": "Globex"}, {"country": "Ukraine", "status": "Refunded", "industry": "Financial Services", "city": "Rio Bonito", "name": "Ivan Adams", "job_title": "Sales Representative", "company": "Hooli"}, {"country": "Nauru", "status": "Pending", "industry": "Energy", "city": "Santa Teresa", "name": "Frank Smith", "job_title": "Sales Representative", "company": "Globex"}, {"country": "Pakistan", "status": "Paid", "industry": "Energy", "city": "Kasungu", "name": "Omar Brandt", "job_title": "Sales Representative", "company": "Stark Industries"}, {"country": "Ukraine", "status": "Pending", "industry": "Financial Services", "city": "Santa Teresa", "name": "Alice Khan", "job_title": "Product Manager", "company": "Soylent Corp"}, {"country": "Pakistan", "status": "Paid", "industry": "Financial Services", "city": "Taquari", "name": "Grace Petrov", "job_title": "Product Manager", "company": "Acme Inc"}, {"country": "Nauru", "status": "Overdue", "industry": "Financial Services", "city": "Santa Teresa", "name": "Carol Park", "job_title": "Product Manager", "company": "Initech"}, {"country": "Somalia", "status": "Overdue", "industry": "Financial Services", "city": "Rio Bonito", "name": "Bob Park", "job_title": "Product Manager", "company": "Initech"}, {"country": "Somalia", "status": "Refunded", "industry": "Financial Services", "city": "Kasungu", "name": "Lena Johnson", "job_title": "Sales Representative", "company": "Acme Inc"}, {"country": "Malta", "status": "Paid", "industry": "Energy", "city": "Taquari", "name": "David Khan", "job_title": "Product Manager", "company": "Hooli"}, {"country": "Malta", "status": "Overdue", "industry": "Financial Services", "city": "Kasungu", "name": "Heidi Brandt", "job_title": "Product Manager", "company": "Stark Industries"}, {"country": "Malta", "status": "Paid", "industry": "Financial Services", "city": "Taquari", "name": "Frank Novak", "job_title": "Product Manager", "company": "Initech"}, {"country": "Malta", "status": "Refunded", "industry": "Financial Services", "city": "Taquari", "name": "David Ali", "job_title": "Sales Representative", "company": "Globex"}, {"country": "Ukraine", "status": "Overdue", "industry": "Financial Services", "city": "Santa Teresa", "name": "Eve Diaz", "job_title": "Sales Representative", "company": "Cyberdyne"}, {"country": "Pakistan", "status": "Overdue", "industry": "Energy", "city": "Santa Teresa", "name": "Lena Petrov", "job_title": "Product Manager", "company": "Globex"}, {"country": "Pakistan", "status": "Paid", "industry": "Financial Services", "city": "Rio Bonito", "name": "Priya Novak", "job_title": "Sales Representative", "company": "Acme Inc"}, {"country": "Somalia", "status": "Pending", "industry": "Energy", "city": "Taquari", "name": "Heidi Wong", "job_title": "Sales Representative", "company": "Cyberdyne"}, {"country": "Pakistan", "status": "Overdue", "industry": "Financial Services", "city": "Taquari", "name": "Ivan Park", "job_title": "Product Manager", "company": "Stark Industries"}, {"country": "Malta", "status": "Paid", "industry": "Energy", "city": "Santa Teresa", "name": "Lena Petrov", "job_title": "Product Manager", "company": "Hooli"}, {"country": "Somalia", "status": "Refunded", "industry": "Energy", "city": "Kasungu", "name": "Eve Smith", "job_title": "Product Manager", "company": "Wonka Co"}, {"country": "Somalia", "status": "Paid", "industry": "Financial Services", "city": "Kasungu", "name": "Frank Novak", "job_title": "Sales Representative", "company": "Soylent Corp"}, {"country": "Somalia", "status": "Paid", "industry": "Energy", "city": "Kasungu", "name": "Sara Ali", "job_title": "Product Manager", "company": "Soylent Corp"}, {"country": "Ukraine", "status": "Paid", "industry": "Financial Services", "city": "Santa Teresa", "name": "Lena Moore", "job_title": "Sales Representative", "company": "Vehement"}, {"country": "Ukraine", "status": "Overdue", "industry": "Financial Services", "city": "Rio Bonito", "name": "Bob Cruz", "job_title": "Sales Representative", "company": "Globex"}, {"country": "Ukraine", "status": "Pending", "industry": "Energy", "city": "Taquari", "name": "Karl Moore", "job_title": "Sales Representative", "company": "Acme Inc"}, {"country": "Somalia", "status": "Refunded", "industry": "Energy", "city": "Kasungu", "name": "Carol Novak", "job_title": "Sales Representative", "company": "Acme Inc"}, {"country": "Ukraine", "status": "Overdue", "industry": "Energy", "city": "Rio Bonito", "name": "Eve Ali", "job_title": "Sales Representative", "company": "Soylent Corp"}, {"country": "Somalia", "status": "Paid", "industry": "Financial Services", "city": "Santa Teresa", "name": "Judy Ali", "job_title": "Product Manager", "company": "Acme Inc"}, {"country": "Ukraine", "status": "Paid", "industry": "Energy", "city": "Santa Teresa", "name": "Grace Fischer", "job_title": "Sales Representative", "company": "Stark Industries"}], "dirty_cols": ["country", "status", "industry", "city", "name", "job_title", "company"], "clean_cols": ["country", "status", "industry", "city", "name", "job_title", "company"], "plan": {"dataset_summary": "55 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NR": "Nauru", "nr": "Nauru", "Republic Of Malta": "Malta", "UKR": "Ukraine", "Mlt": "Malta", "ukr": "Ukraine", "FEDERAL REPUBLIC OF SOMALIA": "Somalia", "pk": "Pakistan", "PK": "Pakistan", "NRU": "Nauru", "republic of nauru": "Nauru", "MLT": "Malta", "ua": "Ukraine", "MT": "Malta", "Federal Republic of Somalia": "Somalia", "somalia": "Somalia", "malta": "Malta", "ukraine": "Ukraine", "Republic of Malta": "Malta", "UA": "Ukraine", "nauru": "Nauru", "PAK": "Pakistan", "Republic of Nauru": "Nauru", "Somalta": "Somalia", "mt": "Malta", "mlt": "Malta", "pakistan": "Pakistan", "pak": "Pakistan", "so": "Somalia", "SO": "Somalia", "SOM": "Somalia", "Ukr": "Ukraine", "Ua": "Ukraine"}, "rationale": "Unified 33 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PAID": "Paid", "refunded": "Refunded", "pakd": "Paid", "laet": "Overdue", "pending": "Pending", "due": "Pending", "late": "Overdue", "unpaid": "Pending", "REFUNDED": "Refunded", "DUE": "Pending", "overdue": "Overdue", "OVERDUE": "Overdue", "refund": "Refunded", "Refuned": "Refunded", "unaid": "Pending", "Refund": "Refunded", "overde": "Overdue", "paiid": "Paid", "Late": "Overdue", "paid": "Paid", "piad": "Paid"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"FinServ": "Financial Services", "ENERGY": "Energy", "FINSREV": "Financial Services", "energy": "Energy", "finsrv": "Financial Services", "utilities": "Energy", "Banking": "Financial Services", "finserv": "Financial Services", "Fintech": "Financial Services", "banking": "Financial Services", "Utilities": "Energy", "Oil & Gas": "Energy", "fnitech": "Financial Services", "Finance": "Financial Services", "FINANCE": "Financial Services", "BANKING": "Financial Services", "finsserv": "Financial Services", "fintech": "Financial Services", "finance": "Financial Services", "fiannce": "Financial Services"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Kasugnu": "Kasungu", "santa teresa": "Santa Teresa", "taquari": "Taquari", "rio bonito": "Rio Bonito", "snata teresa": "Santa Teresa", "Tqauari": "Taquari", "SANTA TERESA": "Santa Teresa", "Taquai": "Taquari", "TAQQUARI": "Taquari", "kasvngu": "Kasungu", "Santa Tersa": "Santa Teresa", "rio konito": "Rio Bonito", "kasungu": "Kasungu", "TAQUARI": "Taquari", "ksaungu": "Kasungu"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SALES REP": "Sales Representative", "prod manager": "Product Manager", "Rep": "Sales Representative", "prdduct manager": "Product Manager", "PM": "Product Manager", "PROD MANAGER": "Product Manager", "sales rep": "Sales Representative", "Prod Mgr": "Product Manager", "Sales Rep": "Sales Representative", "sales representative": "Sales Representative", "Prod Manager": "Product Manager", "product manager": "Product Manager", "PRDO MANAGER": "Product Manager", "Prxd Mgr": "Product Manager", "rep": "Sales Representative", "Product anager": "Product Manager", "Prod aMnager": "Product Manager", "Sales Reprzsentative": "Sales Representative", "Pm": "Product Manager", "prod mgr": "Product Manager", "REP": "Sales Representative"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"state": "WY", "industry": "EDUCATION", "signup_date": "4/16/2023", "rate": "66.1%", "company": "  Vehement", "city": "Miracema do Tocantins"}, {"state": "WY   ", "industry": " EDUCATION ", "signup_date": "44973", "rate": "59.5%", "company": "Hooli", "city": " stephenville "}, {"state": "Pennsylvania", "industry": " Comms", "signup_date": "1 Jan 2023", "rate": "27.6%", "company": "Cyberdyne  ", "city": "Qujsng"}, {"state": "pennsylvania", "industry": "   Energy", "signup_date": "5/17/2023", "rate": "42.5%", "company": "Umbrella ", "city": "QUJING"}, {"state": "PA", "industry": "Telco", "signup_date": "9/19/2023", "rate": "81.8%", "company": "Acme Inc", "city": "Stephenville"}, {"state": " ct ", "industry": "Utilities", "signup_date": "8/28/2023", "rate": "29.8%", "company": "Stark  Industries", "city": "Miracema do Tocantins   "}, {"state": "wyoming", "industry": "Comms", "signup_date": "2023-07-10", "rate": "21.7%", "company": " Initech ", "city": " Stephenville"}, {"state": "wy", "industry": " Education", "signup_date": "10/6/2023", "rate": "24.4%", "company": "Acme  Inc", "city": "Miracema do Tocantins"}, {"state": "ct", "industry": "TELECOM", "signup_date": "26 Apr 2023", "rate": "11.4%", "company": "Vehement", "city": "miracema do tocantins"}, {"state": "ct", "industry": "Energy", "signup_date": "2023-10-10", "rate": "77.7%", "company": "   Soylent Corp", "city": "QUJINNG"}, {"state": "Guam", "industry": "Education", "signup_date": "2023-11-15", "rate": "19.2%", "company": "Globex", "city": "QUJIG"}, {"state": "Wyoming", "industry": "ED", "signup_date": "10 Mar 2023", "rate": "80.8%", "company": "Soylent Corp", "city": "Stephenville"}, {"state": "WV  ", "industry": "utilities", "signup_date": "45127", "rate": "83.9%", "company": "Soylent Corp   ", "city": "qujing"}, {"state": "Guam", "industry": "Oil & Gas", "signup_date": "9/6/2023", "rate": "17.5%", "company": "Soylent Corp", "city": "   Qujing"}, {"state": "PA   ", "industry": "education", "signup_date": "45162", "rate": "85.5%", "company": "Stark Industries", "city": "qujing"}, {"state": "Connecticut", "industry": "   Telecommunications", "signup_date": "2023-06-19", "rate": "51.4%", "company": "Umbrella", "city": "quzing"}, {"state": "Wyoming", "industry": "ENERGY", "signup_date": "2023-12-02", "rate": "34.6%", "company": "Stark Industries", "city": "QUJING"}, {"state": "wyoming", "industry": "Comms", "signup_date": "2023-07-10", "rate": "21.7%", "company": " Initech ", "city": " Stephenville"}, {"state": "wy", "industry": " Education", "signup_date": "10/6/2023", "rate": "24.4%", "company": "Acme  Inc", "city": "Miracema do Tocantins"}], "clean": [{"state": "Wyoming", "industry": "Education", "signup_date": "2023-04-16", "rate": 0.6609999999999999, "company": "Vehement", "city": "Miracema do Tocantins"}, {"state": "Wyoming", "industry": "Education", "signup_date": "2023-02-16", "rate": 0.595, "company": "Hooli", "city": "Stephenville"}, {"state": "Pennsylvania", "industry": "Telecommunications", "signup_date": "2023-01-01", "rate": 0.276, "company": "Cyberdyne", "city": "Qujing"}, {"state": "Pennsylvania", "industry": "Energy", "signup_date": "2023-05-17", "rate": 0.425, "company": "Umbrella", "city": "Qujing"}, {"state": "Pennsylvania", "industry": "Telecommunications", "signup_date": "2023-09-19", "rate": 0.818, "company": "Acme Inc", "city": "Stephenville"}, {"state": "Connecticut", "industry": "Energy", "signup_date": "2023-08-28", "rate": 0.298, "company": "Stark Industries", "city": "Miracema do Tocantins"}, {"state": "Wyoming", "industry": "Telecommunications", "signup_date": "2023-07-10", "rate": 0.217, "company": "Initech", "city": "Stephenville"}, {"state": "Wyoming", "industry": "Education", "signup_date": "2023-10-06", "rate": 0.244, "company": "Acme Inc", "city": "Miracema do Tocantins"}, {"state": "Connecticut", "industry": "Telecommunications", "signup_date": "2023-04-26", "rate": 0.114, "company": "Vehement", "city": "Miracema do Tocantins"}, {"state": "Connecticut", "industry": "Energy", "signup_date": "2023-10-10", "rate": 0.777, "company": "Soylent Corp", "city": "Qujing"}, {"state": "Guam", "industry": "Education", "signup_date": "2023-11-15", "rate": 0.192, "company": "Globex", "city": "Qujing"}, {"state": "Wyoming", "industry": "Education", "signup_date": "2023-03-10", "rate": 0.8079999999999999, "company": "Soylent Corp", "city": "Stephenville"}, {"state": "West Virginia", "industry": "Energy", "signup_date": "2023-07-20", "rate": 0.8390000000000001, "company": "Soylent Corp", "city": "Qujing"}, {"state": "Guam", "industry": "Energy", "signup_date": "2023-09-06", "rate": 0.175, "company": "Soylent Corp", "city": "Qujing"}, {"state": "Pennsylvania", "industry": "Education", "signup_date": "2023-08-24", "rate": 0.855, "company": "Stark Industries", "city": "Qujing"}, {"state": "Connecticut", "industry": "Telecommunications", "signup_date": "2023-06-19", "rate": 0.514, "company": "Umbrella", "city": "Qujing"}, {"state": "Wyoming", "industry": "Energy", "signup_date": "2023-12-02", "rate": 0.34600000000000003, "company": "Stark Industries", "city": "Qujing"}], "dirty_cols": ["state", "industry", "signup_date", "rate", "company", "city"], "clean_cols": ["state", "industry", "signup_date", "rate", "company", "city"], "plan": {"dataset_summary": "19 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WY": "Wyoming", "pennsylvania": "Pennsylvania", "PA": "Pennsylvania", "ct": "Connecticut", "wyoming": "Wyoming", "wy": "Wyoming", "WV": "West Virginia"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"EDUCATION": "Education", "Comms": "Telecommunications", "Telco": "Telecommunications", "Utilities": "Energy", "TELECOM": "Telecommunications", "ED": "Education", "utilities": "Energy", "Oil & Gas": "Energy", "education": "Education", "ENERGY": "Energy"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"stephenville": "Stephenville", "Qujsng": "Qujing", "QUJING": "Qujing", "miracema do tocantins": "Miracema do Tocantins", "QUJINNG": "Qujing", "QUJIG": "Qujing", "qujing": "Qujing", "quzing": "Qujing"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "   CEO", "phone": "(534)042-1232", "signup_date": "10/14/2023", "state": "NEBRASKA", "department": "Accounting"}, {"job_title": "product manager", "phone": "(502)904-7350", "signup_date": "12/17/2023", "state": "   vermont", "department": "Fin"}, {"job_title": "PM", "phone": "595.978.6010", "signup_date": "2023-06-27", "state": " NE", "department": "CUSTOMER SUPPORT"}, {"job_title": "PROD MGR", "phone": "586-937-0807", "signup_date": "20 Sep 2023", "state": "NE", "department": "Ops"}, {"job_title": "chief executive officer  ", "phone": "(501)326-9546", "signup_date": "2023-09-07", "state": "VT", "department": "ops"}, {"job_title": "product manager", "phone": "502.912.6441", "signup_date": "2023-04-02", "state": "  ne", "department": "ops"}, {"job_title": "prod manager", "phone": "(561)440-9123", "signup_date": "2023-10-15", "state": "Nebraska", "department": "Ops"}, {"job_title": "Pm", "phone": "(545)869-2413", "signup_date": "45066", "state": "IA", "department": " accounting "}, {"job_title": "CEO", "phone": "5406971423", "signup_date": "2023-12-23", "state": "Vemont", "department": "Support"}, {"job_title": "Prod Mgr", "phone": "(594)207-8975", "signup_date": "14 Jan 2023", "state": "IA", "department": "Finance"}, {"job_title": "   PROFUCT MANAGER", "phone": "5046613185", "signup_date": "2023-10-03", "state": "Ia", "department": "Cuts Support"}, {"job_title": "product manager", "phone": "5164770472", "signup_date": "45233", "state": "vermont", "department": "Accounting"}, {"job_title": "ProdManager", "phone": "5046771740", "signup_date": "2023-08-01", "state": "Nebraska", "department": "OPERATIONS"}, {"job_title": "  CEO", "phone": "5425824814", "signup_date": "2023-10-22", "state": "NE", "department": "Cust Support "}, {"job_title": "pm", "phone": "(519)649-3695", "signup_date": "2023-11-24", "state": "NE", "department": "Finance"}, {"job_title": "c.e.o.", "phone": "556.954.1382", "signup_date": "2023-05-09", "state": " vt ", "department": "Customer Support"}, {"job_title": "Prod Manager", "phone": "(550)897-8979", "signup_date": "45273", "state": "vermont", "department": "OPS"}, {"job_title": "PROD MANAGER", "phone": "580-169-0006", "signup_date": "2023-02-05", "state": "VT", "department": "Ops"}, {"job_title": "prod  mgr", "phone": "5076788104", "signup_date": "45080", "state": "Ioowa", "department": "accounting"}, {"job_title": "Chief Executive Officer", "phone": "565.418.6903", "signup_date": "16 Mar 2023", "state": "Nebrasa", "department": "Operations"}, {"job_title": "prod manager ", "phone": "(594)898-5447", "signup_date": "45001", "state": "ioa", "department": "cust support"}, {"job_title": "PROD MR", "phone": "(570)195-7584", "signup_date": "15 Dec 2023", "state": "NEBRASKA", "department": "Ops"}, {"job_title": "Chief Executive Officer", "phone": "5497156217", "signup_date": "2023-02-02", "state": "Nebraska", "department": "FINANCE"}, {"job_title": "PROD MGR", "phone": "(510)634-3318", "signup_date": "2 Jan 2023", "state": "Nebraska", "department": "Fin"}, {"job_title": "   CEO", "phone": "(534)042-1232", "signup_date": "10/14/2023", "state": "NEBRASKA", "department": "Accounting"}, {"job_title": "PROD MR", "phone": "(570)195-7584", "signup_date": "15 Dec 2023", "state": "NEBRASKA", "department": "Ops"}, {"job_title": "", "phone": "", "signup_date": "", "state": "", "department": ""}], "clean": [{"job_title": "Chief Executive Officer", "phone": "(534) 042-1232", "signup_date": "2023-10-14", "state": "Nebraska", "department": "Finance"}, {"job_title": "Product Manager", "phone": "(502) 904-7350", "signup_date": "2023-12-17", "state": "Vermont", "department": "Finance"}, {"job_title": "Product Manager", "phone": "(595) 978-6010", "signup_date": "2023-06-27", "state": "Nebraska", "department": "Customer Support"}, {"job_title": "Product Manager", "phone": "(586) 937-0807", "signup_date": "2023-09-20", "state": "Nebraska", "department": "Operations"}, {"job_title": "Chief Executive Officer", "phone": "(501) 326-9546", "signup_date": "2023-09-07", "state": "Vermont", "department": "Operations"}, {"job_title": "Product Manager", "phone": "(502) 912-6441", "signup_date": "2023-04-02", "state": "Nebraska", "department": "Operations"}, {"job_title": "Product Manager", "phone": "(561) 440-9123", "signup_date": "2023-10-15", "state": "Nebraska", "department": "Operations"}, {"job_title": "Product Manager", "phone": "(545) 869-2413", "signup_date": "2023-05-20", "state": "Iowa", "department": "Finance"}, {"job_title": "Chief Executive Officer", "phone": "(540) 697-1423", "signup_date": "2023-12-23", "state": "Vermont", "department": "Customer Support"}, {"job_title": "Product Manager", "phone": "(594) 207-8975", "signup_date": "2023-01-14", "state": "Iowa", "department": "Finance"}, {"job_title": "Product Manager", "phone": "(504) 661-3185", "signup_date": "2023-10-03", "state": "Iowa", "department": "Customer Support"}, {"job_title": "Product Manager", "phone": "(516) 477-0472", "signup_date": "2023-11-03", "state": "Vermont", "department": "Finance"}, {"job_title": "Product Manager", "phone": "(504) 677-1740", "signup_date": "2023-08-01", "state": "Nebraska", "department": "Operations"}, {"job_title": "Chief Executive Officer", "phone": "(542) 582-4814", "signup_date": "2023-10-22", "state": "Nebraska", "department": "Customer Support"}, {"job_title": "Product Manager", "phone": "(519) 649-3695", "signup_date": "2023-11-24", "state": "Nebraska", "department": "Finance"}, {"job_title": "Chief Executive Officer", "phone": "(556) 954-1382", "signup_date": "2023-05-09", "state": "Vermont", "department": "Customer Support"}, {"job_title": "Product Manager", "phone": "(550) 897-8979", "signup_date": "2023-12-13", "state": "Vermont", "department": "Operations"}, {"job_title": "Product Manager", "phone": "(580) 169-0006", "signup_date": "2023-02-05", "state": "Vermont", "department": "Operations"}, {"job_title": "Product Manager", "phone": "(507) 678-8104", "signup_date": "2023-06-03", "state": "Iowa", "department": "Finance"}, {"job_title": "Chief Executive Officer", "phone": "(565) 418-6903", "signup_date": "2023-03-16", "state": "Nebraska", "department": "Operations"}, {"job_title": "Product Manager", "phone": "(594) 898-5447", "signup_date": "2023-03-16", "state": "Iowa", "department": "Customer Support"}, {"job_title": "Product Manager", "phone": "(570) 195-7584", "signup_date": "2023-12-15", "state": "Nebraska", "department": "Operations"}, {"job_title": "Chief Executive Officer", "phone": "(549) 715-6217", "signup_date": "2023-02-02", "state": "Nebraska", "department": "Finance"}, {"job_title": "Product Manager", "phone": "(510) 634-3318", "signup_date": "2023-01-02", "state": "Nebraska", "department": "Finance"}], "dirty_cols": ["job_title", "phone", "signup_date", "state", "department"], "clean_cols": ["job_title", "phone", "signup_date", "state", "department"], "plan": {"dataset_summary": "27 rows × 5 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CEO": "Chief Executive Officer", "product manager": "Product Manager", "PM": "Product Manager", "PROD MGR": "Product Manager", "chief executive officer": "Chief Executive Officer", "prod manager": "Product Manager", "Pm": "Product Manager", "Prod Mgr": "Product Manager", "PROFUCT MANAGER": "Product Manager", "ProdManager": "Product Manager", "pm": "Product Manager", "c.e.o.": "Chief Executive Officer", "Prod Manager": "Product Manager", "PROD MANAGER": "Product Manager", "prod mgr": "Product Manager", "PROD MR": "Product Manager"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NEBRASKA": "Nebraska", "vermont": "Vermont", "NE": "Nebraska", "VT": "Vermont", "ne": "Nebraska", "IA": "Iowa", "Vemont": "Vermont", "Ia": "Iowa", "vt": "Vermont", "Ioowa": "Iowa", "Nebrasa": "Nebraska", "ioa": "Iowa"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Accounting": "Finance", "Fin": "Finance", "CUSTOMER SUPPORT": "Customer Support", "Ops": "Operations", "ops": "Operations", "accounting": "Finance", "Support": "Customer Support", "Cuts Support": "Customer Support", "OPERATIONS": "Operations", "Cust Support": "Customer Support", "OPS": "Operations", "cust support": "Customer Support", "FINANCE": "Finance"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "34.6%", "amount": "6.937,36", "job_title": " SALES REPRESENTATIVE", "company": "Vehement", "status": "free trial", "email": "bfgpef@example.com   ", "unnamed": ""}, {"rate": "10.9%", "amount": "1.877,72", "job_title": " drector ", "company": "Cyberdyne ", "status": "trial", "email": "JBBEA@MAIL.COM   ", "unnamed": ""}, {"rate": "71.9%", "amount": "7.786,34", "job_title": "administrative  assistant", "company": "Soylent  Corp", "status": "Trial", "email": "jjacnj@mail.com  ", "unnamed": ""}, {"rate": "7.4%", "amount": "$4,358.03", "job_title": "Dir", "company": "Globex", "status": "TRIAL", "email": "IHJDII@EXAMPLE.COM", "unnamed": ""}, {"rate": "41.2%", "amount": "$6,884.38", "job_title": "Dir", "company": "Stark Industries", "status": "ACTIVE", "email": "akokb@test.org", "unnamed": ""}, {"rate": "97.4%", "amount": "8.115,39", "job_title": "Administrative Assistant", "company": "Acme Inc ", "status": "TRIAL", "email": "EPHGE@TEST.ORG", "unnamed": ""}, {"rate": "95.3%", "amount": "541,27", "job_title": "Dir ", "company": "   Umbrella", "status": "paused ", "email": "dabf@corp.io", "unnamed": ""}, {"rate": "6.5%", "amount": "$4,551.75", "job_title": "  admin", "company": "Vehement", "status": "trial", "email": "CACOIO@TEST.ORG", "unnamed": ""}, {"rate": "52.9%", "amount": "$6,508.93", "job_title": " Director ", "company": "Acme Inc", "status": "ON HOLD", "email": "cahgcaf@corp.io ", "unnamed": ""}, {"rate": "86.7%", "amount": "7.232,99", "job_title": " Safes Rep", "company": "Cyberdyne  ", "status": "free trial", "email": "  mkjhdg@mail.com", "unnamed": ""}, {"rate": "81.1%", "amount": "$3,084.66", "job_title": "Director   ", "company": "  Umbrella", "status": "paused  ", "email": "POKDE@MAIL.COM", "unnamed": ""}, {"rate": "81.5%", "amount": "$784.34", "job_title": "di.", "company": "Stark Industries", "status": "paused  ", "email": "ddpog@mail.com", "unnamed": ""}, {"rate": "17.9%", "amount": "$3,957.43", "job_title": "Dir", "company": "Cyberdyne", "status": "altive", "email": " DPOMKA@EXAMPLE.COM ", "unnamed": ""}, {"rate": "61.4%", "amount": "5.556,50", "job_title": "dir", "company": "   Vehement", "status": "Active ", "email": "lljg@mail.com", "unnamed": ""}, {"rate": "45.6%", "amount": "3.627,06", "job_title": "amdin assistant", "company": "Umbrella", "status": " On Hod", "email": "GHBI@MAIL.COM", "unnamed": ""}, {"rate": "90.6%", "amount": "$2,166.93", "job_title": "dir", "company": "Initech", "status": "on hold", "email": "PHNBDD@CORP.IO", "unnamed": ""}, {"rate": "57.4%", "amount": "3.644,77", "job_title": " Admin Assistant", "company": "Cyberdyne  ", "status": "Ajtive   ", "email": "  pbpcl@test.org", "unnamed": ""}, {"rate": "81.7%", "amount": "3.443,69", "job_title": "Director", "company": "Acme Inc", "status": "Paused", "email": "jeoin@test.org", "unnamed": ""}, {"rate": "68.8%", "amount": "$3,307.34", "job_title": "sales representative", "company": "Cyberdyne   ", "status": "   paused", "email": "FHHJ@EXAMPLE.COM", "unnamed": ""}, {"rate": "76.6%", "amount": "$7,681.58", "job_title": "REP", "company": "Soylent Corp", "status": "Pauhed", "email": " labaco@test.org", "unnamed": ""}, {"rate": "93.5%", "amount": "0", "job_title": "Administrative Assistant", "company": "Cyberdyne", "status": "ACTIVE", "email": "   PFPKCP@EXAMPLE.COM", "unnamed": ""}, {"rate": "28.0%", "amount": "None", "job_title": "Sales Representative   ", "company": "Hooli", "status": "Active", "email": "gmcgj@mail.com", "unnamed": ""}, {"rate": "94.9%", "amount": "#N/A", "job_title": "administrative assistant", "company": "Cyberdyne", "status": " ACTIVE ", "email": "OLPKE@CORP.IO", "unnamed": ""}, {"rate": "97.1%", "amount": "8.194,13", "job_title": "Dir   ", "company": "Initech", "status": "on hold", "email": "fhjg@mail.com", "unnamed": ""}, {"rate": "", "amount": "", "job_title": "", "company": "", "status": "", "email": "", "unnamed": ""}], "clean": [{"rate": 0.34600000000000003, "amount": 6937.36, "job_title": "Sales Representative", "company": "Vehement", "status": "Trial", "email": "bfgpef@example.com"}, {"rate": 0.109, "amount": 1877.72, "job_title": "Director", "company": "Cyberdyne", "status": "Trial", "email": "jbbea@mail.com"}, {"rate": 0.7190000000000001, "amount": 7786.34, "job_title": "Administrative Assistant", "company": "Soylent Corp", "status": "Trial", "email": "jjacnj@mail.com"}, {"rate": 0.07400000000000001, "amount": 4358.03, "job_title": "Director", "company": "Globex", "status": "Trial", "email": "ihjdii@example.com"}, {"rate": 0.41200000000000003, "amount": 6884.38, "job_title": "Director", "company": "Stark Industries", "status": "Active", "email": "akokb@test.org"}, {"rate": 0.9740000000000001, "amount": 8115.39, "job_title": "Administrative Assistant", "company": "Acme Inc", "status": "Trial", "email": "ephge@test.org"}, {"rate": 0.953, "amount": 541.27, "job_title": "Director", "company": "Umbrella", "status": "Paused", "email": "dabf@corp.io"}, {"rate": 0.065, "amount": 4551.75, "job_title": "Administrative Assistant", "company": "Vehement", "status": "Trial", "email": "cacoio@test.org"}, {"rate": 0.529, "amount": 6508.93, "job_title": "Director", "company": "Acme Inc", "status": "Paused", "email": "cahgcaf@corp.io"}, {"rate": 0.867, "amount": 7232.99, "job_title": "Sales Representative", "company": "Cyberdyne", "status": "Trial", "email": "mkjhdg@mail.com"}, {"rate": 0.8109999999999999, "amount": 3084.66, "job_title": "Director", "company": "Umbrella", "status": "Paused", "email": "pokde@mail.com"}, {"rate": 0.815, "amount": 784.34, "job_title": "Director", "company": "Stark Industries", "status": "Paused", "email": "ddpog@mail.com"}, {"rate": 0.179, "amount": 3957.43, "job_title": "Director", "company": "Cyberdyne", "status": "Active", "email": "dpomka@example.com"}, {"rate": 0.614, "amount": 5556.5, "job_title": "Director", "company": "Vehement", "status": "Active", "email": "lljg@mail.com"}, {"rate": 0.456, "amount": 3627.06, "job_title": "Administrative Assistant", "company": "Umbrella", "status": "Paused", "email": "ghbi@mail.com"}, {"rate": 0.9059999999999999, "amount": 2166.93, "job_title": "Director", "company": "Initech", "status": "Paused", "email": "phnbdd@corp.io"}, {"rate": 0.574, "amount": 3644.77, "job_title": "Administrative Assistant", "company": "Cyberdyne", "status": "Active", "email": "pbpcl@test.org"}, {"rate": 0.8170000000000001, "amount": 3443.69, "job_title": "Director", "company": "Acme Inc", "status": "Paused", "email": "jeoin@test.org"}, {"rate": 0.688, "amount": 3307.34, "job_title": "Sales Representative", "company": "Cyberdyne", "status": "Paused", "email": "fhhj@example.com"}, {"rate": 0.7659999999999999, "amount": 7681.58, "job_title": "Sales Representative", "company": "Soylent Corp", "status": "Paused", "email": "labaco@test.org"}, {"rate": 0.935, "amount": 0.0, "job_title": "Administrative Assistant", "company": "Cyberdyne", "status": "Active", "email": "pfpkcp@example.com"}, {"rate": 0.28, "amount": NaN, "job_title": "Sales Representative", "company": "Hooli", "status": "Active", "email": "gmcgj@mail.com"}, {"rate": 0.9490000000000001, "amount": NaN, "job_title": "Administrative Assistant", "company": "Cyberdyne", "status": "Active", "email": "olpke@corp.io"}, {"rate": 0.971, "amount": 8194.13, "job_title": "Director", "company": "Initech", "status": "Paused", "email": "fhjg@mail.com"}], "dirty_cols": ["rate", "amount", "job_title", "company", "status", "email", "unnamed"], "clean_cols": ["rate", "amount", "job_title", "company", "status", "email"], "plan": {"dataset_summary": "25 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SALES REPRESENTATIVE": "Sales Representative", "drector": "Director", "administrative assistant": "Administrative Assistant", "Dir": "Director", "admin": "Administrative Assistant", "Safes Rep": "Sales Representative", "di.": "Director", "dir": "Director", "amdin assistant": "Administrative Assistant", "Admin Assistant": "Administrative Assistant", "sales representative": "Sales Representative", "REP": "Sales Representative"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"free trial": "Trial", "trial": "Trial", "TRIAL": "Trial", "ACTIVE": "Active", "paused": "Paused", "ON HOLD": "Paused", "altive": "Active", "On Hod": "Paused", "on hold": "Paused", "Ajtive": "Active", "Pauhed": "Paused"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"rate": "57.8%", "email": "IKBNFEL@TEST.ORG", "unit": "centimeters", "state": "MN", "phone": "585.958.4156", "currency": " boliviano ", "col_x": ""}, {"rate": "80.7%", "email": " fkbckme@test.org", "unit": "POUND", "state": "MN  ", "phone": "(528)546-3636", "currency": "ALGERIAN DINAR", "col_x": ""}, {"rate": "69.4%", "email": "FMNCLI@MAIL.COM  ", "unit": "lbs  ", "state": "mn   ", "phone": "502.946.0886", "currency": "DZD", "col_x": ""}, {"rate": "36.3%", "email": "fgba@mail.com", "unit": "KG", "state": " CA ", "phone": "5232676509", "currency": " Boliviano ", "col_x": ""}, {"rate": "36.2%", "email": "KOMNG@TEST.ORG", "unit": "cm", "state": "minnesota", "phone": "529-336-3055", "currency": " BOB ", "col_x": ""}, {"rate": "78.8%", "email": "HDAM@EXAMPLE.COM", "unit": "Mi", "state": "ca", "phone": "590-287-3724", "currency": " bob", "col_x": ""}, {"rate": "70.8%", "email": "NAIK@CORP.IO", "unit": "lbs", "state": "Montana", "phone": "505.472.4429", "currency": "Algerian Dinar", "col_x": ""}, {"rate": "63.3%", "email": " CFBHBMC@MAIL.COM ", "unit": "cm", "state": "   mt", "phone": "5237492691", "currency": " BOB", "col_x": ""}, {"rate": "36.0%", "email": "GIOA@EXAMPLE.COM", "unit": " Mi ", "state": "mn", "phone": "5490784351", "currency": "boliviano", "col_x": ""}, {"rate": "7.8%", "email": "  effg@corp.io", "unit": " kg", "state": "Minnesota", "phone": "562.148.8621", "currency": "Boliviano", "col_x": ""}, {"rate": "57.7%", "email": "igmfglf@example.com", "unit": "Centimeter", "state": "minnesota", "phone": "539-456-7613", "currency": "Boliviano", "col_x": ""}, {"rate": "46.1%", "email": "CBICDGI@TEST.ORG ", "unit": "KILOGRAMS", "state": "MONTANA", "phone": "5505211679", "currency": "Algerian Dinar", "col_x": ""}, {"rate": "98.3%", "email": "mogmgc@test.org", "unit": "ceytimeters", "state": " ca ", "phone": "(580)267-6689", "currency": "boliviano", "col_x": ""}, {"rate": "84.1%", "email": "MDHJ@EXAMPLE.COM", "unit": "   Kilograms", "state": "MT ", "phone": "589.945.5400", "currency": "Algerian Dinar  ", "col_x": ""}, {"rate": "15.8%", "email": "fmholik@mail.com", "unit": "lb", "state": "ca", "phone": "529.068.1768", "currency": "Algerian Oinar", "col_x": ""}, {"rate": "3.7%", "email": "fank@test.org", "unit": "lbs", "state": "California", "phone": "5663536550", "currency": "DZD", "col_x": ""}, {"rate": "8.1%", "email": "CDBKHF@EXAMPLE.COM ", "unit": "Miles", "state": "ca", "phone": "5046162953", "currency": "dzd", "col_x": ""}, {"rate": "45.2%", "email": "jkignod@example.com", "unit": "MILE", "state": "Minnesota", "phone": "5841841228", "currency": "DZD", "col_x": ""}, {"rate": "12.1%", "email": " GELA@TEST.ORG", "unit": "mi", "state": "Montana", "phone": "5264819914", "currency": "BOB", "col_x": ""}, {"rate": "64.9%", "email": "   CBLOOBC@TEST.ORG", "unit": "CENTIMETERS", "state": "CA", "phone": "535.660.5466", "currency": " BOB", "col_x": ""}, {"rate": "90.0%", "email": "adgjeg@mail.com", "unit": "miles", "state": " MN ", "phone": "(545)249-2900", "currency": "   Dzd", "col_x": ""}, {"rate": "35.1%", "email": "   ipmkk@test.org", "unit": "Cm", "state": "minnesoat", "phone": "(501)003-4865", "currency": "Algerian Dinar", "col_x": ""}, {"rate": "52.0%", "email": " EGHJDDI@EXAMPLE.COM ", "unit": " kilograms ", "state": "California", "phone": "513-089-0827", "currency": "Algeiran Dinar", "col_x": ""}, {"rate": "45.6%", "email": "EOBHJ@MAIL.COM", "unit": "centimeters", "state": "mt", "phone": "526-954-1262", "currency": "Algerian Dinar", "col_x": ""}, {"rate": "51.2%", "email": "mndjgi@test.org", "unit": "kg", "state": " minnesota", "phone": "5495981942", "currency": "DZD", "col_x": ""}, {"rate": "27.4%", "email": "JMBLP@EXAMPLE.COM", "unit": "   Pound", "state": "MINESOTA", "phone": "578-254-5087", "currency": "   BOB", "col_x": ""}, {"rate": "13.8%", "email": "PHAB@CORP.IO", "unit": "lb   ", "state": "Montana", "phone": "540-128-1935", "currency": "Algerian Dinar  ", "col_x": ""}, {"rate": "3.6%", "email": "IOIKBLG@TEST.ORG", "unit": "KG", "state": "California", "phone": "(549)536-8239", "currency": "Boliviuno", "col_x": ""}, {"rate": "70.1%", "email": "kpmif@test.org", "unit": "cm", "state": "montana", "phone": "5006258784", "currency": "   Algerian Dinar", "col_x": ""}, {"rate": "96.4%", "email": "KBNGNLM@CORP.IO ", "unit": "lb", "state": "California", "phone": "5428909881", "currency": "Algerian Dinar", "col_x": ""}, {"rate": "25.9%", "email": "PKKOGH@MAIL.COM", "unit": "mi", "state": "Mt", "phone": "595.115.8345", "currency": "Algerian Dinar", "col_x": ""}, {"rate": "88.6%", "email": " laaid@mail.com", "unit": " centimeters ", "state": "Mn", "phone": "527.217.7885", "currency": " Boliviano ", "col_x": ""}, {"rate": "61.6%", "email": "  GBCBPD@TEST.ORG", "unit": "Lb", "state": " Mn ", "phone": "529-794-1430", "currency": " bob ", "col_x": ""}, {"rate": "70.6%", "email": "ogfh@example.com", "unit": "KG", "state": "MT", "phone": "597-470-2589", "currency": "   BOB", "col_x": ""}, {"rate": "81.8%", "email": "dbnboe@test.org", "unit": "   kg", "state": "california", "phone": "5666418132", "currency": "BOB", "col_x": ""}, {"rate": "48.8%", "email": " ebkank@example.com ", "unit": "   mile", "state": "CA", "phone": "5097602256", "currency": "dzd", "col_x": ""}, {"rate": "25.2%", "email": "ACHK@CORP.IO", "unit": "lbs", "state": "MT", "phone": "598-661-4446", "currency": "DZD", "col_x": ""}, {"rate": "63.6%", "email": "hmhggk@test.org", "unit": " pound ", "state": "minnesota", "phone": "570.841.7398", "currency": "Boliviano", "col_x": ""}, {"rate": "70.2%", "email": "ADHPOD@EXAMPLE.COM", "unit": "miles ", "state": "california", "phone": "(577)392-6559", "currency": "Dzd", "col_x": ""}, {"rate": "63.7%", "email": "pkmffpf@corp.io", "unit": "pounds", "state": "MONTANA", "phone": "(597)849-7075", "currency": "BOB", "col_x": ""}, {"rate": "76.5%", "email": "PMNGF@TEST.ORG", "unit": "kilograms", "state": "MT", "phone": "(524)066-9502", "currency": "aglerian dinar", "col_x": ""}, {"rate": "79.3%", "email": "PIBLKAJ@EXAMPLE.COM", "unit": "POUND  ", "state": "MN  ", "phone": "545.879.3239", "currency": "Boliviano", "col_x": ""}, {"rate": "76.2%", "email": "ODPMG@TEST.ORG", "unit": "   kg", "state": "mt", "phone": "(523)234-7639", "currency": "BOB", "col_x": ""}, {"rate": "13.8%", "email": "PHAB@CORP.IO", "unit": "lb   ", "state": "Montana", "phone": "540-128-1935", "currency": "Algerian Dinar  ", "col_x": ""}, {"rate": "57.8%", "email": "IKBNFEL@TEST.ORG", "unit": "centimeters", "state": "MN", "phone": "585.958.4156", "currency": " boliviano ", "col_x": ""}], "clean": [{"rate": 0.578, "email": "ikbnfel@test.org", "unit": "cm", "state": "Minnesota", "phone": "(585) 958-4156", "currency": "BOB"}, {"rate": 0.807, "email": "fkbckme@test.org", "unit": "lb", "state": "Minnesota", "phone": "(528) 546-3636", "currency": "DZD"}, {"rate": 0.6940000000000001, "email": "fmncli@mail.com", "unit": "lb", "state": "Minnesota", "phone": "(502) 946-0886", "currency": "DZD"}, {"rate": 0.363, "email": "fgba@mail.com", "unit": "kg", "state": "California", "phone": "(523) 267-6509", "currency": "BOB"}, {"rate": 0.36200000000000004, "email": "komng@test.org", "unit": "cm", "state": "Minnesota", "phone": "(529) 336-3055", "currency": "BOB"}, {"rate": 0.7879999999999999, "email": "hdam@example.com", "unit": "mi", "state": "California", "phone": "(590) 287-3724", "currency": "BOB"}, {"rate": 0.708, "email": "naik@corp.io", "unit": "lb", "state": "Montana", "phone": "(505) 472-4429", "currency": "DZD"}, {"rate": 0.633, "email": "cfbhbmc@mail.com", "unit": "cm", "state": "Montana", "phone": "(523) 749-2691", "currency": "BOB"}, {"rate": 0.36, "email": "gioa@example.com", "unit": "mi", "state": "Minnesota", "phone": "(549) 078-4351", "currency": "BOB"}, {"rate": 0.078, "email": "effg@corp.io", "unit": "kg", "state": "Minnesota", "phone": "(562) 148-8621", "currency": "BOB"}, {"rate": 0.5770000000000001, "email": "igmfglf@example.com", "unit": "cm", "state": "Minnesota", "phone": "(539) 456-7613", "currency": "BOB"}, {"rate": 0.461, "email": "cbicdgi@test.org", "unit": "kg", "state": "Montana", "phone": "(550) 521-1679", "currency": "DZD"}, {"rate": 0.983, "email": "mogmgc@test.org", "unit": "cm", "state": "California", "phone": "(580) 267-6689", "currency": "BOB"}, {"rate": 0.841, "email": "mdhj@example.com", "unit": "kg", "state": "Montana", "phone": "(589) 945-5400", "currency": "DZD"}, {"rate": 0.158, "email": "fmholik@mail.com", "unit": "lb", "state": "California", "phone": "(529) 068-1768", "currency": "DZD"}, {"rate": 0.037000000000000005, "email": "fank@test.org", "unit": "lb", "state": "California", "phone": "(566) 353-6550", "currency": "DZD"}, {"rate": 0.081, "email": "cdbkhf@example.com", "unit": "mi", "state": "California", "phone": "(504) 616-2953", "currency": "DZD"}, {"rate": 0.452, "email": "jkignod@example.com", "unit": "mi", "state": "Minnesota", "phone": "(584) 184-1228", "currency": "DZD"}, {"rate": 0.121, "email": "gela@test.org", "unit": "mi", "state": "Montana", "phone": "(526) 481-9914", "currency": "BOB"}, {"rate": 0.649, "email": "cbloobc@test.org", "unit": "cm", "state": "California", "phone": "(535) 660-5466", "currency": "BOB"}, {"rate": 0.9, "email": "adgjeg@mail.com", "unit": "mi", "state": "Minnesota", "phone": "(545) 249-2900", "currency": "DZD"}, {"rate": 0.35100000000000003, "email": "ipmkk@test.org", "unit": "cm", "state": "Minnesota", "phone": "(501) 003-4865", "currency": "DZD"}, {"rate": 0.52, "email": "eghjddi@example.com", "unit": "kg", "state": "California", "phone": "(513) 089-0827", "currency": "DZD"}, {"rate": 0.456, "email": "eobhj@mail.com", "unit": "cm", "state": "Montana", "phone": "(526) 954-1262", "currency": "DZD"}, {"rate": 0.512, "email": "mndjgi@test.org", "unit": "kg", "state": "Minnesota", "phone": "(549) 598-1942", "currency": "DZD"}, {"rate": 0.27399999999999997, "email": "jmblp@example.com", "unit": "lb", "state": "Minnesota", "phone": "(578) 254-5087", "currency": "BOB"}, {"rate": 0.138, "email": "phab@corp.io", "unit": "lb", "state": "Montana", "phone": "(540) 128-1935", "currency": "DZD"}, {"rate": 0.036000000000000004, "email": "ioikblg@test.org", "unit": "kg", "state": "California", "phone": "(549) 536-8239", "currency": "BOB"}, {"rate": 0.701, "email": "kpmif@test.org", "unit": "cm", "state": "Montana", "phone": "(500) 625-8784", "currency": "DZD"}, {"rate": 0.9640000000000001, "email": "kbngnlm@corp.io", "unit": "lb", "state": "California", "phone": "(542) 890-9881", "currency": "DZD"}, {"rate": 0.259, "email": "pkkogh@mail.com", "unit": "mi", "state": "Montana", "phone": "(595) 115-8345", "currency": "DZD"}, {"rate": 0.8859999999999999, "email": "laaid@mail.com", "unit": "cm", "state": "Minnesota", "phone": "(527) 217-7885", "currency": "BOB"}, {"rate": 0.616, "email": "gbcbpd@test.org", "unit": "lb", "state": "Minnesota", "phone": "(529) 794-1430", "currency": "BOB"}, {"rate": 0.706, "email": "ogfh@example.com", "unit": "kg", "state": "Montana", "phone": "(597) 470-2589", "currency": "BOB"}, {"rate": 0.818, "email": "dbnboe@test.org", "unit": "kg", "state": "California", "phone": "(566) 641-8132", "currency": "BOB"}, {"rate": 0.488, "email": "ebkank@example.com", "unit": "mi", "state": "California", "phone": "(509) 760-2256", "currency": "DZD"}, {"rate": 0.252, "email": "achk@corp.io", "unit": "lb", "state": "Montana", "phone": "(598) 661-4446", "currency": "DZD"}, {"rate": 0.636, "email": "hmhggk@test.org", "unit": "lb", "state": "Minnesota", "phone": "(570) 841-7398", "currency": "BOB"}, {"rate": 0.7020000000000001, "email": "adhpod@example.com", "unit": "mi", "state": "California", "phone": "(577) 392-6559", "currency": "DZD"}, {"rate": 0.637, "email": "pkmffpf@corp.io", "unit": "lb", "state": "Montana", "phone": "(597) 849-7075", "currency": "BOB"}, {"rate": 0.765, "email": "pmngf@test.org", "unit": "kg", "state": "Montana", "phone": "(524) 066-9502", "currency": "DZD"}, {"rate": 0.7929999999999999, "email": "piblkaj@example.com", "unit": "lb", "state": "Minnesota", "phone": "(545) 879-3239", "currency": "BOB"}, {"rate": 0.762, "email": "odpmg@test.org", "unit": "kg", "state": "Montana", "phone": "(523) 234-7639", "currency": "BOB"}], "dirty_cols": ["rate", "email", "unit", "state", "phone", "currency", "col_x"], "clean_cols": ["rate", "email", "unit", "state", "phone", "currency"], "plan": {"dataset_summary": "45 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"centimeters": "cm", "POUND": "lb", "lbs": "lb", "KG": "kg", "Mi": "mi", "Centimeter": "cm", "KILOGRAMS": "kg", "ceytimeters": "cm", "Kilograms": "kg", "Miles": "mi", "MILE": "mi", "CENTIMETERS": "cm", "miles": "mi", "Cm": "cm", "kilograms": "kg", "Pound": "lb", "Lb": "lb", "mile": "mi", "pound": "lb", "pounds": "lb"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MN": "Minnesota", "mn": "Minnesota", "CA": "California", "minnesota": "Minnesota", "ca": "California", "mt": "Montana", "MONTANA": "Montana", "MT": "Montana", "minnesoat": "Minnesota", "MINESOTA": "Minnesota", "montana": "Montana", "Mt": "Montana", "Mn": "Minnesota", "california": "California"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"boliviano": "BOB", "ALGERIAN DINAR": "DZD", "Boliviano": "BOB", "bob": "BOB", "Algerian Dinar": "DZD", "Algerian Oinar": "DZD", "dzd": "DZD", "Dzd": "DZD", "Algeiran Dinar": "DZD", "Boliviuno": "BOB", "aglerian dinar": "DZD"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"company": "Hooli  ", "amount": "2.481,65", "city": "de meern", "state": "AR", "phone": "5119635344", "col_x": ""}, {"company": "  Globex", "amount": "None", "city": "Appleton", "state": "Arkansas", "phone": "5514524412", "col_x": ""}, {"company": "Wonka Co ", "amount": "TBD", "city": "Appleton", "state": "Virgin Islands, U.S.", "phone": "(507)396-4991", "col_x": ""}, {"company": "Initech", "amount": "0", "city": "appleton", "state": "ar", "phone": "525.088.3480", "col_x": ""}, {"company": "Hooli", "amount": "$906.64", "city": "de meern", "state": "virgin islands, u.s.", "phone": "501-173-8500", "col_x": ""}, {"company": "Stark Industries", "amount": "$5,772.45", "city": "de meern", "state": "Arkansas", "phone": "(549)594-2660", "col_x": ""}, {"company": "Stark  Industries", "amount": "$3,993.56", "city": "de meern  ", "state": "Virgin Islands US", "phone": "(546)874-4270", "col_x": ""}, {"company": "Vehement", "amount": "1.897,41", "city": "Appleton", "state": "virgin islalds, u.s.", "phone": "548-402-6670", "col_x": ""}, {"company": "Stark Industries", "amount": "-", "city": "de meerrn", "state": " vi", "phone": "596.888.9614", "col_x": ""}, {"company": "Umbrella  ", "amount": "6.225,33", "city": "De MMeern", "state": "vi", "phone": "(547)706-1154", "col_x": ""}, {"company": " Stark Industries", "amount": "TBD", "city": "demeern", "state": "ar", "phone": "5969757828", "col_x": ""}, {"company": "   Cyberdyne", "amount": "4.540,97", "city": "de meern", "state": "vi", "phone": "5126997814", "col_x": ""}, {"company": "Soylent Corp   ", "amount": "$6,908.08", "city": "DE MEERN", "state": "VIRGIN ISLANDS, U.S.   ", "phone": "578-207-5557", "col_x": ""}, {"company": "   Hooli", "amount": "-", "city": "de meern", "state": "   VI", "phone": "5591626821", "col_x": ""}, {"company": "Acme Inc", "amount": "5.473,16", "city": "Appleton   ", "state": "AR", "phone": "540.362.0217", "col_x": ""}, {"company": "   Acme Inc", "amount": "4.972,30", "city": "Appleton", "state": "AR   ", "phone": "571.177.2675", "col_x": ""}, {"company": "Umbrella", "amount": "null", "city": "De Meern", "state": "VI  ", "phone": "5199143474", "col_x": ""}, {"company": "Acme Inc", "amount": "$5,326.71", "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "5426735867", "col_x": ""}, {"company": "   Globex", "amount": "7.464,14", "city": "   de meern", "state": "VI   ", "phone": "525.626.3461", "col_x": ""}, {"company": "Hooli", "amount": "$4,064.43", "city": "  Appleton", "state": "Virgin Islnads, U.S.", "phone": "524.225.2496", "col_x": ""}], "clean": [{"company": "Hooli", "amount": 2481.65, "city": "De Meern", "state": "Arkansas", "phone": "(511) 963-5344"}, {"company": "Globex", "amount": NaN, "city": "Appleton", "state": "Arkansas", "phone": "(551) 452-4412"}, {"company": "Wonka Co", "amount": NaN, "city": "Appleton", "state": "Virgin Islands, U.S.", "phone": "(507) 396-4991"}, {"company": "Initech", "amount": 0.0, "city": "Appleton", "state": "Arkansas", "phone": "(525) 088-3480"}, {"company": "Hooli", "amount": 906.64, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(501) 173-8500"}, {"company": "Stark Industries", "amount": 5772.45, "city": "De Meern", "state": "Arkansas", "phone": "(549) 594-2660"}, {"company": "Stark Industries", "amount": 3993.56, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(546) 874-4270"}, {"company": "Vehement", "amount": 1897.41, "city": "Appleton", "state": "Virgin Islands, U.S.", "phone": "(548) 402-6670"}, {"company": "Stark Industries", "amount": NaN, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(596) 888-9614"}, {"company": "Umbrella", "amount": 6225.33, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(547) 706-1154"}, {"company": "Stark Industries", "amount": NaN, "city": "De Meern", "state": "Arkansas", "phone": "(596) 975-7828"}, {"company": "Cyberdyne", "amount": 4540.97, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(512) 699-7814"}, {"company": "Soylent Corp", "amount": 6908.08, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(578) 207-5557"}, {"company": "Hooli", "amount": NaN, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(559) 162-6821"}, {"company": "Acme Inc", "amount": 5473.16, "city": "Appleton", "state": "Arkansas", "phone": "(540) 362-0217"}, {"company": "Acme Inc", "amount": 4972.3, "city": "Appleton", "state": "Arkansas", "phone": "(571) 177-2675"}, {"company": "Umbrella", "amount": NaN, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(519) 914-3474"}, {"company": "Acme Inc", "amount": 5326.71, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(542) 673-5867"}, {"company": "Globex", "amount": 7464.14, "city": "De Meern", "state": "Virgin Islands, U.S.", "phone": "(525) 626-3461"}, {"company": "Hooli", "amount": 4064.43, "city": "Appleton", "state": "Virgin Islands, U.S.", "phone": "(524) 225-2496"}], "dirty_cols": ["company", "amount", "city", "state", "phone", "col_x"], "clean_cols": ["company", "amount", "city", "state", "phone"], "plan": {"dataset_summary": "20 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"de meern": "De Meern", "appleton": "Appleton", "de meerrn": "De Meern", "De MMeern": "De Meern", "demeern": "De Meern", "DE MEERN": "De Meern"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AR": "Arkansas", "ar": "Arkansas", "virgin islands, u.s.": "Virgin Islands, U.S.", "Virgin Islands US": "Virgin Islands, U.S.", "virgin islalds, u.s.": "Virgin Islands, U.S.", "vi": "Virgin Islands, U.S.", "VIRGIN ISLANDS, U.S.": "Virgin Islands, U.S.", "VI": "Virgin Islands, U.S.", "Virgin Islnads, U.S.": "Virgin Islands, U.S."}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"name": "Eve  Diaz", "status": "high", "phone": "577-951-1765", "unit": "ml", "is_active": "FALSE", "rate": "55.9%", "signup_date": "2023-08-21", "email": "BKLMDED@EXAMPLE.COM"}, {"name": "Judy Diaz", "status": "Medium", "phone": "557.543.9387", "unit": "milliliter", "is_active": "Yes", "rate": "66.5%", "signup_date": "5 Aug 2023", "email": "KLFLOF@TEST.ORG"}, {"name": "Omar Johnson", "status": "H", "phone": "(502)869-3368", "unit": "mi", "is_active": "T", "rate": "70.0%", "signup_date": "3/8/2023", "email": "ehgpjk@mail.com"}, {"name": "Frank Reyes", "status": "med", "phone": "5809817319", "unit": "LBS  ", "is_active": "Yes", "rate": "97.4%", "signup_date": "45207", "email": "ENOFCHG@EXAMPLE.COM"}, {"name": "Sara Moore", "status": "P2", "phone": "512-254-6769", "unit": "ml", "is_active": "false", "rate": "29.9%", "signup_date": "45093", "email": "  obnbooa@example.com"}, {"name": "Alice Diaz", "status": "med", "phone": "(528)839-5297", "unit": "MI", "is_active": "N", "rate": "30.9%", "signup_date": "2/11/2023", "email": "hkcmnpg@example.com"}, {"name": "Priya Moore ", "status": "high", "phone": "5557417273", "unit": " mi", "is_active": "Y", "rate": "88.5%", "signup_date": "2023-02-14", "email": "HFHPLEH@CORP.IO"}, {"name": "Heidi Lee", "status": "L", "phone": "(534)104-0312", "unit": "milliliter", "is_active": "Yes", "rate": "99.2%", "signup_date": "4/9/2023", "email": " jhbjic@test.org "}, {"name": "Lena Fischer   ", "status": "P3", "phone": "587-764-7097", "unit": "ml", "is_active": "Yes", "rate": "38.6%", "signup_date": "2023-04-08", "email": "NOCHPEH@TEST.ORG"}, {"name": "Grace Lee", "status": "medium", "phone": "569-777-1755", "unit": "mi", "is_active": "Yes", "rate": "60.1%", "signup_date": "6 Oct 2023", "email": " KAJFDLE@CORP.IO "}, {"name": "Mona Khan", "status": "P2", "phone": "515.343.1745", "unit": "   milliliter", "is_active": "Yes", "rate": "44.4%", "signup_date": "8/25/2023", "email": "kclfl@example.com"}, {"name": "Judy Petrov", "status": " low", "phone": "560.639.4168", "unit": "lb", "is_active": "1", "rate": "2.6%", "signup_date": "7/14/2023", "email": "padh@example.com"}, {"name": "  Ivan Novak", "status": "medium", "phone": "522.846.7612", "unit": "lb", "is_active": "1", "rate": "70.6%", "signup_date": "2023-04-04", "email": "akmhfd@mail.com"}, {"name": "Priya Adams", "status": "low", "phone": "5108831712", "unit": "mls", "is_active": "F", "rate": "51.8%", "signup_date": "3/20/2023", "email": " oeoddjh@example.com"}, {"name": " Mona Petrov", "status": "MEDIUM", "phone": "5272286150", "unit": "ML", "is_active": "FALSE", "rate": "43.7%", "signup_date": "2023-07-02", "email": "PABB@CORP.IO"}, {"name": "   Judy Cruz", "status": "Medium", "phone": "(588)158-7051", "unit": "lb", "is_active": "T", "rate": "92.8%", "signup_date": "45061", "email": "bkecjn@mail.com"}, {"name": "Bob Adams", "status": "H", "phone": "549.068.9509", "unit": " mL", "is_active": "Yes", "rate": "75.8%", "signup_date": "2023-10-19", "email": " JFGMMI@TEST.ORG "}, {"name": "Ivan Johnson", "status": "P1", "phone": "590-166-3288", "unit": "miles  ", "is_active": "T", "rate": "58.8%", "signup_date": "2023-06-16", "email": "ahjkgc@example.com"}, {"name": "Karl Johnson", "status": " Low ", "phone": "(570)209-6144", "unit": "lb", "is_active": "N", "rate": "32.1%", "signup_date": "2023-08-06", "email": "mggl@example.com"}, {"name": "Grace Smith   ", "status": "Low", "phone": "519.801.2427", "unit": "Mi", "is_active": "TRUE", "rate": "65.4%", "signup_date": "2023-11-26", "email": "AJKKFI@EXAMPLE.COM"}, {"name": "Heidi Park ", "status": "H", "phone": "5408885328", "unit": "mi", "is_active": "F", "rate": "34.1%", "signup_date": "2023-05-12", "email": "kepccfh@test.org "}, {"name": "Carol Ali", "status": "LOW", "phone": "5750716581", "unit": "ml", "is_active": "false", "rate": "95.9%", "signup_date": "45142", "email": "JILPH@EXAMPLE.COM"}, {"name": " Ivan Diaz", "status": "M", "phone": "521.415.0152", "unit": "lb", "is_active": "1", "rate": "0.4%", "signup_date": "45048", "email": "DGPDGD@MAIL.COM"}, {"name": "Judy Moore", "status": "L", "phone": "579-776-6248", "unit": " pounds ", "is_active": "TRUE", "rate": "55.2%", "signup_date": "7/1/2023", "email": "HDDGLAN@MAIL.COM"}, {"name": "Heidi Johnson", "status": "low   ", "phone": "(593)278-5920", "unit": "miles", "is_active": "false", "rate": "63.7%", "signup_date": "8/4/2023", "email": "GBJDDA@TEST.ORG"}, {"name": "Alice Lee ", "status": "high", "phone": "5554383209", "unit": "mile", "is_active": "F", "rate": "89.0%", "signup_date": "2023-05-12", "email": "fkig@corp.io"}, {"name": "Karl  Petrov", "status": "HIGH", "phone": "512.588.1177", "unit": " mile ", "is_active": "F", "rate": "16.6%", "signup_date": "45039", "email": " naoj@test.org "}, {"name": "Carol Cruz", "status": "High", "phone": "(597)710-1528", "unit": "miles ", "is_active": "Y", "rate": "24.0%", "signup_date": "2023-04-27", "email": "nfgplmf@example.com "}, {"name": "Karl Ali", "status": "high", "phone": "5523892849", "unit": "lb", "is_active": "TRUE", "rate": "94.9%", "signup_date": "2023-05-22", "email": " AJMP@EXAMPLE.COM"}, {"name": "Priya Smith", "status": "Medium", "phone": "572-932-1358", "unit": "POUND", "is_active": "false", "rate": "72.4%", "signup_date": "45160", "email": "echl@test.org"}, {"name": "   Ivan Park", "status": "  LOW", "phone": "531.953.6259", "unit": "Lb", "is_active": "Y", "rate": "53.1%", "signup_date": "2023-02-14", "email": "LCNDK@TEST.ORG"}, {"name": "  Bob Smith", "status": "low", "phone": "576.587.9247", "unit": "Mi", "is_active": "Yes", "rate": "31.3%", "signup_date": "2023-10-23", "email": "liepl@test.org"}, {"name": "Mona Petrov", "status": "M", "phone": "(549)766-5950", "unit": "MI", "is_active": "0", "rate": "73.0%", "signup_date": "7/2/2023", "email": "  incp@corp.io"}, {"name": "Grace Novak", "status": "M   ", "phone": "511.309.1873", "unit": "Lbs", "is_active": "1", "rate": "53.6%", "signup_date": "2023-08-14", "email": "NOCML@EXAMPLE.COM"}, {"name": "Priya Park", "status": "high", "phone": "(529)797-6107", "unit": "pound", "is_active": "true", "rate": "49.9%", "signup_date": "45001", "email": "MCCCN@MAIL.COM"}, {"name": "Bob Cruz", "status": "med", "phone": "532-141-9273", "unit": "mL", "is_active": "Y", "rate": "89.0%", "signup_date": "10/25/2023", "email": " DCDEA@EXAMPLE.COM"}, {"name": "Bob  Adams", "status": "M", "phone": "571-484-2033", "unit": " milliliter ", "is_active": "T", "rate": "86.1%", "signup_date": "9/25/2023", "email": " OPIFC@EXAMPLE.COM "}, {"name": "David  Petrov", "status": "low", "phone": "579.827.1265", "unit": "Ml", "is_active": "false", "rate": "55.1%", "signup_date": "4/1/2023", "email": "gfhohof@corp.io  "}, {"name": "Judy  Reyes", "status": "M", "phone": "568.252.8362", "unit": "mies", "is_active": "N", "rate": "11.6%", "signup_date": "8/5/2023", "email": "ONNPCN@TEST.ORG"}, {"name": "Mona Lee", "status": "low ", "phone": "(554)479-4080", "unit": "Lb", "is_active": "false", "rate": "14.6%", "signup_date": "45096", "email": "JCNC@CORP.IO"}, {"name": "Carol Fischer ", "status": "HIGH", "phone": "503-122-9763", "unit": " ml ", "is_active": "F", "rate": "82.5%", "signup_date": "45244", "email": "jhpn@corp.io"}, {"name": "Carol Khan ", "status": "medium", "phone": "535.158.5281", "unit": "ml", "is_active": "0", "rate": "30.6%", "signup_date": "27 Mar 2023", "email": "HMIK@EXAMPLE.COM"}, {"name": "   Eve Lee", "status": "low", "phone": "521-410-5925", "unit": "Mi", "is_active": "Y", "rate": "89.7%", "signup_date": "2023-02-06", "email": "JNKF@CORP.IO"}, {"name": "  Alice Novak", "status": "High", "phone": "595.315.7277", "unit": "milliliter", "is_active": "1", "rate": "73.5%", "signup_date": "45173", "email": "oecpc@mail.com"}, {"name": "  Karl Brandt", "status": "L", "phone": "534-762-6325", "unit": "mls", "is_active": "false", "rate": "74.0%", "signup_date": "4 Aug 2023", "email": " IGFGIA@EXAMPLE.COM "}, {"name": "Heidi  Lee", "status": "Low", "phone": "(536)491-7042", "unit": " miles ", "is_active": "No", "rate": "40.3%", "signup_date": "2023-10-27", "email": "fjfjjph@test.org"}, {"name": "Bob Novak", "status": "low", "phone": "517-719-7475", "unit": "lbs  ", "is_active": "N", "rate": "63.0%", "signup_date": "45124", "email": "CGGEID@CORP.IO"}, {"name": "Omar  Smith", "status": "l", "phone": "5045242810", "unit": "mL", "is_active": "1", "rate": "26.9%", "signup_date": "45182", "email": " KMBLIFC@MAIL.COM "}, {"name": "Grace Johnson", "status": "HIGH  ", "phone": "500-351-9501", "unit": "miles", "is_active": "TRUE", "rate": "98.1%", "signup_date": "2023-03-26", "email": "NEPM@EXAMPLE.COM  "}, {"name": "Mona Ali", "status": "HIGH", "phone": "5981843369", "unit": "ml  ", "is_active": "0", "rate": "40.4%", "signup_date": "4/18/2023", "email": "belkl@corp.io"}, {"name": "Bob Moore  ", "status": "low", "phone": "596-727-7310", "unit": "millilietr", "is_active": "false", "rate": "71.5%", "signup_date": "8 Oct 2023", "email": " dlojaa@corp.io "}, {"name": "  Lena Moore", "status": "low", "phone": "(595)008-7546", "unit": "ml", "is_active": "FALSE", "rate": "85.1%", "signup_date": "2023-06-21", "email": "JEECAG@CORP.IO"}, {"name": "Karl Petrov ", "status": "l", "phone": "530-566-6836", "unit": "mi", "is_active": "1", "rate": "87.8%", "signup_date": "45200", "email": "  DGFPCNH@MAIL.COM"}, {"name": "", "status": "", "phone": "", "unit": "", "is_active": "", "rate": "", "signup_date": "", "email": ""}], "clean": [{"name": "Eve Diaz", "status": "High", "phone": "(577) 951-1765", "unit": "mL", "is_active": false, "rate": 0.5589999999999999, "signup_date": "2023-08-21", "email": "bklmded@example.com"}, {"name": "Judy Diaz", "status": "Medium", "phone": "(557) 543-9387", "unit": "mL", "is_active": true, "rate": 0.665, "signup_date": "2023-08-05", "email": "klflof@test.org"}, {"name": "Omar Johnson", "status": "High", "phone": "(502) 869-3368", "unit": "mi", "is_active": true, "rate": 0.7, "signup_date": "2023-03-08", "email": "ehgpjk@mail.com"}, {"name": "Frank Reyes", "status": "Medium", "phone": "(580) 981-7319", "unit": "lb", "is_active": true, "rate": 0.9740000000000001, "signup_date": "2023-10-08", "email": "enofchg@example.com"}, {"name": "Sara Moore", "status": "Medium", "phone": "(512) 254-6769", "unit": "mL", "is_active": false, "rate": 0.299, "signup_date": "2023-06-16", "email": "obnbooa@example.com"}, {"name": "Alice Diaz", "status": "Medium", "phone": "(528) 839-5297", "unit": "mi", "is_active": false, "rate": 0.309, "signup_date": "2023-02-11", "email": "hkcmnpg@example.com"}, {"name": "Priya Moore", "status": "High", "phone": "(555) 741-7273", "unit": "mi", "is_active": true, "rate": 0.885, "signup_date": "2023-02-14", "email": "hfhpleh@corp.io"}, {"name": "Heidi Lee", "status": "Low", "phone": "(534) 104-0312", "unit": "mL", "is_active": true, "rate": 0.992, "signup_date": "2023-04-09", "email": "jhbjic@test.org"}, {"name": "Lena Fischer", "status": "Low", "phone": "(587) 764-7097", "unit": "mL", "is_active": true, "rate": 0.386, "signup_date": "2023-04-08", "email": "nochpeh@test.org"}, {"name": "Grace Lee", "status": "Medium", "phone": "(569) 777-1755", "unit": "mi", "is_active": true, "rate": 0.601, "signup_date": "2023-10-06", "email": "kajfdle@corp.io"}, {"name": "Mona Khan", "status": "Medium", "phone": "(515) 343-1745", "unit": "mL", "is_active": true, "rate": 0.444, "signup_date": "2023-08-25", "email": "kclfl@example.com"}, {"name": "Judy Petrov", "status": "Low", "phone": "(560) 639-4168", "unit": "lb", "is_active": true, "rate": 0.026000000000000002, "signup_date": "2023-07-14", "email": "padh@example.com"}, {"name": "Ivan Novak", "status": "Medium", "phone": "(522) 846-7612", "unit": "lb", "is_active": true, "rate": 0.706, "signup_date": "2023-04-04", "email": "akmhfd@mail.com"}, {"name": "Priya Adams", "status": "Low", "phone": "(510) 883-1712", "unit": "mL", "is_active": false, "rate": 0.518, "signup_date": "2023-03-20", "email": "oeoddjh@example.com"}, {"name": "Mona Petrov", "status": "Medium", "phone": "(527) 228-6150", "unit": "mL", "is_active": false, "rate": 0.43700000000000006, "signup_date": "2023-07-02", "email": "pabb@corp.io"}, {"name": "Judy Cruz", "status": "Medium", "phone": "(588) 158-7051", "unit": "lb", "is_active": true, "rate": 0.9279999999999999, "signup_date": "2023-05-15", "email": "bkecjn@mail.com"}, {"name": "Bob Adams", "status": "High", "phone": "(549) 068-9509", "unit": "mL", "is_active": true, "rate": 0.758, "signup_date": "2023-10-19", "email": "jfgmmi@test.org"}, {"name": "Ivan Johnson", "status": "High", "phone": "(590) 166-3288", "unit": "mi", "is_active": true, "rate": 0.588, "signup_date": "2023-06-16", "email": "ahjkgc@example.com"}, {"name": "Karl Johnson", "status": "Low", "phone": "(570) 209-6144", "unit": "lb", "is_active": false, "rate": 0.321, "signup_date": "2023-08-06", "email": "mggl@example.com"}, {"name": "Grace Smith", "status": "Low", "phone": "(519) 801-2427", "unit": "mi", "is_active": true, "rate": 0.654, "signup_date": "2023-11-26", "email": "ajkkfi@example.com"}, {"name": "Heidi Park", "status": "High", "phone": "(540) 888-5328", "unit": "mi", "is_active": false, "rate": 0.341, "signup_date": "2023-05-12", "email": "kepccfh@test.org"}, {"name": "Carol Ali", "status": "Low", "phone": "(575) 071-6581", "unit": "mL", "is_active": false, "rate": 0.9590000000000001, "signup_date": "2023-08-04", "email": "jilph@example.com"}, {"name": "Ivan Diaz", "status": "Medium", "phone": "(521) 415-0152", "unit": "lb", "is_active": true, "rate": 0.004, "signup_date": "2023-05-02", "email": "dgpdgd@mail.com"}, {"name": "Judy Moore", "status": "Low", "phone": "(579) 776-6248", "unit": "lb", "is_active": true, "rate": 0.552, "signup_date": "2023-07-01", "email": "hddglan@mail.com"}, {"name": "Heidi Johnson", "status": "Low", "phone": "(593) 278-5920", "unit": "mi", "is_active": false, "rate": 0.637, "signup_date": "2023-08-04", "email": "gbjdda@test.org"}, {"name": "Alice Lee", "status": "High", "phone": "(555) 438-3209", "unit": "mi", "is_active": false, "rate": 0.89, "signup_date": "2023-05-12", "email": "fkig@corp.io"}, {"name": "Karl Petrov", "status": "High", "phone": "(512) 588-1177", "unit": "mi", "is_active": false, "rate": 0.166, "signup_date": "2023-04-23", "email": "naoj@test.org"}, {"name": "Carol Cruz", "status": "High", "phone": "(597) 710-1528", "unit": "mi", "is_active": true, "rate": 0.24, "signup_date": "2023-04-27", "email": "nfgplmf@example.com"}, {"name": "Karl Ali", "status": "High", "phone": "(552) 389-2849", "unit": "lb", "is_active": true, "rate": 0.9490000000000001, "signup_date": "2023-05-22", "email": "ajmp@example.com"}, {"name": "Priya Smith", "status": "Medium", "phone": "(572) 932-1358", "unit": "lb", "is_active": false, "rate": 0.7240000000000001, "signup_date": "2023-08-22", "email": "echl@test.org"}, {"name": "Ivan Park", "status": "Low", "phone": "(531) 953-6259", "unit": "lb", "is_active": true, "rate": 0.531, "signup_date": "2023-02-14", "email": "lcndk@test.org"}, {"name": "Bob Smith", "status": "Low", "phone": "(576) 587-9247", "unit": "mi", "is_active": true, "rate": 0.313, "signup_date": "2023-10-23", "email": "liepl@test.org"}, {"name": "Mona Petrov", "status": "Medium", "phone": "(549) 766-5950", "unit": "mi", "is_active": false, "rate": 0.73, "signup_date": "2023-07-02", "email": "incp@corp.io"}, {"name": "Grace Novak", "status": "Medium", "phone": "(511) 309-1873", "unit": "lb", "is_active": true, "rate": 0.536, "signup_date": "2023-08-14", "email": "nocml@example.com"}, {"name": "Priya Park", "status": "High", "phone": "(529) 797-6107", "unit": "lb", "is_active": true, "rate": 0.499, "signup_date": "2023-03-16", "email": "mcccn@mail.com"}, {"name": "Bob Cruz", "status": "Medium", "phone": "(532) 141-9273", "unit": "mL", "is_active": true, "rate": 0.89, "signup_date": "2023-10-25", "email": "dcdea@example.com"}, {"name": "Bob Adams", "status": "Medium", "phone": "(571) 484-2033", "unit": "mL", "is_active": true, "rate": 0.861, "signup_date": "2023-09-25", "email": "opifc@example.com"}, {"name": "David Petrov", "status": "Low", "phone": "(579) 827-1265", "unit": "mL", "is_active": false, "rate": 0.551, "signup_date": "2023-04-01", "email": "gfhohof@corp.io"}, {"name": "Judy Reyes", "status": "Medium", "phone": "(568) 252-8362", "unit": "mi", "is_active": false, "rate": 0.11599999999999999, "signup_date": "2023-08-05", "email": "onnpcn@test.org"}, {"name": "Mona Lee", "status": "Low", "phone": "(554) 479-4080", "unit": "lb", "is_active": false, "rate": 0.146, "signup_date": "2023-06-19", "email": "jcnc@corp.io"}, {"name": "Carol Fischer", "status": "High", "phone": "(503) 122-9763", "unit": "mL", "is_active": false, "rate": 0.825, "signup_date": "2023-11-14", "email": "jhpn@corp.io"}, {"name": "Carol Khan", "status": "Medium", "phone": "(535) 158-5281", "unit": "mL", "is_active": false, "rate": 0.306, "signup_date": "2023-03-27", "email": "hmik@example.com"}, {"name": "Eve Lee", "status": "Low", "phone": "(521) 410-5925", "unit": "mi", "is_active": true, "rate": 0.897, "signup_date": "2023-02-06", "email": "jnkf@corp.io"}, {"name": "Alice Novak", "status": "High", "phone": "(595) 315-7277", "unit": "mL", "is_active": true, "rate": 0.735, "signup_date": "2023-09-04", "email": "oecpc@mail.com"}, {"name": "Karl Brandt", "status": "Low", "phone": "(534) 762-6325", "unit": "mL", "is_active": false, "rate": 0.74, "signup_date": "2023-08-04", "email": "igfgia@example.com"}, {"name": "Heidi Lee", "status": "Low", "phone": "(536) 491-7042", "unit": "mi", "is_active": false, "rate": 0.40299999999999997, "signup_date": "2023-10-27", "email": "fjfjjph@test.org"}, {"name": "Bob Novak", "status": "Low", "phone": "(517) 719-7475", "unit": "lb", "is_active": false, "rate": 0.63, "signup_date": "2023-07-17", "email": "cggeid@corp.io"}, {"name": "Omar Smith", "status": "Low", "phone": "(504) 524-2810", "unit": "mL", "is_active": true, "rate": 0.26899999999999996, "signup_date": "2023-09-13", "email": "kmblifc@mail.com"}, {"name": "Grace Johnson", "status": "High", "phone": "(500) 351-9501", "unit": "mi", "is_active": true, "rate": 0.981, "signup_date": "2023-03-26", "email": "nepm@example.com"}, {"name": "Mona Ali", "status": "High", "phone": "(598) 184-3369", "unit": "mL", "is_active": false, "rate": 0.40399999999999997, "signup_date": "2023-04-18", "email": "belkl@corp.io"}, {"name": "Bob Moore", "status": "Low", "phone": "(596) 727-7310", "unit": "mL", "is_active": false, "rate": 0.715, "signup_date": "2023-10-08", "email": "dlojaa@corp.io"}, {"name": "Lena Moore", "status": "Low", "phone": "(595) 008-7546", "unit": "mL", "is_active": false, "rate": 0.851, "signup_date": "2023-06-21", "email": "jeecag@corp.io"}, {"name": "Karl Petrov", "status": "Low", "phone": "(530) 566-6836", "unit": "mi", "is_active": true, "rate": 0.878, "signup_date": "2023-10-01", "email": "dgfpcnh@mail.com"}], "dirty_cols": ["name", "status", "phone", "unit", "is_active", "rate", "signup_date", "email"], "clean_cols": ["name", "status", "phone", "unit", "is_active", "rate", "signup_date", "email"], "plan": {"dataset_summary": "54 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"high": "High", "H": "High", "med": "Medium", "P2": "Medium", "L": "Low", "P3": "Low", "medium": "Medium", "low": "Low", "MEDIUM": "Medium", "P1": "High", "LOW": "Low", "M": "Medium", "HIGH": "High", "l": "Low"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ml": "mL", "milliliter": "mL", "LBS": "lb", "MI": "mi", "mls": "mL", "ML": "mL", "miles": "mi", "Mi": "mi", "pounds": "lb", "mile": "mi", "POUND": "lb", "Lb": "lb", "Lbs": "lb", "pound": "lb", "Ml": "mL", "mies": "mi", "lbs": "lb", "millilietr": "mL"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"job_title": "dir", "state": "MD", "rate": "7.6%", "industry": " Tourism "}, {"job_title": "dir", "state": "mt", "rate": "40.9%", "industry": "Mfg"}, {"job_title": "DYR.", "state": "west virginia", "rate": "95.7%", "industry": "Hotels & Travel"}, {"job_title": "Director  ", "state": "New Mexico", "rate": "38.4%", "industry": "INDUSTRIAL"}, {"job_title": "CTO", "state": "  WV", "rate": "87.1%", "industry": "TRAEVL"}, {"job_title": "Admin Assistant", "state": "MD", "rate": "7.9%", "industry": "Tech"}, {"job_title": "  CTO", "state": "La", "rate": "3.7%", "industry": "manufacturing"}, {"job_title": "Chief Executive Officer", "state": "LA", "rate": "65.4%", "industry": "Travel"}, {"job_title": "Dir  ", "state": "   louisiana", "rate": "0.4%", "industry": " SOFTWARE"}, {"job_title": " C.E.O.", "state": "LA", "rate": "60.4%", "industry": "MANUFACTURING"}, {"job_title": "Dir.", "state": "west virginia", "rate": "6.7%", "industry": "Software"}, {"job_title": "CEO", "state": "MD", "rate": "60.5%", "industry": "tech"}, {"job_title": "Dir", "state": "MD   ", "rate": "49.3%", "industry": "Hotels & Travel"}, {"job_title": "C..E.O.", "state": "West Virginia", "rate": "82.8%", "industry": "Tfch  "}, {"job_title": "Dir", "state": "Nm", "rate": "89.8%", "industry": "tech   "}, {"job_title": "DIR ", "state": "West Virginia", "rate": "5.1%", "industry": "Technology"}, {"job_title": "Chief Executive Officer", "state": "Montana", "rate": "90.5%", "industry": "HOSPIATLITY"}, {"job_title": "CHIEF TECHNOLOGY OFFICER", "state": "Maryland", "rate": "30.0%", "industry": "tech"}, {"job_title": "C.T.O.", "state": "Md", "rate": "44.7%", "industry": "MFG"}, {"job_title": "C.T.O.", "state": "   WV", "rate": "79.8%", "industry": "manufacturing"}, {"job_title": "chief technology officer", "state": "MD", "rate": "99.2%", "industry": "Manufacturing"}, {"job_title": "C.T.O.", "state": "   WV", "rate": "79.8%", "industry": "manufacturing"}, {"job_title": "C.T.O.", "state": "Md", "rate": "44.7%", "industry": "MFG"}], "clean": [{"job_title": "Director", "state": "Maryland", "rate": 0.076, "industry": "Hospitality"}, {"job_title": "Director", "state": "Montana", "rate": 0.409, "industry": "Manufacturing"}, {"job_title": "Director", "state": "West Virginia", "rate": 0.9570000000000001, "industry": "Hospitality"}, {"job_title": "Director", "state": "New Mexico", "rate": 0.384, "industry": "Manufacturing"}, {"job_title": "Chief Technology Officer", "state": "West Virginia", "rate": 0.871, "industry": "Hospitality"}, {"job_title": "Administrative Assistant", "state": "Maryland", "rate": 0.079, "industry": "Technology"}, {"job_title": "Chief Technology Officer", "state": "Louisiana", "rate": 0.037000000000000005, "industry": "Manufacturing"}, {"job_title": "Chief Executive Officer", "state": "Louisiana", "rate": 0.654, "industry": "Hospitality"}, {"job_title": "Director", "state": "Louisiana", "rate": 0.004, "industry": "Technology"}, {"job_title": "Chief Executive Officer", "state": "Louisiana", "rate": 0.604, "industry": "Manufacturing"}, {"job_title": "Director", "state": "West Virginia", "rate": 0.067, "industry": "Technology"}, {"job_title": "Chief Executive Officer", "state": "Maryland", "rate": 0.605, "industry": "Technology"}, {"job_title": "Director", "state": "Maryland", "rate": 0.493, "industry": "Hospitality"}, {"job_title": "Chief Executive Officer", "state": "West Virginia", "rate": 0.828, "industry": "Technology"}, {"job_title": "Director", "state": "New Mexico", "rate": 0.898, "industry": "Technology"}, {"job_title": "Director", "state": "West Virginia", "rate": 0.051, "industry": "Technology"}, {"job_title": "Chief Executive Officer", "state": "Montana", "rate": 0.905, "industry": "Hospitality"}, {"job_title": "Chief Technology Officer", "state": "Maryland", "rate": 0.3, "industry": "Technology"}, {"job_title": "Chief Technology Officer", "state": "Maryland", "rate": 0.447, "industry": "Manufacturing"}, {"job_title": "Chief Technology Officer", "state": "West Virginia", "rate": 0.7979999999999999, "industry": "Manufacturing"}, {"job_title": "Chief Technology Officer", "state": "Maryland", "rate": 0.992, "industry": "Manufacturing"}], "dirty_cols": ["job_title", "state", "rate", "industry"], "clean_cols": ["job_title", "state", "rate", "industry"], "plan": {"dataset_summary": "23 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"dir": "Director", "DYR.": "Director", "CTO": "Chief Technology Officer", "Admin Assistant": "Administrative Assistant", "Dir": "Director", "C.E.O.": "Chief Executive Officer", "Dir.": "Director", "CEO": "Chief Executive Officer", "C..E.O.": "Chief Executive Officer", "DIR": "Director", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MD": "Maryland", "mt": "Montana", "west virginia": "West Virginia", "WV": "West Virginia", "La": "Louisiana", "LA": "Louisiana", "louisiana": "Louisiana", "Nm": "New Mexico", "Md": "Maryland"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Tourism": "Hospitality", "Mfg": "Manufacturing", "Hotels & Travel": "Hospitality", "INDUSTRIAL": "Manufacturing", "TRAEVL": "Hospitality", "Tech": "Technology", "manufacturing": "Manufacturing", "Travel": "Hospitality", "SOFTWARE": "Technology", "MANUFACTURING": "Manufacturing", "Software": "Technology", "tech": "Technology", "Tfch": "Technology", "HOSPIATLITY": "Hospitality", "MFG": "Manufacturing"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"industry": "REALESTATE", "country": "Côte D'Ivoire", "phone": "(541)217-6258", "job_title": "Prod Manager"}, {"industry": " Edtech", "country": "Republic of Singapore", "phone": "5330267443", "job_title": "Product Manager "}, {"industry": "PROPTECH", "country": "CI", "phone": "567-824-2267", "job_title": "CTO"}, {"industry": "Property", "country": "CIV ", "phone": "599.393.2640", "job_title": "Chief Technology Officer"}, {"industry": "education", "country": "sg", "phone": "(590)065-7921", "job_title": "CTO"}, {"industry": "education", "country": "  flk", "phone": "508.407.7632", "job_title": "engineering mgr"}, {"industry": "Education", "country": "côte d'ivoire", "phone": "(598)659-6395", "job_title": "PM"}, {"industry": " Edtech ", "country": "   FK", "phone": "5495295919", "job_title": "chief technology officer"}, {"industry": "realestate", "country": "falkland islands (malvinas)", "phone": "5491764307", "job_title": "C.T.O."}, {"industry": "education ", "country": "CI", "phone": "543.727.1575", "job_title": "eng mgr"}, {"industry": "EDTECH", "country": "FLK", "phone": "(566)761-4966", "job_title": "   Cto"}, {"industry": "ed", "country": "Côte d'Ivoire   ", "phone": "572-248-8606", "job_title": "Eng Mgr"}, {"industry": "ed", "country": "côte dwivoire", "phone": "5815379285", "job_title": "Engineering Mgr"}, {"industry": "Property", "country": "Republic of Singapore", "phone": "575-481-9753", "job_title": "Chief Technology Officer"}, {"industry": "edtech", "country": "flk ", "phone": "564-413-7756", "job_title": "Engineering Manager"}, {"industry": "Property ", "country": "CÔTE D'IVOIRE", "phone": "500-685-8614", "job_title": "ENG MGR"}, {"industry": "rbalestate", "country": "FK", "phone": "548.339.7942", "job_title": "pm"}, {"industry": " education ", "country": "Falkland Islands (Malvinas)", "phone": "508-481-1556", "job_title": "Eng Manager"}, {"industry": "Retail", "country": "FLK", "phone": "5551055825", "job_title": " cto "}, {"industry": "REALESTATE", "country": "Côte D'Ivoire", "phone": "(541)217-6258", "job_title": "Prod Manager"}], "clean": [{"industry": "Real Estate", "country": "Côte d'Ivoire", "phone": "(541) 217-6258", "job_title": "Product Manager"}, {"industry": "Education", "country": "Singapore", "phone": "(533) 026-7443", "job_title": "Product Manager"}, {"industry": "Real Estate", "country": "Côte d'Ivoire", "phone": "(567) 824-2267", "job_title": "Chief Technology Officer"}, {"industry": "Real Estate", "country": "Côte d'Ivoire", "phone": "(599) 393-2640", "job_title": "Chief Technology Officer"}, {"industry": "Education", "country": "Singapore", "phone": "(590) 065-7921", "job_title": "Chief Technology Officer"}, {"industry": "Education", "country": "Falkland Islands (Malvinas)", "phone": "(508) 407-7632", "job_title": "Engineering Manager"}, {"industry": "Education", "country": "Côte d'Ivoire", "phone": "(598) 659-6395", "job_title": "Product Manager"}, {"industry": "Education", "country": "Falkland Islands (Malvinas)", "phone": "(549) 529-5919", "job_title": "Chief Technology Officer"}, {"industry": "Real Estate", "country": "Falkland Islands (Malvinas)", "phone": "(549) 176-4307", "job_title": "Chief Technology Officer"}, {"industry": "Education", "country": "Côte d'Ivoire", "phone": "(543) 727-1575", "job_title": "Engineering Manager"}, {"industry": "Education", "country": "Falkland Islands (Malvinas)", "phone": "(566) 761-4966", "job_title": "Chief Technology Officer"}, {"industry": "Education", "country": "Côte d'Ivoire", "phone": "(572) 248-8606", "job_title": "Engineering Manager"}, {"industry": "Education", "country": "Côte d'Ivoire", "phone": "(581) 537-9285", "job_title": "Engineering Manager"}, {"industry": "Real Estate", "country": "Singapore", "phone": "(575) 481-9753", "job_title": "Chief Technology Officer"}, {"industry": "Education", "country": "Falkland Islands (Malvinas)", "phone": "(564) 413-7756", "job_title": "Engineering Manager"}, {"industry": "Real Estate", "country": "Côte d'Ivoire", "phone": "(500) 685-8614", "job_title": "Engineering Manager"}, {"industry": "Real Estate", "country": "Falkland Islands (Malvinas)", "phone": "(548) 339-7942", "job_title": "Product Manager"}, {"industry": "Education", "country": "Falkland Islands (Malvinas)", "phone": "(508) 481-1556", "job_title": "Engineering Manager"}, {"industry": "Retail", "country": "Falkland Islands (Malvinas)", "phone": "(555) 105-5825", "job_title": "Chief Technology Officer"}], "dirty_cols": ["industry", "country", "phone", "job_title"], "clean_cols": ["industry", "country", "phone", "job_title"], "plan": {"dataset_summary": "20 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"REALESTATE": "Real Estate", "Edtech": "Education", "PROPTECH": "Real Estate", "Property": "Real Estate", "education": "Education", "realestate": "Real Estate", "EDTECH": "Education", "ed": "Education", "edtech": "Education", "rbalestate": "Real Estate"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Côte D'Ivoire": "Côte d'Ivoire", "Republic of Singapore": "Singapore", "CI": "Côte d'Ivoire", "CIV": "Côte d'Ivoire", "sg": "Singapore", "flk": "Falkland Islands (Malvinas)", "côte d'ivoire": "Côte d'Ivoire", "FK": "Falkland Islands (Malvinas)", "falkland islands (malvinas)": "Falkland Islands (Malvinas)", "FLK": "Falkland Islands (Malvinas)", "côte dwivoire": "Côte d'Ivoire", "CÔTE D'IVOIRE": "Côte d'Ivoire"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Prod Manager": "Product Manager", "CTO": "Chief Technology Officer", "engineering mgr": "Engineering Manager", "PM": "Product Manager", "chief technology officer": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "eng mgr": "Engineering Manager", "Cto": "Chief Technology Officer", "Eng Mgr": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "ENG MGR": "Engineering Manager", "pm": "Product Manager", "Eng Manager": "Engineering Manager", "cto": "Chief Technology Officer"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "55.5%", "country": "Guyana", "phone": "(542)500-9205", "city": "nova milanese", "unit": "L"}, {"rate": "47.4%", "country": "GUM", "phone": "568-465-4867", "city": "Nova Milaanese", "unit": "centimeter"}, {"rate": "95.7%", "country": "federal democratic republic of nepal", "phone": "551-178-5458", "city": "HOHOE", "unit": "LITER"}, {"rate": "32.6%", "country": " GU", "phone": "(588)049-2295", "city": " Nova Milanese", "unit": "cm"}, {"rate": "55.4%", "country": "NPL", "phone": "(547)683-5947", "city": "nva milanese", "unit": "l"}, {"rate": "55.9%", "country": "  GUAM", "phone": "(597)629-3679", "city": " Hhooe ", "unit": "Cm"}, {"rate": "91.6%", "country": "guy", "phone": "(586)642-7869", "city": "   HOHOE", "unit": "  l"}, {"rate": "0.2%", "country": "gum", "phone": "(585)099-8591", "city": "Nova Milanese", "unit": " cm "}, {"rate": "70.9%", "country": "gy ", "phone": "5000171737", "city": "HHOE", "unit": "liters"}, {"rate": "32.1%", "country": " NP", "phone": "5033138509", "city": "Nova  Milanese", "unit": "litre"}, {"rate": "65.4%", "country": "Nepal", "phone": "540.674.3768", "city": "Nova Milanese", "unit": "l"}, {"rate": "36.4%", "country": " np ", "phone": "(570)695-0156", "city": "nova milanese", "unit": "centimeters"}, {"rate": "2.3%", "country": "GUM", "phone": "564.345.9586", "city": "Hohoe", "unit": "cm"}, {"rate": "14.8%", "country": "NP", "phone": "596.965.3836", "city": "Hohoe", "unit": "centimeter"}, {"rate": "53.4%", "country": "Np", "phone": "(591)288-6198", "city": "Hohoe", "unit": "liters"}, {"rate": "22.2%", "country": " guuam ", "phone": "577-758-4949", "city": "HOHOE", "unit": "L"}, {"rate": "98.0%", "country": "GUAM", "phone": "5654851922", "city": "nova milanese", "unit": "L"}, {"rate": "76.1%", "country": " gu ", "phone": "(573)969-1796", "city": "Nova Milanese  ", "unit": "cm"}, {"rate": "22.8%", "country": "gu", "phone": "593.047.5201", "city": "nova milanese", "unit": "cm"}, {"rate": "76.1%", "country": "Gum", "phone": "(597)185-6799", "city": "HOHOE   ", "unit": " liter "}, {"rate": "55.5%", "country": "GU   ", "phone": "537.912.9680", "city": "Hohoe", "unit": "CENTIMETERS"}, {"rate": "38.5%", "country": "gum", "phone": "587.758.1654", "city": "Nova Milanese", "unit": "L"}, {"rate": "2.3%", "country": "GUM", "phone": "564.345.9586", "city": "Hohoe", "unit": "cm"}], "clean": [{"rate": 0.555, "country": "Guyana", "phone": "(542) 500-9205", "city": "Nova Milanese", "unit": "L"}, {"rate": 0.474, "country": "Guam", "phone": "(568) 465-4867", "city": "Nova Milanese", "unit": "cm"}, {"rate": 0.9570000000000001, "country": "Nepal", "phone": "(551) 178-5458", "city": "Hohoe", "unit": "L"}, {"rate": 0.326, "country": "Guam", "phone": "(588) 049-2295", "city": "Nova Milanese", "unit": "cm"}, {"rate": 0.5539999999999999, "country": "Nepal", "phone": "(547) 683-5947", "city": "Nova Milanese", "unit": "L"}, {"rate": 0.5589999999999999, "country": "Guam", "phone": "(597) 629-3679", "city": "Hohoe", "unit": "cm"}, {"rate": 0.9159999999999999, "country": "Guyana", "phone": "(586) 642-7869", "city": "Hohoe", "unit": "L"}, {"rate": 0.002, "country": "Guam", "phone": "(585) 099-8591", "city": "Nova Milanese", "unit": "cm"}, {"rate": 0.7090000000000001, "country": "Guyana", "phone": "(500) 017-1737", "city": "Hohoe", "unit": "L"}, {"rate": 0.321, "country": "Nepal", "phone": "(503) 313-8509", "city": "Nova Milanese", "unit": "L"}, {"rate": 0.654, "country": "Nepal", "phone": "(540) 674-3768", "city": "Nova Milanese", "unit": "L"}, {"rate": 0.364, "country": "Nepal", "phone": "(570) 695-0156", "city": "Nova Milanese", "unit": "cm"}, {"rate": 0.023, "country": "Guam", "phone": "(564) 345-9586", "city": "Hohoe", "unit": "cm"}, {"rate": 0.14800000000000002, "country": "Nepal", "phone": "(596) 965-3836", "city": "Hohoe", "unit": "cm"}, {"rate": 0.534, "country": "Nepal", "phone": "(591) 288-6198", "city": "Hohoe", "unit": "L"}, {"rate": 0.222, "country": "Guam", "phone": "(577) 758-4949", "city": "Hohoe", "unit": "L"}, {"rate": 0.98, "country": "Guam", "phone": "(565) 485-1922", "city": "Nova Milanese", "unit": "L"}, {"rate": 0.7609999999999999, "country": "Guam", "phone": "(573) 969-1796", "city": "Nova Milanese", "unit": "cm"}, {"rate": 0.228, "country": "Guam", "phone": "(593) 047-5201", "city": "Nova Milanese", "unit": "cm"}, {"rate": 0.7609999999999999, "country": "Guam", "phone": "(597) 185-6799", "city": "Hohoe", "unit": "L"}, {"rate": 0.555, "country": "Guam", "phone": "(537) 912-9680", "city": "Hohoe", "unit": "cm"}, {"rate": 0.385, "country": "Guam", "phone": "(587) 758-1654", "city": "Nova Milanese", "unit": "L"}], "dirty_cols": ["rate", "country", "phone", "city", "unit"], "clean_cols": ["rate", "country", "phone", "city", "unit"], "plan": {"dataset_summary": "23 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GUM": "Guam", "federal democratic republic of nepal": "Nepal", "GU": "Guam", "NPL": "Nepal", "GUAM": "Guam", "guy": "Guyana", "gum": "Guam", "gy": "Guyana", "NP": "Nepal", "np": "Nepal", "Np": "Nepal", "guuam": "Guam", "gu": "Guam", "Gum": "Guam"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"nova milanese": "Nova Milanese", "Nova Milaanese": "Nova Milanese", "HOHOE": "Hohoe", "nva milanese": "Nova Milanese", "Hhooe": "Hohoe", "HHOE": "Hohoe"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"centimeter": "cm", "LITER": "L", "l": "L", "Cm": "cm", "liters": "L", "litre": "L", "centimeters": "cm", "liter": "L", "CENTIMETERS": "cm"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "Vt", "country": " glp", "job_title": "Dir", "email": "  cghabd@test.org", "unit": "KILOGRAM", "unnamed": ""}, {"state": "WA", "country": "  romania", "job_title": "dir", "email": "EBIIBJB@TEST.ORG", "unit": "L", "unnamed": ""}, {"state": " WA ", "country": "ROU", "job_title": " CTO ", "email": "KMALPN@EXAMPLE.COM", "unit": "LITRE", "unnamed": ""}, {"state": "washington", "country": "   Ro", "job_title": " Engineering Manager", "email": " pjeg@example.com ", "unit": "liters", "unnamed": ""}, {"state": "  Washington", "country": "ROU", "job_title": " Chief Technology Officer", "email": "  BMBN@EXAMPLE.COM", "unit": "Ml", "unnamed": ""}, {"state": "wa", "country": "GLP", "job_title": "C.T.O.", "email": "joej@example.com   ", "unit": "kilograms", "unnamed": ""}, {"state": "vermont", "country": " gp ", "job_title": "ENG MGR   ", "email": "BOLKHG@TEST.ORG", "unit": "MLS", "unnamed": ""}, {"state": "Vermont", "country": "ROU", "job_title": " DIR ", "email": " PFKP@CORP.IO ", "unit": "mls ", "unnamed": ""}, {"state": "VT", "country": "GLP", "job_title": "Chief Technology Officer", "email": "AECCON@CORP.IO", "unit": "litre", "unnamed": ""}, {"state": "VERMONT", "country": "RO", "job_title": "c.t.o.   ", "email": " KFAMOL@EXAMPLE.COM ", "unit": "kilogram", "unnamed": ""}, {"state": "Vermont", "country": "GLP", "job_title": "  DIRECTOR", "email": "gdblja@test.org", "unit": " liters", "unnamed": ""}, {"state": "Vermont", "country": "ROU", "job_title": " DIR ", "email": " PFKP@CORP.IO ", "unit": "mls ", "unnamed": ""}], "clean": [{"state": "Vermont", "country": "Guadeloupe", "job_title": "Director", "email": "cghabd@test.org", "unit": "kg"}, {"state": "Washington", "country": "Romania", "job_title": "Director", "email": "ebiibjb@test.org", "unit": "L"}, {"state": "Washington", "country": "Romania", "job_title": "Chief Technology Officer", "email": "kmalpn@example.com", "unit": "L"}, {"state": "Washington", "country": "Romania", "job_title": "Engineering Manager", "email": "pjeg@example.com", "unit": "L"}, {"state": "Washington", "country": "Romania", "job_title": "Chief Technology Officer", "email": "bmbn@example.com", "unit": "mL"}, {"state": "Washington", "country": "Guadeloupe", "job_title": "Chief Technology Officer", "email": "joej@example.com", "unit": "kg"}, {"state": "Vermont", "country": "Guadeloupe", "job_title": "Engineering Manager", "email": "bolkhg@test.org", "unit": "mL"}, {"state": "Vermont", "country": "Romania", "job_title": "Director", "email": "pfkp@corp.io", "unit": "mL"}, {"state": "Vermont", "country": "Guadeloupe", "job_title": "Chief Technology Officer", "email": "aeccon@corp.io", "unit": "L"}, {"state": "Vermont", "country": "Romania", "job_title": "Chief Technology Officer", "email": "kfamol@example.com", "unit": "kg"}, {"state": "Vermont", "country": "Guadeloupe", "job_title": "Director", "email": "gdblja@test.org", "unit": "L"}], "dirty_cols": ["state", "country", "job_title", "email", "unit", "unnamed"], "clean_cols": ["state", "country", "job_title", "email", "unit"], "plan": {"dataset_summary": "12 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Vt": "Vermont", "WA": "Washington", "washington": "Washington", "wa": "Washington", "vermont": "Vermont", "VT": "Vermont", "VERMONT": "Vermont"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"glp": "Guadeloupe", "romania": "Romania", "ROU": "Romania", "Ro": "Romania", "GLP": "Guadeloupe", "gp": "Guadeloupe", "RO": "Romania"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dir": "Director", "dir": "Director", "CTO": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "ENG MGR": "Engineering Manager", "DIR": "Director", "c.t.o.": "Chief Technology Officer", "DIRECTOR": "Director"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KILOGRAM": "kg", "LITRE": "L", "liters": "L", "Ml": "mL", "kilograms": "kg", "MLS": "mL", "mls": "mL", "litre": "L", "kilogram": "kg"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"status": "L", "is_active": "true", "industry": "Industrial", "phone": "(564)870-8063", "rate": "57.6%", "company": "Wonka  Co", "state": "Idaho   ", "email": "KFEA@TEST.ORG"}, {"status": "P3", "is_active": "Y", "industry": "Telco", "phone": "509-986-3772", "rate": "45.1%", "company": "Globex", "state": "CALIFORNIA", "email": "OFDM@CORP.IO"}, {"status": "P3", "is_active": "No", "industry": "Mfg", "phone": "5930652397", "rate": "34.5%", "company": "Vehement ", "state": "LOUISIANA", "email": "gnlap@corp.io"}, {"status": " MED ", "is_active": "Y", "industry": "Manufacturing", "phone": "(521)772-4070", "rate": "25.6%", "company": "Initech", "state": "North Carolina", "email": "dcbollb@test.org  "}, {"status": "med", "is_active": "Y", "industry": "Industrial", "phone": "(564)962-2980", "rate": "13.2%", "company": "Stark Industries  ", "state": "  MARYLAND", "email": "ccedn@mail.com"}, {"status": "p3", "is_active": "0", "industry": "Telco", "phone": "559.840.4353", "rate": "18.7%", "company": "Umbrella", "state": "Id   ", "email": "CPNOHJH@TEST.ORG "}, {"status": "low", "is_active": "TRUE", "industry": " comms ", "phone": "588.693.0857", "rate": "79.8%", "company": " Soylent Corp", "state": "CA", "email": "HPCF@CORP.IO"}, {"status": "medium", "is_active": "N", "industry": "Mfg", "phone": "527-753-2258", "rate": "19.1%", "company": "Vehement", "state": "CA", "email": "bamja@test.org"}, {"status": "MEDIUM", "is_active": "0", "industry": "Mfg", "phone": "5797795993", "rate": "17.0%", "company": "Globex", "state": "la", "email": "ljgoi@test.org"}, {"status": "med", "is_active": "TRUE", "industry": "comms", "phone": "563-330-9843", "rate": "71.4%", "company": "Stark  Industries", "state": "md", "email": " bajdfc@test.org"}, {"status": "low", "is_active": "1", "industry": "Mfg", "phone": "(585)130-7834", "rate": "83.9%", "company": "  Stark Industries", "state": "Id", "email": "PPKNCE@EXAMPLE.COM"}, {"status": "  p2", "is_active": "0", "industry": "telco", "phone": "530-963-6235", "rate": "81.5%", "company": "Cyberdyne", "state": "LA", "email": " AFLM@EXAMPLE.COM "}, {"status": "M", "is_active": "TRUE", "industry": "Telecom", "phone": "531-951-1898", "rate": "90.2%", "company": "Wonka Co ", "state": "  Ca", "email": "CADNJMG@EXAMPLE.COM"}, {"status": "medium", "is_active": "0", "industry": "TELECOM", "phone": "545.885.0009", "rate": "21.7%", "company": "Initech", "state": "Idho", "email": " jagiad@example.com "}, {"status": "P3", "is_active": "Yes", "industry": "Comms", "phone": "526-430-2335", "rate": "87.3%", "company": " Hooli ", "state": "LA", "email": "FJCFK@CORP.IO"}, {"status": "P2", "is_active": "1", "industry": "Mfg", "phone": "507.769.8159", "rate": "30.4%", "company": "Stark Industries", "state": "NC", "email": "   iagi@corp.io"}, {"status": "P3", "is_active": "Yes", "industry": "manufacturing", "phone": "(540)173-0679", "rate": "42.5%", "company": "   Hooli", "state": " Nc ", "email": "cgdcc@test.org"}, {"status": "p3  ", "is_active": "F", "industry": "mfg", "phone": "(576)590-4753", "rate": "9.1%", "company": "   Globex", "state": " MD ", "email": "ECJNLGO@MAIL.COM"}, {"status": "L", "is_active": "No", "industry": "telecom", "phone": "(565)394-0065", "rate": "76.0%", "company": "Soylent Corp", "state": " id", "email": " kfjkml@test.org "}, {"status": "med", "is_active": "0", "industry": "   manufacturing", "phone": "(501)743-4443", "rate": "89.1%", "company": "Vehement", "state": "Idao", "email": "  lghk@mail.com"}, {"status": " MED", "is_active": "TRUE", "industry": "mfg", "phone": "5571804304", "rate": "6.6%", "company": "  Hooli", "state": "   louisiana", "email": "   aelc@corp.io"}, {"status": "P3", "is_active": "FALSE", "industry": "MFG", "phone": "597.596.4486", "rate": "70.4%", "company": "Hooli", "state": "LA", "email": "malk@example.com"}, {"status": " l ", "is_active": "N", "industry": "Mfg  ", "phone": "536-062-5059", "rate": "54.7%", "company": "Acme Inc   ", "state": "maryland", "email": "CIIMOIM@TEST.ORG"}, {"status": "   m", "is_active": "T", "industry": "Telecommunications", "phone": "5868225458", "rate": "21.2%", "company": "Soylent  Corp", "state": "LA", "email": "PBCAFBN@EXAMPLE.COM"}, {"status": "med", "is_active": "Y", "industry": "Manufacturing", "phone": "5819195371", "rate": "82.3%", "company": "Vehement", "state": "Ca", "email": "bpffai@test.org"}, {"status": "   m", "is_active": "No", "industry": "Telecommunications", "phone": "505-109-2685", "rate": "85.5%", "company": "Globex", "state": "id", "email": "GDFGLFD@TEST.ORG"}, {"status": "med", "is_active": "0", "industry": "telecom   ", "phone": "(568)450-7692", "rate": "32.9%", "company": "Wonka Co", "state": "LOUISIANA", "email": "bkmieb@corp.io"}, {"status": "P3", "is_active": "1", "industry": "INDUSTRIAL   ", "phone": "517.050.2340", "rate": "8.0%", "company": "  Umbrella", "state": "California", "email": "  JLFGNCJ@TEST.ORG"}, {"status": "P3", "is_active": "Yes", "industry": "Telecom", "phone": "546.028.9284", "rate": "49.8%", "company": "   Cyberdyne", "state": "louisiana ", "email": "   EJCPB@TEST.ORG"}, {"status": "Low", "is_active": "N", "industry": " Manufacturing ", "phone": "546.703.0252", "rate": "83.9%", "company": "Soylent Corp", "state": "north carolina ", "email": "AGGDJJ@MAIL.COM"}, {"status": "P3", "is_active": "TRUE", "industry": " MANUFQCTURING ", "phone": "(572)073-9890", "rate": "22.5%", "company": "Soylent Corp", "state": "ID", "email": "blbfmm@corp.io"}, {"status": "low  ", "is_active": "TRUE", "industry": "industial", "phone": "(547)930-5599", "rate": "80.8%", "company": "Stark  Industries", "state": "IDAOH", "email": "hipoh@test.org  "}, {"status": "LOW", "is_active": "Yes", "industry": "Manufacturing   ", "phone": "(542)833-1219", "rate": "36.3%", "company": "Vehement  ", "state": "MARYLAND", "email": "elmp@mail.com"}, {"status": " P2", "is_active": "F", "industry": "telecommunications", "phone": "537.423.9636", "rate": "50.0%", "company": "Wonka Co  ", "state": "North Carolina", "email": "OJPEP@MAIL.COM"}, {"status": "P3", "is_active": "1", "industry": "telco", "phone": "517-563-4845", "rate": "55.1%", "company": "Globex", "state": "NC", "email": "GKDL@EXAMPLE.COM"}, {"status": "Low", "is_active": "No", "industry": "   Telco", "phone": "5752424250", "rate": "59.1%", "company": "Hooli   ", "state": "ID", "email": " GPNLLAE@TEST.ORG "}, {"status": "low", "is_active": "T", "industry": "Comms", "phone": "5725245284", "rate": "41.8%", "company": " Cyberdyne ", "state": "MD", "email": "EEKH@CORP.IO"}, {"status": "Low", "is_active": "FALSE", "industry": "industrial", "phone": "(512)204-2986", "rate": "89.4%", "company": "Stark  Industries", "state": "Idaho", "email": "jjae@test.org"}, {"status": "Medium", "is_active": "FALSE", "industry": "mfg", "phone": "564-690-8993", "rate": "8.9%", "company": " Globex ", "state": "MD", "email": "ABFI@MAIL.COM"}, {"status": "p3", "is_active": "FALSE", "industry": "Telco", "phone": "(561)439-2891", "rate": "54.3%", "company": " Globex ", "state": "NORTH CAROLINA", "email": " PNHBDIC@CORP.IO "}, {"status": "Medium", "is_active": "TRUE", "industry": "Mfg", "phone": "5442588509", "rate": "65.4%", "company": "Vehement", "state": "North Carolina", "email": "cjcgii@mail.com"}, {"status": "Low", "is_active": "0", "industry": "Telco", "phone": "5051414611", "rate": "53.8%", "company": "Soylent Corp", "state": "   CALIFORNIA", "email": "   KKGALOH@EXAMPLE.COM"}, {"status": "M", "is_active": "N", "industry": "industrial", "phone": "(540)435-7304", "rate": "57.7%", "company": "Stark Industries  ", "state": "Maryland", "email": "bhge@corp.io"}, {"status": "L", "is_active": "false", "industry": "manufacturing", "phone": "570-051-8919", "rate": "57.4%", "company": "Globex   ", "state": "   Id", "email": "GPNFAPO@TEST.ORG"}, {"status": "P2", "is_active": "No", "industry": "Telecom", "phone": "(511)794-9916", "rate": "6.6%", "company": "Cyberdyne  ", "state": "California", "email": "OHHO@CORP.IO  "}, {"status": "low", "is_active": "TRUE", "industry": "Manufacturing", "phone": "548.868.6727", "rate": "7.5%", "company": "Soylent Corp   ", "state": "id", "email": "   fcinmj@corp.io"}, {"status": "   P3", "is_active": "0", "industry": "TELECOMMUNICATIONS", "phone": "(598)697-2686", "rate": "37.0%", "company": "Acme Inc", "state": "Marylnd", "email": "deficcb@example.com   "}, {"status": "p2", "is_active": "1", "industry": "TEBCO", "phone": "(509)575-2461", "rate": "41.4%", "company": "   Globex", "state": "Idaho", "email": "KFHGP@EXAMPLE.COM"}, {"status": "med", "is_active": "1", "industry": " TELECOM", "phone": "(500)903-8137", "rate": "62.7%", "company": "Globex ", "state": "Louisiana", "email": "GLHH@TEST.ORG"}, {"status": " P3 ", "is_active": "T", "industry": "teecom", "phone": "504-151-8978", "rate": "15.5%", "company": "Globex", "state": "Id", "email": "PMMLO@MAIL.COM"}, {"status": "low", "is_active": "TRUE", "industry": "Industrial", "phone": "(558)195-4149", "rate": "56.5%", "company": "  Cyberdyne", "state": "NC", "email": " NLEGP@MAIL.COM "}, {"status": "L", "is_active": "false", "industry": "Manufacturing", "phone": "565-555-9573", "rate": "51.5%", "company": " Cyberdyne ", "state": "MD", "email": "PHIFHKB@CORP.IO  "}, {"status": "Low", "is_active": "Y", "industry": "Telecommuniications", "phone": "546.592.2244", "rate": "84.2%", "company": "Acme Inc   ", "state": "Nc", "email": " KHEKG@EXAMPLE.COM"}, {"status": "med", "is_active": "0", "industry": "   manufacturing", "phone": "(501)743-4443", "rate": "89.1%", "company": "Vehement", "state": "Idao", "email": "  lghk@mail.com"}, {"status": "medium", "is_active": "0", "industry": "TELECOM", "phone": "545.885.0009", "rate": "21.7%", "company": "Initech", "state": "Idho", "email": " jagiad@example.com "}, {"status": "", "is_active": "", "industry": "", "phone": "", "rate": "", "company": "", "state": "", "email": ""}], "clean": [{"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(564) 870-8063", "rate": 0.5760000000000001, "company": "Wonka Co", "state": "Idaho", "email": "kfea@test.org"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(509) 986-3772", "rate": 0.451, "company": "Globex", "state": "California", "email": "ofdm@corp.io"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(593) 065-2397", "rate": 0.345, "company": "Vehement", "state": "Louisiana", "email": "gnlap@corp.io"}, {"status": "Medium", "is_active": true, "industry": "Manufacturing", "phone": "(521) 772-4070", "rate": 0.256, "company": "Initech", "state": "North Carolina", "email": "dcbollb@test.org"}, {"status": "Medium", "is_active": true, "industry": "Manufacturing", "phone": "(564) 962-2980", "rate": 0.132, "company": "Stark Industries", "state": "Maryland", "email": "ccedn@mail.com"}, {"status": "Low", "is_active": false, "industry": "Telecommunications", "phone": "(559) 840-4353", "rate": 0.187, "company": "Umbrella", "state": "Idaho", "email": "cpnohjh@test.org"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(588) 693-0857", "rate": 0.7979999999999999, "company": "Soylent Corp", "state": "California", "email": "hpcf@corp.io"}, {"status": "Medium", "is_active": false, "industry": "Manufacturing", "phone": "(527) 753-2258", "rate": 0.191, "company": "Vehement", "state": "California", "email": "bamja@test.org"}, {"status": "Medium", "is_active": false, "industry": "Manufacturing", "phone": "(579) 779-5993", "rate": 0.17, "company": "Globex", "state": "Louisiana", "email": "ljgoi@test.org"}, {"status": "Medium", "is_active": true, "industry": "Telecommunications", "phone": "(563) 330-9843", "rate": 0.7140000000000001, "company": "Stark Industries", "state": "Maryland", "email": "bajdfc@test.org"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(585) 130-7834", "rate": 0.8390000000000001, "company": "Stark Industries", "state": "Idaho", "email": "ppknce@example.com"}, {"status": "Medium", "is_active": false, "industry": "Telecommunications", "phone": "(530) 963-6235", "rate": 0.815, "company": "Cyberdyne", "state": "Louisiana", "email": "aflm@example.com"}, {"status": "Medium", "is_active": true, "industry": "Telecommunications", "phone": "(531) 951-1898", "rate": 0.902, "company": "Wonka Co", "state": "California", "email": "cadnjmg@example.com"}, {"status": "Medium", "is_active": false, "industry": "Telecommunications", "phone": "(545) 885-0009", "rate": 0.217, "company": "Initech", "state": "Idaho", "email": "jagiad@example.com"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(526) 430-2335", "rate": 0.873, "company": "Hooli", "state": "Louisiana", "email": "fjcfk@corp.io"}, {"status": "Medium", "is_active": true, "industry": "Manufacturing", "phone": "(507) 769-8159", "rate": 0.304, "company": "Stark Industries", "state": "North Carolina", "email": "iagi@corp.io"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(540) 173-0679", "rate": 0.425, "company": "Hooli", "state": "North Carolina", "email": "cgdcc@test.org"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(576) 590-4753", "rate": 0.091, "company": "Globex", "state": "Maryland", "email": "ecjnlgo@mail.com"}, {"status": "Low", "is_active": false, "industry": "Telecommunications", "phone": "(565) 394-0065", "rate": 0.76, "company": "Soylent Corp", "state": "Idaho", "email": "kfjkml@test.org"}, {"status": "Medium", "is_active": false, "industry": "Manufacturing", "phone": "(501) 743-4443", "rate": 0.8909999999999999, "company": "Vehement", "state": "Idaho", "email": "lghk@mail.com"}, {"status": "Medium", "is_active": true, "industry": "Manufacturing", "phone": "(557) 180-4304", "rate": 0.066, "company": "Hooli", "state": "Louisiana", "email": "aelc@corp.io"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(597) 596-4486", "rate": 0.7040000000000001, "company": "Hooli", "state": "Louisiana", "email": "malk@example.com"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(536) 062-5059", "rate": 0.547, "company": "Acme Inc", "state": "Maryland", "email": "ciimoim@test.org"}, {"status": "Medium", "is_active": true, "industry": "Telecommunications", "phone": "(586) 822-5458", "rate": 0.212, "company": "Soylent Corp", "state": "Louisiana", "email": "pbcafbn@example.com"}, {"status": "Medium", "is_active": true, "industry": "Manufacturing", "phone": "(581) 919-5371", "rate": 0.823, "company": "Vehement", "state": "California", "email": "bpffai@test.org"}, {"status": "Medium", "is_active": false, "industry": "Telecommunications", "phone": "(505) 109-2685", "rate": 0.855, "company": "Globex", "state": "Idaho", "email": "gdfglfd@test.org"}, {"status": "Medium", "is_active": false, "industry": "Telecommunications", "phone": "(568) 450-7692", "rate": 0.32899999999999996, "company": "Wonka Co", "state": "Louisiana", "email": "bkmieb@corp.io"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(517) 050-2340", "rate": 0.08, "company": "Umbrella", "state": "California", "email": "jlfgncj@test.org"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(546) 028-9284", "rate": 0.498, "company": "Cyberdyne", "state": "Louisiana", "email": "ejcpb@test.org"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(546) 703-0252", "rate": 0.8390000000000001, "company": "Soylent Corp", "state": "North Carolina", "email": "aggdjj@mail.com"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(572) 073-9890", "rate": 0.225, "company": "Soylent Corp", "state": "Idaho", "email": "blbfmm@corp.io"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(547) 930-5599", "rate": 0.8079999999999999, "company": "Stark Industries", "state": "Idaho", "email": "hipoh@test.org"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(542) 833-1219", "rate": 0.363, "company": "Vehement", "state": "Maryland", "email": "elmp@mail.com"}, {"status": "Medium", "is_active": false, "industry": "Telecommunications", "phone": "(537) 423-9636", "rate": 0.5, "company": "Wonka Co", "state": "North Carolina", "email": "ojpep@mail.com"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(517) 563-4845", "rate": 0.551, "company": "Globex", "state": "North Carolina", "email": "gkdl@example.com"}, {"status": "Low", "is_active": false, "industry": "Telecommunications", "phone": "(575) 242-4250", "rate": 0.591, "company": "Hooli", "state": "Idaho", "email": "gpnllae@test.org"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(572) 524-5284", "rate": 0.418, "company": "Cyberdyne", "state": "Maryland", "email": "eekh@corp.io"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(512) 204-2986", "rate": 0.894, "company": "Stark Industries", "state": "Idaho", "email": "jjae@test.org"}, {"status": "Medium", "is_active": false, "industry": "Manufacturing", "phone": "(564) 690-8993", "rate": 0.08900000000000001, "company": "Globex", "state": "Maryland", "email": "abfi@mail.com"}, {"status": "Low", "is_active": false, "industry": "Telecommunications", "phone": "(561) 439-2891", "rate": 0.5429999999999999, "company": "Globex", "state": "North Carolina", "email": "pnhbdic@corp.io"}, {"status": "Medium", "is_active": true, "industry": "Manufacturing", "phone": "(544) 258-8509", "rate": 0.654, "company": "Vehement", "state": "North Carolina", "email": "cjcgii@mail.com"}, {"status": "Low", "is_active": false, "industry": "Telecommunications", "phone": "(505) 141-4611", "rate": 0.5379999999999999, "company": "Soylent Corp", "state": "California", "email": "kkgaloh@example.com"}, {"status": "Medium", "is_active": false, "industry": "Manufacturing", "phone": "(540) 435-7304", "rate": 0.5770000000000001, "company": "Stark Industries", "state": "Maryland", "email": "bhge@corp.io"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(570) 051-8919", "rate": 0.574, "company": "Globex", "state": "Idaho", "email": "gpnfapo@test.org"}, {"status": "Medium", "is_active": false, "industry": "Telecommunications", "phone": "(511) 794-9916", "rate": 0.066, "company": "Cyberdyne", "state": "California", "email": "ohho@corp.io"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(548) 868-6727", "rate": 0.075, "company": "Soylent Corp", "state": "Idaho", "email": "fcinmj@corp.io"}, {"status": "Low", "is_active": false, "industry": "Telecommunications", "phone": "(598) 697-2686", "rate": 0.37, "company": "Acme Inc", "state": "Maryland", "email": "deficcb@example.com"}, {"status": "Medium", "is_active": true, "industry": "Telecommunications", "phone": "(509) 575-2461", "rate": 0.414, "company": "Globex", "state": "Idaho", "email": "kfhgp@example.com"}, {"status": "Medium", "is_active": true, "industry": "Telecommunications", "phone": "(500) 903-8137", "rate": 0.627, "company": "Globex", "state": "Louisiana", "email": "glhh@test.org"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(504) 151-8978", "rate": 0.155, "company": "Globex", "state": "Idaho", "email": "pmmlo@mail.com"}, {"status": "Low", "is_active": true, "industry": "Manufacturing", "phone": "(558) 195-4149", "rate": 0.565, "company": "Cyberdyne", "state": "North Carolina", "email": "nlegp@mail.com"}, {"status": "Low", "is_active": false, "industry": "Manufacturing", "phone": "(565) 555-9573", "rate": 0.515, "company": "Cyberdyne", "state": "Maryland", "email": "phifhkb@corp.io"}, {"status": "Low", "is_active": true, "industry": "Telecommunications", "phone": "(546) 592-2244", "rate": 0.8420000000000001, "company": "Acme Inc", "state": "North Carolina", "email": "khekg@example.com"}], "dirty_cols": ["status", "is_active", "industry", "phone", "rate", "company", "state", "email"], "clean_cols": ["status", "is_active", "industry", "phone", "rate", "company", "state", "email"], "plan": {"dataset_summary": "56 rows × 8 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"L": "Low", "P3": "Low", "MED": "Medium", "med": "Medium", "p3": "Low", "low": "Low", "medium": "Medium", "MEDIUM": "Medium", "p2": "Medium", "M": "Medium", "P2": "Medium", "l": "Low", "m": "Medium", "LOW": "Low"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Industrial": "Manufacturing", "Telco": "Telecommunications", "Mfg": "Manufacturing", "comms": "Telecommunications", "telco": "Telecommunications", "Telecom": "Telecommunications", "TELECOM": "Telecommunications", "Comms": "Telecommunications", "manufacturing": "Manufacturing", "mfg": "Manufacturing", "telecom": "Telecommunications", "MFG": "Manufacturing", "INDUSTRIAL": "Manufacturing", "MANUFQCTURING": "Manufacturing", "industial": "Manufacturing", "telecommunications": "Telecommunications", "industrial": "Manufacturing", "TELECOMMUNICATIONS": "Telecommunications", "TEBCO": "Telecommunications", "teecom": "Telecommunications", "Telecommuniications": "Telecommunications"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CALIFORNIA": "California", "LOUISIANA": "Louisiana", "MARYLAND": "Maryland", "Id": "Idaho", "CA": "California", "la": "Louisiana", "md": "Maryland", "LA": "Louisiana", "Ca": "California", "Idho": "Idaho", "NC": "North Carolina", "Nc": "North Carolina", "MD": "Maryland", "id": "Idaho", "Idao": "Idaho", "louisiana": "Louisiana", "maryland": "Maryland", "north carolina": "North Carolina", "ID": "Idaho", "IDAOH": "Idaho", "NORTH CAROLINA": "North Carolina", "Marylnd": "Maryland"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"city": "MAULDIN", "job_title": "Chief Executive Officer", "status": "P2", "name": "Bob Lee "}, {"city": "mauldin", "job_title": " c.e.o. ", "status": "l", "name": "Omar Brandt"}, {"city": "Dubai", "job_title": "Product Manager ", "status": " P2 ", "name": "Carol Johnson "}, {"city": "Cairo", "job_title": " Ceo ", "status": "Low", "name": "Sara Brandt "}, {"city": "Cairo", "job_title": " prod mgr", "status": "m", "name": "Eve Wong"}, {"city": "Dubai", "job_title": "Sr. Engineer", "status": " low ", "name": "Mona Fischer"}, {"city": "cairo", "job_title": "PM", "status": "low ", "name": "Lena Brandt"}, {"city": "Cairo", "job_title": "Sr. Engineer", "status": "p3", "name": "Priya Novak"}, {"city": "dubai", "job_title": "Prod gr", "status": "low", "name": " Judy Khan"}, {"city": "Cairo", "job_title": "snr engineer", "status": "Medium", "name": "Eve Moore"}, {"city": " mauldin", "job_title": "CEO", "status": " LOW ", "name": "Omar Smith"}, {"city": "Dubai ", "job_title": "chief executive officer", "status": "L", "name": "Karl Moore"}, {"city": "Cario   ", "job_title": "Chief Executive Officer", "status": "Medium", "name": "   Alice Lee"}, {"city": "Dubai", "job_title": "prod mgr", "status": "   l", "name": "  Priya Adams"}, {"city": "   dubai", "job_title": "prod manager", "status": "M   ", "name": "Grace Moore"}, {"city": "MAULDIN", "job_title": "pm", "status": "medium", "name": "Frank Khan  "}, {"city": "Mauldin", "job_title": "ceo  ", "status": "L", "name": "  Ivan Smith"}, {"city": "Mauldin", "job_title": "SNR ENGINEER", "status": "p2", "name": "  Eve Wong"}, {"city": "Dubai", "job_title": "Sr. Engineer", "status": "P3", "name": "Eve Khan   "}, {"city": "Dubai ", "job_title": "SENIOR ENG  ", "status": "m", "name": " Eve Johnson"}, {"city": "cairo", "job_title": "chief executive officer", "status": "M", "name": "Frank Reyes"}, {"city": " MAULDDIN", "job_title": "sr engineer", "status": "M", "name": "Lena  Ali"}, {"city": " Dubai ", "job_title": "PRROD MGR", "status": "low", "name": "Carol Smith"}, {"city": "dubai", "job_title": "Product Manager", "status": "P3", "name": "  Ivan Khan"}, {"city": "Dubai ", "job_title": "SENIOR ENG  ", "status": "m", "name": " Eve Johnson"}, {"city": "dubai", "job_title": "Product Manager", "status": "P3", "name": "  Ivan Khan"}], "clean": [{"city": "Mauldin", "job_title": "Chief Executive Officer", "status": "Medium", "name": "Bob Lee"}, {"city": "Mauldin", "job_title": "Chief Executive Officer", "status": "Low", "name": "Omar Brandt"}, {"city": "Dubai", "job_title": "Product Manager", "status": "Medium", "name": "Carol Johnson"}, {"city": "Cairo", "job_title": "Chief Executive Officer", "status": "Low", "name": "Sara Brandt"}, {"city": "Cairo", "job_title": "Product Manager", "status": "Medium", "name": "Eve Wong"}, {"city": "Dubai", "job_title": "Senior Engineer", "status": "Low", "name": "Mona Fischer"}, {"city": "Cairo", "job_title": "Product Manager", "status": "Low", "name": "Lena Brandt"}, {"city": "Cairo", "job_title": "Senior Engineer", "status": "Low", "name": "Priya Novak"}, {"city": "Dubai", "job_title": "Product Manager", "status": "Low", "name": "Judy Khan"}, {"city": "Cairo", "job_title": "Senior Engineer", "status": "Medium", "name": "Eve Moore"}, {"city": "Mauldin", "job_title": "Chief Executive Officer", "status": "Low", "name": "Omar Smith"}, {"city": "Dubai", "job_title": "Chief Executive Officer", "status": "Low", "name": "Karl Moore"}, {"city": "Cairo", "job_title": "Chief Executive Officer", "status": "Medium", "name": "Alice Lee"}, {"city": "Dubai", "job_title": "Product Manager", "status": "Low", "name": "Priya Adams"}, {"city": "Dubai", "job_title": "Product Manager", "status": "Medium", "name": "Grace Moore"}, {"city": "Mauldin", "job_title": "Product Manager", "status": "Medium", "name": "Frank Khan"}, {"city": "Mauldin", "job_title": "Chief Executive Officer", "status": "Low", "name": "Ivan Smith"}, {"city": "Mauldin", "job_title": "Senior Engineer", "status": "Medium", "name": "Eve Wong"}, {"city": "Dubai", "job_title": "Senior Engineer", "status": "Low", "name": "Eve Khan"}, {"city": "Dubai", "job_title": "Senior Engineer", "status": "Medium", "name": "Eve Johnson"}, {"city": "Cairo", "job_title": "Chief Executive Officer", "status": "Medium", "name": "Frank Reyes"}, {"city": "Mauldin", "job_title": "Senior Engineer", "status": "Medium", "name": "Lena Ali"}, {"city": "Dubai", "job_title": "Product Manager", "status": "Low", "name": "Carol Smith"}, {"city": "Dubai", "job_title": "Product Manager", "status": "Low", "name": "Ivan Khan"}], "dirty_cols": ["city", "job_title", "status", "name"], "clean_cols": ["city", "job_title", "status", "name"], "plan": {"dataset_summary": "26 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MAULDIN": "Mauldin", "mauldin": "Mauldin", "cairo": "Cairo", "dubai": "Dubai", "Cario": "Cairo", "MAULDDIN": "Mauldin"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"c.e.o.": "Chief Executive Officer", "Ceo": "Chief Executive Officer", "prod mgr": "Product Manager", "Sr. Engineer": "Senior Engineer", "PM": "Product Manager", "Prod gr": "Product Manager", "snr engineer": "Senior Engineer", "CEO": "Chief Executive Officer", "chief executive officer": "Chief Executive Officer", "prod manager": "Product Manager", "pm": "Product Manager", "ceo": "Chief Executive Officer", "SNR ENGINEER": "Senior Engineer", "SENIOR ENG": "Senior Engineer", "sr engineer": "Senior Engineer", "PRROD MGR": "Product Manager"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P2": "Medium", "l": "Low", "m": "Medium", "low": "Low", "p3": "Low", "LOW": "Low", "L": "Low", "M": "Medium", "medium": "Medium", "p2": "Medium", "P3": "Low"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "people ops", "email": "OOMMEL@MAIL.COM", "unit": "kg", "signup_date": "4/16/2023", "rate": "92.5%"}, {"city": "San Antonio Oeste", "job_title": "CEO  ", "department": "Ops", "email": "PLMJPP@MAIL.COM", "unit": "pouods", "signup_date": "11/27/2023", "rate": "31.9%"}, {"city": "Nowshera Kalan", "job_title": "   chief executive officer", "department": "Human Resources   ", "email": "dnlcpd@example.com", "unit": "KG", "signup_date": "2 Sep 2023", "rate": "95.0%"}, {"city": "Noswhera Kalan", "job_title": "c.e.o.", "department": "   OPERATIONS", "email": "HPHNG@TEST.ORG", "unit": "kg", "signup_date": "2023-10-09", "rate": "97.6%"}, {"city": "BADARGANJ", "job_title": "   Chief Executive Officer", "department": "operations", "email": "GILP@TEST.ORG", "unit": "mi", "signup_date": "12/13/2023", "rate": "11.7%"}, {"city": "KODUVAYUR", "job_title": "Chief Executivj Officer", "department": " OPS ", "email": "gomjfo@mail.com  ", "unit": "lbs", "signup_date": "20 Nov 2023", "rate": "3.8%"}, {"city": "Baadrganj", "job_title": "Dir", "department": "People Ops", "email": "ogdha@example.com", "unit": "kg", "signup_date": "45004", "rate": "72.4%"}, {"city": "San Antonio OOeste ", "job_title": "Chief Executive Officer", "department": "HUMAN RESOURCES", "email": "  mcbkom@test.org", "unit": "lb", "signup_date": "2023-05-10", "rate": "97.9%"}, {"city": "Koduvayur", "job_title": "DIR", "department": "Ops", "email": "obnngl@example.com  ", "unit": "kilogram   ", "signup_date": "20 Oct 2023", "rate": "92.1%"}, {"city": "Badarganj", "job_title": "director", "department": "hr", "email": "   LJHDBNJ@MAIL.COM", "unit": "pound", "signup_date": "1 Feb 2023", "rate": "66.8%"}, {"city": "Nowshera Kalan", "job_title": "CEO", "department": "ops", "email": "bjeae@example.com", "unit": " Lb ", "signup_date": "2/11/2023", "rate": "1.2%"}, {"city": "Koduvayur", "job_title": "ceo", "department": "HR", "email": "KFHFBDK@EXAMPLE.COM", "unit": "Kilogram ", "signup_date": "13 Mar 2023", "rate": "58.4%"}, {"city": " koduvayur", "job_title": "director", "department": "OPS", "email": "cdfjp@mail.com", "unit": " kilogram ", "signup_date": "1/9/2023", "rate": "57.1%"}, {"city": "rudow", "job_title": "Dir.", "department": "Ops", "email": "ELOEPD@EXAMPLE.COM", "unit": "miile", "signup_date": "2023-03-25", "rate": "7.5%"}, {"city": "badarganj", "job_title": "Dir", "department": "operationns", "email": "cbcep@mail.com", "unit": "kg", "signup_date": "2023-09-19", "rate": "71.9%"}, {"city": "Rudow", "job_title": "DIR", "department": "hr", "email": "BFOCK@TEST.ORG", "unit": "MI ", "signup_date": "3/7/2023", "rate": "20.8%"}, {"city": "Nowshera Kalan", "job_title": "ceo", "department": "Hr", "email": " JPMCPC@EXAMPLE.COM ", "unit": "miles", "signup_date": "10/12/2023", "rate": "65.8%"}, {"city": "nowshera kalan   ", "job_title": "Chief Executive Officer", "department": "Human eRsources", "email": "   INIMEP@MAIL.COM", "unit": "   kg", "signup_date": "2023-01-20", "rate": "88.4%"}, {"city": "  nowshera kalan", "job_title": "Dir. ", "department": "   Operations", "email": "   ikikeg@test.org", "unit": "   lbs", "signup_date": "2023-11-17", "rate": "74.1%"}, {"city": "Kodduvayur", "job_title": "DIRECTOR ", "department": "people", "email": "GIFEAK@EXAMPLE.COM", "unit": "lb", "signup_date": "2023-04-26", "rate": "30.9%"}, {"city": "san antonio oeste", "job_title": "chief executive officer", "department": "OPS", "email": "   AING@EXAMPLE.COM", "unit": " kilograms ", "signup_date": "2023-04-23", "rate": "95.6%"}, {"city": "badarganj   ", "job_title": " C.E.O. ", "department": "Ops", "email": "fdlk@test.org", "unit": "   lbs", "signup_date": "2023-08-22", "rate": "13.9%"}, {"city": "Nowshera Kalan", "job_title": "Director", "department": "  HR", "email": "GAPPBLI@EXAMPLE.COM", "unit": "pounds", "signup_date": "2023-01-26", "rate": "12.6%"}, {"city": "Rudow", "job_title": "   C.E.O.", "department": "Ops", "email": "GOAGFLB@MAIL.COM ", "unit": "MI  ", "signup_date": "3/12/2023", "rate": "55.9%"}, {"city": "ruudow", "job_title": "chief executive officer", "department": "human resources", "email": "plfpfga@test.org  ", "unit": "pound", "signup_date": "14 Apr 2023", "rate": "84.9%"}, {"city": "Nowshera Kalan", "job_title": "DR.  ", "department": "People  Ops", "email": "IHOEHH@EXAMPLE.COM ", "unit": "mi  ", "signup_date": "45279", "rate": "30.8%"}, {"city": "Rudow", "job_title": "C.EO..", "department": "People", "email": "LHCBBJL@TEST.ORG", "unit": "pound", "signup_date": "2023-06-28", "rate": "15.6%"}, {"city": "badarganj", "job_title": "Chief Executive Officer", "department": "People Bps", "email": "bbifgb@corp.io", "unit": "   kilogram", "signup_date": "22 May 2023", "rate": "65.5%"}, {"city": " rudow ", "job_title": "Dir.", "department": "Ops", "email": "jnhj@mail.com", "unit": "   mile", "signup_date": "2/9/2023", "rate": "57.5%"}, {"city": "rudow", "job_title": "Chief Executive Officer ", "department": "Ops", "email": "AEPHLPI@MAIL.COM", "unit": "MILES", "signup_date": "2023-07-16", "rate": "83.4%"}, {"city": "san antofio oeste", "job_title": "C.E.O.", "department": " operations", "email": "ifihmj@example.com", "unit": " KG ", "signup_date": "45135", "rate": "94.1%"}, {"city": "koduvayur", "job_title": "Chief Executive Officer", "department": "People Ops", "email": "  AEGIHAA@TEST.ORG", "unit": "pounds", "signup_date": "9 Sep 2023", "rate": "59.4%"}, {"city": "badarganj", "job_title": "ceo", "department": "operations", "email": " fehm@mail.com ", "unit": "Mi", "signup_date": "44967", "rate": "16.8%"}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Operations", "email": "  KILIMEP@EXAMPLE.COM", "unit": "lb", "signup_date": "2023-02-10", "rate": "6.6%"}, {"city": "Nowshera Kalan", "job_title": "CEO", "department": "People", "email": "HKNL@CORP.IO", "unit": "POUNDS", "signup_date": "2023-10-09", "rate": "37.9%"}, {"city": "Rudow", "job_title": "CEO", "department": "hr", "email": "NCJDKE@EXAMPLE.COM", "unit": "Mi", "signup_date": "9/16/2023", "rate": "37.6%"}, {"city": "   Rudow", "job_title": "Dir.", "department": "Human  Resources", "email": " cipkb@mail.com ", "unit": "pounds", "signup_date": "20 May 2023", "rate": "83.8%"}, {"city": "badarganj", "job_title": "director", "department": "operatsons", "email": "APHCIDA@MAIL.COM  ", "unit": "Lb", "signup_date": "44991", "rate": "68.3%"}, {"city": "Rudow", "job_title": "Director", "department": "Operations", "email": "  LKGP@MAIL.COM", "unit": "lbs", "signup_date": "10/27/2023", "rate": "21.6%"}, {"city": "koduvayur", "job_title": "Dzr.", "department": "PEOPLE", "email": "OIJODN@TEST.ORG  ", "unit": "mi", "signup_date": "11/15/2023", "rate": "41.1%"}, {"city": "Rudow", "job_title": "CEO", "department": "Operations", "email": "EJGBIHF@MAIL.COM", "unit": "kg", "signup_date": "28 Sep 2023", "rate": "68.9%"}, {"city": "Koduvayur", "job_title": "C.E.O.", "department": " oeprations ", "email": "DACA@EXAMPLE.COM", "unit": "kilogram", "signup_date": "28 Feb 2023", "rate": "25.9%"}, {"city": "SAN ANTONIO OESTE", "job_title": "dir.", "department": "ops", "email": "   PHIJL@MAIL.COM", "unit": "mi", "signup_date": "2023-12-23", "rate": "93.7%"}, {"city": "BADARGANJ", "job_title": "Director", "department": "PEOPLE OPS ", "email": "EFFIFGP@MAIL.COM", "unit": "kilogram", "signup_date": "4/2/2023", "rate": "96.4%"}, {"city": "badarganj", "job_title": "C..EO.", "department": "People Ops", "email": "  LGFGGA@MAIL.COM", "unit": "lb", "signup_date": "25 Jul 2023", "rate": "57.0%"}, {"city": "badarganj  ", "job_title": "Chief Executive Officer   ", "department": "OPS", "email": "  lapll@example.com", "unit": "pound", "signup_date": "45135", "rate": "33.9%"}, {"city": "san antonio oeste", "job_title": "di.", "department": "operations", "email": "fangd@mail.com", "unit": "Lb", "signup_date": "45181", "rate": "60.0%"}, {"city": "BADARGANJ", "job_title": "dir.", "department": "operations", "email": "lkdmdcg@example.com", "unit": "pounds", "signup_date": "2023-03-16", "rate": "53.8%"}, {"city": "BADARGANJ", "job_title": "   Chief Executive Officer", "department": "operations", "email": "GILP@TEST.ORG", "unit": "mi", "signup_date": "12/13/2023", "rate": "11.7%"}], "clean": [{"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "oommel@mail.com", "unit": "kg", "signup_date": "2023-04-16", "rate": 0.925}, {"city": "San Antonio Oeste", "job_title": "Chief Executive Officer", "department": "Operations", "email": "plmjpp@mail.com", "unit": "lb", "signup_date": "2023-11-27", "rate": 0.319}, {"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "dnlcpd@example.com", "unit": "kg", "signup_date": "2023-09-02", "rate": 0.95}, {"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Operations", "email": "hphng@test.org", "unit": "kg", "signup_date": "2023-10-09", "rate": 0.976}, {"city": "Badarganj", "job_title": "Chief Executive Officer", "department": "Operations", "email": "gilp@test.org", "unit": "mi", "signup_date": "2023-12-13", "rate": 0.11699999999999999}, {"city": "Koduvayur", "job_title": "Chief Executive Officer", "department": "Operations", "email": "gomjfo@mail.com", "unit": "lb", "signup_date": "2023-11-20", "rate": 0.038}, {"city": "Badarganj", "job_title": "Director", "department": "Human Resources", "email": "ogdha@example.com", "unit": "kg", "signup_date": "2023-03-19", "rate": 0.7240000000000001}, {"city": "San Antonio Oeste", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "mcbkom@test.org", "unit": "lb", "signup_date": "2023-05-10", "rate": 0.9790000000000001}, {"city": "Koduvayur", "job_title": "Director", "department": "Operations", "email": "obnngl@example.com", "unit": "kg", "signup_date": "2023-10-20", "rate": 0.9209999999999999}, {"city": "Badarganj", "job_title": "Director", "department": "Human Resources", "email": "ljhdbnj@mail.com", "unit": "lb", "signup_date": "2023-02-01", "rate": 0.6679999999999999}, {"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Operations", "email": "bjeae@example.com", "unit": "lb", "signup_date": "2023-02-11", "rate": 0.012}, {"city": "Koduvayur", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "kfhfbdk@example.com", "unit": "kg", "signup_date": "2023-03-13", "rate": 0.584}, {"city": "Koduvayur", "job_title": "Director", "department": "Operations", "email": "cdfjp@mail.com", "unit": "kg", "signup_date": "2023-01-09", "rate": 0.5710000000000001}, {"city": "Rudow", "job_title": "Director", "department": "Operations", "email": "eloepd@example.com", "unit": "mi", "signup_date": "2023-03-25", "rate": 0.075}, {"city": "Badarganj", "job_title": "Director", "department": "Operations", "email": "cbcep@mail.com", "unit": "kg", "signup_date": "2023-09-19", "rate": 0.7190000000000001}, {"city": "Rudow", "job_title": "Director", "department": "Human Resources", "email": "bfock@test.org", "unit": "mi", "signup_date": "2023-03-07", "rate": 0.20800000000000002}, {"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "jpmcpc@example.com", "unit": "mi", "signup_date": "2023-10-12", "rate": 0.6579999999999999}, {"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "inimep@mail.com", "unit": "kg", "signup_date": "2023-01-20", "rate": 0.884}, {"city": "Nowshera Kalan", "job_title": "Director", "department": "Operations", "email": "ikikeg@test.org", "unit": "lb", "signup_date": "2023-11-17", "rate": 0.741}, {"city": "Koduvayur", "job_title": "Director", "department": "Human Resources", "email": "gifeak@example.com", "unit": "lb", "signup_date": "2023-04-26", "rate": 0.309}, {"city": "San Antonio Oeste", "job_title": "Chief Executive Officer", "department": "Operations", "email": "aing@example.com", "unit": "kg", "signup_date": "2023-04-23", "rate": 0.956}, {"city": "Badarganj", "job_title": "Chief Executive Officer", "department": "Operations", "email": "fdlk@test.org", "unit": "lb", "signup_date": "2023-08-22", "rate": 0.139}, {"city": "Nowshera Kalan", "job_title": "Director", "department": "Human Resources", "email": "gappbli@example.com", "unit": "lb", "signup_date": "2023-01-26", "rate": 0.126}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Operations", "email": "goagflb@mail.com", "unit": "mi", "signup_date": "2023-03-12", "rate": 0.5589999999999999}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "plfpfga@test.org", "unit": "lb", "signup_date": "2023-04-14", "rate": 0.8490000000000001}, {"city": "Nowshera Kalan", "job_title": "Director", "department": "Human Resources", "email": "ihoehh@example.com", "unit": "mi", "signup_date": "2023-12-19", "rate": 0.308}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "lhcbbjl@test.org", "unit": "lb", "signup_date": "2023-06-28", "rate": 0.156}, {"city": "Badarganj", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "bbifgb@corp.io", "unit": "kg", "signup_date": "2023-05-22", "rate": 0.655}, {"city": "Rudow", "job_title": "Director", "department": "Operations", "email": "jnhj@mail.com", "unit": "mi", "signup_date": "2023-02-09", "rate": 0.575}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Operations", "email": "aephlpi@mail.com", "unit": "mi", "signup_date": "2023-07-16", "rate": 0.8340000000000001}, {"city": "San Antonio Oeste", "job_title": "Chief Executive Officer", "department": "Operations", "email": "ifihmj@example.com", "unit": "kg", "signup_date": "2023-07-28", "rate": 0.941}, {"city": "Koduvayur", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "aegihaa@test.org", "unit": "lb", "signup_date": "2023-09-09", "rate": 0.594}, {"city": "Badarganj", "job_title": "Chief Executive Officer", "department": "Operations", "email": "fehm@mail.com", "unit": "mi", "signup_date": "2023-02-10", "rate": 0.168}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Operations", "email": "kilimep@example.com", "unit": "lb", "signup_date": "2023-02-10", "rate": 0.066}, {"city": "Nowshera Kalan", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "hknl@corp.io", "unit": "lb", "signup_date": "2023-10-09", "rate": 0.379}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "ncjdke@example.com", "unit": "mi", "signup_date": "2023-09-16", "rate": 0.376}, {"city": "Rudow", "job_title": "Director", "department": "Human Resources", "email": "cipkb@mail.com", "unit": "lb", "signup_date": "2023-05-20", "rate": 0.838}, {"city": "Badarganj", "job_title": "Director", "department": "Operations", "email": "aphcida@mail.com", "unit": "lb", "signup_date": "2023-03-06", "rate": 0.6829999999999999}, {"city": "Rudow", "job_title": "Director", "department": "Operations", "email": "lkgp@mail.com", "unit": "lb", "signup_date": "2023-10-27", "rate": 0.21600000000000003}, {"city": "Koduvayur", "job_title": "Director", "department": "Human Resources", "email": "oijodn@test.org", "unit": "mi", "signup_date": "2023-11-15", "rate": 0.41100000000000003}, {"city": "Rudow", "job_title": "Chief Executive Officer", "department": "Operations", "email": "ejgbihf@mail.com", "unit": "kg", "signup_date": "2023-09-28", "rate": 0.6890000000000001}, {"city": "Koduvayur", "job_title": "Chief Executive Officer", "department": "Operations", "email": "daca@example.com", "unit": "kg", "signup_date": "2023-02-28", "rate": 0.259}, {"city": "San Antonio Oeste", "job_title": "Director", "department": "Operations", "email": "phijl@mail.com", "unit": "mi", "signup_date": "2023-12-23", "rate": 0.937}, {"city": "Badarganj", "job_title": "Director", "department": "Human Resources", "email": "effifgp@mail.com", "unit": "kg", "signup_date": "2023-04-02", "rate": 0.9640000000000001}, {"city": "Badarganj", "job_title": "Chief Executive Officer", "department": "Human Resources", "email": "lgfgga@mail.com", "unit": "lb", "signup_date": "2023-07-25", "rate": 0.57}, {"city": "Badarganj", "job_title": "Chief Executive Officer", "department": "Operations", "email": "lapll@example.com", "unit": "lb", "signup_date": "2023-07-28", "rate": 0.33899999999999997}, {"city": "San Antonio Oeste", "job_title": "Director", "department": "Operations", "email": "fangd@mail.com", "unit": "lb", "signup_date": "2023-09-12", "rate": 0.6}, {"city": "Badarganj", "job_title": "Director", "department": "Operations", "email": "lkdmdcg@example.com", "unit": "lb", "signup_date": "2023-03-16", "rate": 0.5379999999999999}], "dirty_cols": ["city", "job_title", "department", "email", "unit", "signup_date", "rate"], "clean_cols": ["city", "job_title", "department", "email", "unit", "signup_date", "rate"], "plan": {"dataset_summary": "49 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Noswhera Kalan": "Nowshera Kalan", "BADARGANJ": "Badarganj", "KODUVAYUR": "Koduvayur", "Baadrganj": "Badarganj", "San Antonio OOeste": "San Antonio Oeste", "koduvayur": "Koduvayur", "rudow": "Rudow", "badarganj": "Badarganj", "nowshera kalan": "Nowshera Kalan", "Kodduvayur": "Koduvayur", "san antonio oeste": "San Antonio Oeste", "ruudow": "Rudow", "san antofio oeste": "San Antonio Oeste", "SAN ANTONIO OESTE": "San Antonio Oeste"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CEO": "Chief Executive Officer", "chief executive officer": "Chief Executive Officer", "c.e.o.": "Chief Executive Officer", "Chief Executivj Officer": "Chief Executive Officer", "Dir": "Director", "DIR": "Director", "director": "Director", "ceo": "Chief Executive Officer", "Dir.": "Director", "DIRECTOR": "Director", "C.E.O.": "Chief Executive Officer", "DR.": "Director", "C.EO..": "Chief Executive Officer", "Dzr.": "Director", "dir.": "Director", "C..EO.": "Chief Executive Officer", "di.": "Director"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"people ops": "Human Resources", "Ops": "Operations", "OPERATIONS": "Operations", "operations": "Operations", "OPS": "Operations", "People Ops": "Human Resources", "HUMAN RESOURCES": "Human Resources", "hr": "Human Resources", "ops": "Operations", "HR": "Human Resources", "operationns": "Operations", "Hr": "Human Resources", "Human eRsources": "Human Resources", "people": "Human Resources", "human resources": "Human Resources", "People": "Human Resources", "People Bps": "Human Resources", "operatsons": "Operations", "PEOPLE": "Human Resources", "oeprations": "Operations", "PEOPLE OPS": "Human Resources"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pouods": "lb", "KG": "kg", "lbs": "lb", "kilogram": "kg", "pound": "lb", "Lb": "lb", "Kilogram": "kg", "miile": "mi", "MI": "mi", "miles": "mi", "kilograms": "kg", "pounds": "lb", "mile": "mi", "MILES": "mi", "Mi": "mi", "POUNDS": "lb"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"status": "opne", "email": "CLFFM@EXAMPLE.COM", "city": "Eppendorf", "unnamed": ""}, {"status": "   Open", "email": "GCKAGD@TEST.ORG ", "city": "An Nimas  ", "unnamed": ""}, {"status": "in  progress", "email": "  CCBBPHO@CORP.IO", "city": "eppendorf", "unnamed": ""}, {"status": "open", "email": "   pikef@test.org", "city": "Af Nimas", "unnamed": ""}, {"status": "won", "email": "KLMBIIM@CORP.IO ", "city": "an muileann gcearr", "unnamed": ""}, {"status": " closed lost", "email": "NFMIA@EXAMPLE.COM", "city": "Johannesburg", "unnamed": ""}, {"status": "in progress", "email": "BHOJP@CORP.IO", "city": "johannesburg", "unnamed": ""}, {"status": "Closed Lost", "email": "geke@test.org", "city": "johannesburg", "unnamed": ""}, {"status": "ongoing  ", "email": " KDHC@MAIL.COM ", "city": "An Nimas", "unnamed": ""}, {"status": "closed-won  ", "email": "lnabiil@example.com ", "city": " Eppendorf", "unnamed": ""}, {"status": "lost", "email": "dejp@test.org", "city": "Eppendorf", "unnamed": ""}, {"status": "CLOSED-SON  ", "email": "kfogc@example.com", "city": "an nmas", "unnamed": ""}, {"status": "won", "email": " jifhhci@mail.com ", "city": "an muileann gcearr", "unnamed": ""}, {"status": "lost", "email": "dejp@test.org", "city": "Eppendorf", "unnamed": ""}, {"status": "", "email": "", "city": "", "unnamed": ""}], "clean": [{"status": "Open", "email": "clffm@example.com", "city": "Eppendorf"}, {"status": "Open", "email": "gckagd@test.org", "city": "An Nimas"}, {"status": "In Progress", "email": "ccbbpho@corp.io", "city": "Eppendorf"}, {"status": "Open", "email": "pikef@test.org", "city": "An Nimas"}, {"status": "Won", "email": "klmbiim@corp.io", "city": "An Muileann gCearr"}, {"status": "Lost", "email": "nfmia@example.com", "city": "Johannesburg"}, {"status": "In Progress", "email": "bhojp@corp.io", "city": "Johannesburg"}, {"status": "Lost", "email": "geke@test.org", "city": "Johannesburg"}, {"status": "In Progress", "email": "kdhc@mail.com", "city": "An Nimas"}, {"status": "Won", "email": "lnabiil@example.com", "city": "Eppendorf"}, {"status": "Lost", "email": "dejp@test.org", "city": "Eppendorf"}, {"status": "Won", "email": "kfogc@example.com", "city": "An Nimas"}, {"status": "Won", "email": "jifhhci@mail.com", "city": "An Muileann gCearr"}], "dirty_cols": ["status", "email", "city", "unnamed"], "clean_cols": ["status", "email", "city"], "plan": {"dataset_summary": "15 rows × 4 columns. 3 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"opne": "Open", "in progress": "In Progress", "open": "Open", "won": "Won", "closed lost": "Lost", "Closed Lost": "Lost", "ongoing": "In Progress", "closed-won": "Won", "lost": "Lost", "CLOSED-SON": "Won"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"eppendorf": "Eppendorf", "Af Nimas": "An Nimas", "an muileann gcearr": "An Muileann gCearr", "johannesburg": "Johannesburg", "an nmas": "An Nimas"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "15 Aug 2023", "industry": "telco", "city": "MSOTAGANEM", "currency": "Nakfa", "is_active": "0"}, {"signup_date": "7/23/2023", "industry": "technology", "city": "marbach am neckar", "currency": "ern", "is_active": "true"}, {"signup_date": "2023-10-08", "industry": "Software", "city": "marbach  am neckar", "currency": "egyptian pound", "is_active": "N"}, {"signup_date": "2/14/2023", "industry": "telecom", "city": "MARBACH AM NECKAR", "currency": "azn", "is_active": "true"}, {"signup_date": "26 Nov 2023", "industry": "   COMMS", "city": "MARBACH AM NECKAR", "currency": "AZN", "is_active": "F"}, {"signup_date": "11/17/2023", "industry": "Telco", "city": "Mostaganem", "currency": "egp", "is_active": "No"}, {"signup_date": "28 Dec 2023", "industry": "Commms", "city": "mostaganem", "currency": "Egyptian Pound", "is_active": "TRUE"}, {"signup_date": "18 Jul 2023", "industry": "IT", "city": " mostaganem ", "currency": " NAKFA ", "is_active": "true"}, {"signup_date": "45237", "industry": "Software", "city": "marbach am neckar", "currency": " AZN", "is_active": "0"}, {"signup_date": "2023-12-21", "industry": "Technology", "city": "   Mostaganem", "currency": "ERN", "is_active": "N"}, {"signup_date": "24 Feb 2023", "industry": "Telco", "city": "mostaganem", "currency": "EGYPTIAN POUD", "is_active": "T"}, {"signup_date": "28 Aug 2023", "industry": "Tech", "city": "MARBACH AM  NECKAR", "currency": "Nakfa", "is_active": "T"}, {"signup_date": "45276", "industry": "Telco", "city": " MOSTAGANEM ", "currency": "EGP  ", "is_active": "N"}, {"signup_date": "45009", "industry": " Technology", "city": "Mostaganem", "currency": "AZERBAIJAN MANAT  ", "is_active": "No"}, {"signup_date": "22 Jan 2023", "industry": "telecom", "city": "marbach am neckar", "currency": "Egyptian Pound", "is_active": "1"}, {"signup_date": "2023-08-23", "industry": "Comms", "city": "mostaganem", "currency": "AZERBAIJAN MANAT", "is_active": "true"}], "clean": [{"signup_date": "2023-08-15", "industry": "Telecommunications", "city": "Mostaganem", "currency": "ERN", "is_active": false}, {"signup_date": "2023-07-23", "industry": "Technology", "city": "Marbach am Neckar", "currency": "ERN", "is_active": true}, {"signup_date": "2023-10-08", "industry": "Technology", "city": "Marbach am Neckar", "currency": "EGP", "is_active": false}, {"signup_date": "2023-02-14", "industry": "Telecommunications", "city": "Marbach am Neckar", "currency": "AZN", "is_active": true}, {"signup_date": "2023-11-26", "industry": "Telecommunications", "city": "Marbach am Neckar", "currency": "AZN", "is_active": false}, {"signup_date": "2023-11-17", "industry": "Telecommunications", "city": "Mostaganem", "currency": "EGP", "is_active": false}, {"signup_date": "2023-12-28", "industry": "Telecommunications", "city": "Mostaganem", "currency": "EGP", "is_active": true}, {"signup_date": "2023-07-18", "industry": "Technology", "city": "Mostaganem", "currency": "ERN", "is_active": true}, {"signup_date": "2023-11-07", "industry": "Technology", "city": "Marbach am Neckar", "currency": "AZN", "is_active": false}, {"signup_date": "2023-12-21", "industry": "Technology", "city": "Mostaganem", "currency": "ERN", "is_active": false}, {"signup_date": "2023-02-24", "industry": "Telecommunications", "city": "Mostaganem", "currency": "EGP", "is_active": true}, {"signup_date": "2023-08-28", "industry": "Technology", "city": "Marbach am Neckar", "currency": "ERN", "is_active": true}, {"signup_date": "2023-12-16", "industry": "Telecommunications", "city": "Mostaganem", "currency": "EGP", "is_active": false}, {"signup_date": "2023-03-24", "industry": "Technology", "city": "Mostaganem", "currency": "AZN", "is_active": false}, {"signup_date": "2023-01-22", "industry": "Telecommunications", "city": "Marbach am Neckar", "currency": "EGP", "is_active": true}, {"signup_date": "2023-08-23", "industry": "Telecommunications", "city": "Mostaganem", "currency": "AZN", "is_active": true}], "dirty_cols": ["signup_date", "industry", "city", "currency", "is_active"], "clean_cols": ["signup_date", "industry", "city", "currency", "is_active"], "plan": {"dataset_summary": "16 rows × 5 columns. 5 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"telco": "Telecommunications", "technology": "Technology", "Software": "Technology", "telecom": "Telecommunications", "COMMS": "Telecommunications", "Telco": "Telecommunications", "Commms": "Telecommunications", "IT": "Technology", "Tech": "Technology", "Comms": "Telecommunications"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MSOTAGANEM": "Mostaganem", "marbach am neckar": "Marbach am Neckar", "MARBACH AM NECKAR": "Marbach am Neckar", "mostaganem": "Mostaganem", "MOSTAGANEM": "Mostaganem"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Nakfa": "ERN", "ern": "ERN", "egyptian pound": "EGP", "azn": "AZN", "egp": "EGP", "Egyptian Pound": "EGP", "NAKFA": "ERN", "EGYPTIAN POUD": "EGP", "AZERBAIJAN MANAT": "AZN"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"name": "   Carol Park", "city": "Akouda", "company": "Wonka Co"}, {"name": "Mona Moore", "city": "STALLINGS", "company": "Soylent Corp "}, {"name": "Priya  Petrov", "city": "Stallings", "company": "Vehement"}, {"name": "Grace Diaz  ", "city": "stalings", "company": "   Vehement"}, {"name": "Karl Lee  ", "city": "Stallings", "company": "Vehement  "}, {"name": "Priya Petrov", "city": "Akouda", "company": "Vehement"}, {"name": "Grace Ali", "city": "Stallings", "company": "Soylent Corp"}, {"name": "Karl Diaz", "city": "stalings", "company": "Hooli"}, {"name": "Bob Adams", "city": "AKOUDA", "company": "Globex"}, {"name": "Sara Cruz", "city": "STALLINGS", "company": "Hooli"}, {"name": "  Heidi Smith", "city": "Akouda   ", "company": "Cyberdyne"}, {"name": "Sara Cruz", "city": "   Akouda", "company": "Umbrella"}, {"name": "  Frank Fischer", "city": "akouda  ", "company": "Cyberdyne"}, {"name": "  Ivan Cruz", "city": "Stallings", "company": "Acme Inc"}, {"name": "Judy  Khan", "city": "stallnigs ", "company": "Stark Industries"}, {"name": "   Bob Moore", "city": "akouda", "company": "Globex"}], "clean": [{"name": "Carol Park", "city": "Akouda", "company": "Wonka Co"}, {"name": "Mona Moore", "city": "Stallings", "company": "Soylent Corp"}, {"name": "Priya Petrov", "city": "Stallings", "company": "Vehement"}, {"name": "Grace Diaz", "city": "Stallings", "company": "Vehement"}, {"name": "Karl Lee", "city": "Stallings", "company": "Vehement"}, {"name": "Priya Petrov", "city": "Akouda", "company": "Vehement"}, {"name": "Grace Ali", "city": "Stallings", "company": "Soylent Corp"}, {"name": "Karl Diaz", "city": "Stallings", "company": "Hooli"}, {"name": "Bob Adams", "city": "Akouda", "company": "Globex"}, {"name": "Sara Cruz", "city": "Stallings", "company": "Hooli"}, {"name": "Heidi Smith", "city": "Akouda", "company": "Cyberdyne"}, {"name": "Sara Cruz", "city": "Akouda", "company": "Umbrella"}, {"name": "Frank Fischer", "city": "Akouda", "company": "Cyberdyne"}, {"name": "Ivan Cruz", "city": "Stallings", "company": "Acme Inc"}, {"name": "Judy Khan", "city": "Stallings", "company": "Stark Industries"}, {"name": "Bob Moore", "city": "Akouda", "company": "Globex"}], "dirty_cols": ["name", "city", "company"], "clean_cols": ["name", "city", "company"], "plan": {"dataset_summary": "16 rows × 3 columns. 3 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"STALLINGS": "Stallings", "stalings": "Stallings", "AKOUDA": "Akouda", "akouda": "Akouda", "stallnigs": "Stallings"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"department": "Prod", "state": "KETNUCKY", "phone": "(582)886-5085", "col_x": ""}, {"department": "Product", "state": "ut", "phone": "593.680.4456", "col_x": ""}, {"department": "CS", "state": "tennessee", "phone": "564-379-7228", "col_x": ""}, {"department": "  support", "state": "utah", "phone": "(572)938-0617", "col_x": ""}, {"department": "Prod", "state": "   kentucky", "phone": "5640096480", "col_x": ""}, {"department": "Support", "state": " Kentucky", "phone": "5461573723", "col_x": ""}, {"department": "product", "state": "Utah", "phone": "5429324659", "col_x": ""}, {"department": "cust support", "state": "Tennessee", "phone": "502.398.1315", "col_x": ""}, {"department": "Customer Support", "state": " Ut", "phone": "5453539944", "col_x": ""}, {"department": "support", "state": "  kentucky", "phone": "523.049.3592", "col_x": ""}, {"department": " Prod", "state": " tn ", "phone": "5971711818", "col_x": ""}, {"department": "Prodduct", "state": "TN", "phone": "522.526.4720", "col_x": ""}], "clean": [{"department": "Product", "state": "Kentucky", "phone": "(582) 886-5085"}, {"department": "Product", "state": "Utah", "phone": "(593) 680-4456"}, {"department": "Customer Support", "state": "Tennessee", "phone": "(564) 379-7228"}, {"department": "Customer Support", "state": "Utah", "phone": "(572) 938-0617"}, {"department": "Product", "state": "Kentucky", "phone": "(564) 009-6480"}, {"department": "Customer Support", "state": "Kentucky", "phone": "(546) 157-3723"}, {"department": "Product", "state": "Utah", "phone": "(542) 932-4659"}, {"department": "Customer Support", "state": "Tennessee", "phone": "(502) 398-1315"}, {"department": "Customer Support", "state": "Utah", "phone": "(545) 353-9944"}, {"department": "Customer Support", "state": "Kentucky", "phone": "(523) 049-3592"}, {"department": "Product", "state": "Tennessee", "phone": "(597) 171-1818"}, {"department": "Product", "state": "Tennessee", "phone": "(522) 526-4720"}], "dirty_cols": ["department", "state", "phone", "col_x"], "clean_cols": ["department", "state", "phone"], "plan": {"dataset_summary": "12 rows × 4 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Prod": "Product", "CS": "Customer Support", "support": "Customer Support", "Support": "Customer Support", "product": "Product", "cust support": "Customer Support", "Prodduct": "Product"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KETNUCKY": "Kentucky", "ut": "Utah", "tennessee": "Tennessee", "utah": "Utah", "kentucky": "Kentucky", "Ut": "Utah", "tn": "Tennessee", "TN": "Tennessee"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"amount": "$4,829.57", "phone": "506-685-2822", "country": "kna", "unnamed": ""}, {"amount": "3.056,40", "phone": "5184388753", "country": "montenegro", "unnamed": ""}, {"amount": "2.687,45", "phone": "568.834.3884", "country": "Jamaira", "unnamed": ""}, {"amount": "1.216,53", "phone": "520-059-6529", "country": "MONTENEGRO", "unnamed": ""}, {"amount": "3.216,83", "phone": "546.099.0504", "country": " kna", "unnamed": ""}, {"amount": "$6,906.90", "phone": "5578090447", "country": "ME", "unnamed": ""}, {"amount": "$8,674.11", "phone": "5750879334", "country": "KN", "unnamed": ""}, {"amount": "6.453,71", "phone": "5767879323", "country": " KN ", "unnamed": ""}, {"amount": "$6,020.58", "phone": "558.983.5067", "country": "jam", "unnamed": ""}, {"amount": "6.671,83", "phone": "(569)688-1116", "country": "jmaica ", "unnamed": ""}, {"amount": "$7,011.53", "phone": "512.874.0755", "country": "Saint Kitts and Nevis", "unnamed": ""}, {"amount": "$7,390.95", "phone": "512.281.6707", "country": "jam", "unnamed": ""}, {"amount": "$3,189.10", "phone": "(535)022-5678", "country": " CG ", "unnamed": ""}, {"amount": "8.122,07", "phone": "5497041648", "country": "cog", "unnamed": ""}, {"amount": "$607.73", "phone": "565-591-1696", "country": "Jamaica", "unnamed": ""}, {"amount": "4.927,56", "phone": "(516)796-7953", "country": "JAM", "unnamed": ""}, {"amount": "$8,674.11", "phone": "5750879334", "country": "KN", "unnamed": ""}], "clean": [{"amount": 4829.57, "phone": "(506) 685-2822", "country": "Saint Kitts and Nevis"}, {"amount": 3056.4, "phone": "(518) 438-8753", "country": "Montenegro"}, {"amount": 2687.45, "phone": "(568) 834-3884", "country": "Jamaica"}, {"amount": 1216.53, "phone": "(520) 059-6529", "country": "Montenegro"}, {"amount": 3216.83, "phone": "(546) 099-0504", "country": "Saint Kitts and Nevis"}, {"amount": 6906.9, "phone": "(557) 809-0447", "country": "Montenegro"}, {"amount": 8674.11, "phone": "(575) 087-9334", "country": "Saint Kitts and Nevis"}, {"amount": 6453.71, "phone": "(576) 787-9323", "country": "Saint Kitts and Nevis"}, {"amount": 6020.58, "phone": "(558) 983-5067", "country": "Jamaica"}, {"amount": 6671.83, "phone": "(569) 688-1116", "country": "Jamaica"}, {"amount": 7011.53, "phone": "(512) 874-0755", "country": "Saint Kitts and Nevis"}, {"amount": 7390.95, "phone": "(512) 281-6707", "country": "Jamaica"}, {"amount": 3189.1, "phone": "(535) 022-5678", "country": "Congo"}, {"amount": 8122.07, "phone": "(549) 704-1648", "country": "Congo"}, {"amount": 607.73, "phone": "(565) 591-1696", "country": "Jamaica"}, {"amount": 4927.56, "phone": "(516) 796-7953", "country": "Jamaica"}], "dirty_cols": ["amount", "phone", "country", "unnamed"], "clean_cols": ["amount", "phone", "country"], "plan": {"dataset_summary": "17 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kna": "Saint Kitts and Nevis", "montenegro": "Montenegro", "Jamaira": "Jamaica", "MONTENEGRO": "Montenegro", "ME": "Montenegro", "KN": "Saint Kitts and Nevis", "jam": "Jamaica", "jmaica": "Jamaica", "CG": "Congo", "cog": "Congo", "JAM": "Jamaica"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": " Supmort ", "currency": "  lari", "is_active": "false", "email": "PFNJ@TEST.ORG", "company": "Wonka Co  ", "city": "moose jaw   "}, {"department": "People", "currency": "   Bahamian Dollar", "is_active": "TRUE", "email": " ddigi@test.org", "company": "  Vehement", "city": "Moose Jaw"}, {"department": "Customer Support", "currency": "Lari", "is_active": "FALSE", "email": " colnbd@mail.com ", "company": " Initech ", "city": "mxose jaw"}, {"department": "HR", "currency": "Bahamian Dollar", "is_active": "Yes", "email": "EINFJFE@TEST.ORG", "company": "Vehement  ", "city": "Shahecheng"}, {"department": "people ops", "currency": "lari", "is_active": "FALSE", "email": "  amaec@test.org", "company": "Cyberdyne", "city": "Balqash"}, {"department": "CUSTOMER SUPPORT", "currency": "BSD", "is_active": "Yes", "email": "JNJCI@TEST.ORG", "company": "Acme  Inc", "city": "balqash"}, {"department": "CS ", "currency": "BERMUDIAN DOLLAR", "is_active": "1", "email": "mgiojb@test.org", "company": "Soylent Corp", "city": "balqash"}, {"department": "CS", "currency": "Bahamian Dollar", "is_active": "TRUE", "email": "   FHIGMMK@EXAMPLE.COM", "company": "Initech", "city": "Moose Jaw"}, {"department": "cs", "currency": " bmd ", "is_active": "TRUE", "email": " pcnllpn@corp.io ", "company": "  Umbrella", "city": "Shahecheng"}, {"department": " SUPPORT ", "currency": "Cuban Peso", "is_active": "FALSE", "email": "cnjgb@corp.io   ", "company": "Wonka Co", "city": "Soamoso"}, {"department": "People", "currency": "Bermudian Dollar", "is_active": "No", "email": "  NBIG@MAIL.COM", "company": " Vehement ", "city": "sogamoso"}, {"department": "Cust Support", "currency": " BSD ", "is_active": "T", "email": "  cipabij@mail.com", "company": "Cyberdyne ", "city": "balqash"}, {"department": "PEOPLE OPS", "currency": "GEL", "is_active": "TRUE", "email": " gkncaca@test.org ", "company": "Stark Industries ", "city": "Sogamoso"}, {"department": "CS", "currency": "gel", "is_active": "T", "email": "jffp@example.com", "company": "Umbrella ", "city": "Sogamoos"}, {"department": "cust support", "currency": "CUP   ", "is_active": "N", "email": "bpfig@corp.io", "company": "Stark Industries ", "city": "Hong Kong"}, {"department": "Cust Support", "currency": "GEL", "is_active": "F", "email": "gmoipm@mail.com ", "company": " Globex ", "city": "shahecheng"}, {"department": "Customer Support", "currency": "lari", "is_active": "true", "email": "bnobe@corp.io", "company": "Vehement", "city": "shahecheng"}, {"department": "cs ", "currency": "bahamian  dollar", "is_active": "T", "email": "dojpioo@mail.com", "company": "Vehement", "city": "Shahecheng"}, {"department": "  Human Resources", "currency": "cup", "is_active": "Yes", "email": "bdcan@test.org", "company": "Stark Industries", "city": "mose jaw"}, {"department": "people ops", "currency": "Lari", "is_active": "1", "email": " JFJIO@MAIL.COM ", "company": "Umbrella", "city": "Moose Jaw"}], "clean": [{"department": "Customer Support", "currency": "GEL", "is_active": false, "email": "pfnj@test.org", "company": "Wonka Co", "city": "Moose Jaw"}, {"department": "Human Resources", "currency": "BSD", "is_active": true, "email": "ddigi@test.org", "company": "Vehement", "city": "Moose Jaw"}, {"department": "Customer Support", "currency": "GEL", "is_active": false, "email": "colnbd@mail.com", "company": "Initech", "city": "Moose Jaw"}, {"department": "Human Resources", "currency": "BSD", "is_active": true, "email": "einfjfe@test.org", "company": "Vehement", "city": "Shahecheng"}, {"department": "Human Resources", "currency": "GEL", "is_active": false, "email": "amaec@test.org", "company": "Cyberdyne", "city": "Balqash"}, {"department": "Customer Support", "currency": "BSD", "is_active": true, "email": "jnjci@test.org", "company": "Acme Inc", "city": "Balqash"}, {"department": "Customer Support", "currency": "BMD", "is_active": true, "email": "mgiojb@test.org", "company": "Soylent Corp", "city": "Balqash"}, {"department": "Customer Support", "currency": "BSD", "is_active": true, "email": "fhigmmk@example.com", "company": "Initech", "city": "Moose Jaw"}, {"department": "Customer Support", "currency": "BMD", "is_active": true, "email": "pcnllpn@corp.io", "company": "Umbrella", "city": "Shahecheng"}, {"department": "Customer Support", "currency": "CUP", "is_active": false, "email": "cnjgb@corp.io", "company": "Wonka Co", "city": "Sogamoso"}, {"department": "Human Resources", "currency": "BMD", "is_active": false, "email": "nbig@mail.com", "company": "Vehement", "city": "Sogamoso"}, {"department": "Customer Support", "currency": "BSD", "is_active": true, "email": "cipabij@mail.com", "company": "Cyberdyne", "city": "Balqash"}, {"department": "Human Resources", "currency": "GEL", "is_active": true, "email": "gkncaca@test.org", "company": "Stark Industries", "city": "Sogamoso"}, {"department": "Customer Support", "currency": "GEL", "is_active": true, "email": "jffp@example.com", "company": "Umbrella", "city": "Sogamoso"}, {"department": "Customer Support", "currency": "CUP", "is_active": false, "email": "bpfig@corp.io", "company": "Stark Industries", "city": "Hong Kong"}, {"department": "Customer Support", "currency": "GEL", "is_active": false, "email": "gmoipm@mail.com", "company": "Globex", "city": "Shahecheng"}, {"department": "Customer Support", "currency": "GEL", "is_active": true, "email": "bnobe@corp.io", "company": "Vehement", "city": "Shahecheng"}, {"department": "Customer Support", "currency": "BSD", "is_active": true, "email": "dojpioo@mail.com", "company": "Vehement", "city": "Shahecheng"}, {"department": "Human Resources", "currency": "CUP", "is_active": true, "email": "bdcan@test.org", "company": "Stark Industries", "city": "Moose Jaw"}, {"department": "Human Resources", "currency": "GEL", "is_active": true, "email": "jfjio@mail.com", "company": "Umbrella", "city": "Moose Jaw"}], "dirty_cols": ["department", "currency", "is_active", "email", "company", "city"], "clean_cols": ["department", "currency", "is_active", "email", "company", "city"], "plan": {"dataset_summary": "20 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Supmort": "Customer Support", "People": "Human Resources", "HR": "Human Resources", "people ops": "Human Resources", "CUSTOMER SUPPORT": "Customer Support", "CS": "Customer Support", "cs": "Customer Support", "SUPPORT": "Customer Support", "Cust Support": "Customer Support", "PEOPLE OPS": "Human Resources", "cust support": "Customer Support"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"lari": "GEL", "Bahamian Dollar": "BSD", "Lari": "GEL", "BERMUDIAN DOLLAR": "BMD", "bmd": "BMD", "Cuban Peso": "CUP", "Bermudian Dollar": "BMD", "gel": "GEL", "bahamian dollar": "BSD", "cup": "CUP"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"moose jaw": "Moose Jaw", "mxose jaw": "Moose Jaw", "balqash": "Balqash", "Soamoso": "Sogamoso", "sogamoso": "Sogamoso", "Sogamoos": "Sogamoso", "shahecheng": "Shahecheng", "mose jaw": "Moose Jaw"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": " Lena Brandt", "city": "Saint-Constant", "company": "Stark Industries", "department": "PEOPLE", "industry": "Financial Services   ", "col_x": ""}, {"name": "Priya Lee", "city": "sheoganj", "company": "Stark Industries  ", "department": "Ops   ", "industry": "Banking", "col_x": ""}, {"name": "  Mona Khan", "city": "Sheoanj  ", "company": "Acme Inc", "department": "accounting", "industry": "FINSERV", "col_x": ""}, {"name": "Alice Moore", "city": "Yanacancha", "company": "Wonka Co", "department": "Ops  ", "industry": "FinServ", "col_x": ""}, {"name": "Grace Khan ", "city": "SaintConstant", "company": "Vehement", "department": " Fin ", "industry": "It", "col_x": ""}, {"name": "Lena Ali  ", "city": "Saint-Constant", "company": "Soylent Corp", "department": "Finance", "industry": "  hospitality", "col_x": ""}, {"name": "Sara Smith", "city": " Sheoganj ", "company": "Initech", "department": "HR ", "industry": "travel", "col_x": ""}, {"name": "  David Moore", "city": "SaintConstant", "company": " Globex ", "department": "People Ops", "industry": "travel", "col_x": ""}, {"name": "Bob Reyes", "city": "sait-constant", "company": "Wonka  Co", "department": " Fin ", "industry": " hoslitality ", "col_x": ""}, {"name": "Lena Fischer ", "city": "Sheoganj", "company": "   Wonka Co", "department": " Ops ", "industry": "Cimms", "col_x": ""}, {"name": "Grace Johnson", "city": "SHEOGANJ", "company": "Hooli", "department": "Ops  ", "industry": "Turism", "col_x": ""}, {"name": "Judy Johnson", "city": "Yanacancha", "company": "Vehement", "department": "  people ops", "industry": "technology", "col_x": ""}, {"name": "David Reyes", "city": "SHEOGANJ", "company": "Cyberdyne", "department": "FIN", "industry": "Banking", "col_x": ""}, {"name": "  Priya Moore", "city": "Saint-Coonstant", "company": "Vehement   ", "department": "fin", "industry": "comms", "col_x": ""}, {"name": "Mona Reyes", "city": "sheoganj", "company": "Globex", "department": "finacne", "industry": "Comms", "col_x": ""}, {"name": " David Smith", "city": "saint-constant", "company": "Umbrella  ", "department": "PEOPLE OPS", "industry": "   Fintech", "col_x": ""}, {"name": " Heidi Diaz", "city": "Sheogaj  ", "company": "   Acme Inc", "department": "Pepole   ", "industry": "Technology", "col_x": ""}, {"name": "Karl Moore", "city": "Yanacancha", "company": " Wonka Co", "department": "Ops", "industry": " Finance ", "col_x": ""}, {"name": "   Lena Lee", "city": "Saint-Constant", "company": "Soylent Corp ", "department": "Accounting", "industry": "travel", "col_x": ""}, {"name": "Bob Park", "city": "Saint-Constant", "company": "Umbrella", "department": "Fin", "industry": "it", "col_x": ""}, {"name": "Omar Park  ", "city": "SHEOGANJ", "company": "Vehement   ", "department": "Finance", "industry": "Fintech ", "col_x": ""}, {"name": " Heidi Fischer", "city": "saint-constant", "company": "Acme Inc", "department": " ops", "industry": "Software", "col_x": ""}, {"name": "  Bob Ali", "city": "Yanaqancha", "company": "Soylent Corp  ", "department": "operations", "industry": "FINSERV", "col_x": ""}, {"name": "Ivan Brandt   ", "city": "Sheoganj", "company": "Cyberdyne", "department": " ACCOUNTING ", "industry": "COMMS", "col_x": ""}, {"name": "Frank Ali", "city": "saint-constant", "company": "Soylent Corp   ", "department": "Operations", "industry": "Hospitality", "col_x": ""}, {"name": "Carol Cruz   ", "city": "SaintConstant", "company": "   Cyberdyne", "department": "Ops", "industry": "TECRNOLOGY", "col_x": ""}, {"name": "Omar Khan", "city": "Yanacancha", "company": " Wonka Co", "department": "Operations", "industry": "technoloogy", "col_x": ""}, {"name": "David Park", "city": "   yanacancha", "company": "Initech", "department": "   Operations", "industry": "HOSPITALITY", "col_x": ""}, {"name": "Judy Reyes", "city": "Sheoganj ", "company": "  Globex", "department": "Operations", "industry": "tech", "col_x": ""}, {"name": "Priya Johnson   ", "city": "SAINT-CONSTANT   ", "company": "Umbrella", "department": "Ops", "industry": "financial services", "col_x": ""}, {"name": "Alice  Park", "city": "Sheoganj", "company": " Umbrella ", "department": " OPERATIONS ", "industry": "TRAVEL", "col_x": ""}, {"name": "Grace Diaz", "city": "Shoeganj", "company": "Stark Industries", "department": "hr", "industry": "BANKING", "col_x": ""}, {"name": "Judy Lee ", "city": " YANACANCHA ", "company": "Umbrella", "department": "Fitance", "industry": "fintech", "col_x": ""}, {"name": "Priya Petrov", "city": "   Yinacancha", "company": "Soylent Corp", "department": "Hr", "industry": "hospitality", "col_x": ""}, {"name": "Grace Brandt", "city": " Suint-Constant ", "company": "Cyberdyne", "department": "Human Resources", "industry": "tceh", "col_x": ""}, {"name": "Ivan Brandt", "city": "SHEOGANJ", "company": "Vehement", "department": "HR", "industry": "Banking", "col_x": ""}, {"name": "Ivan  Novak", "city": "Sheoganj", "company": "Stark Industries", "department": " Fin ", "industry": "Comms", "col_x": ""}, {"name": "Judy  Cruz", "city": "Saint-Constant", "company": "Soylent Corp", "department": "Operations  ", "industry": "FinServ", "col_x": ""}, {"name": "Eve Ali  ", "city": "yanacancha", "company": "Globex", "department": "Operations", "industry": "travel", "col_x": ""}, {"name": "Ivan Khan  ", "city": "sheoganj", "company": " Wonka Co", "department": "Accounting", "industry": "Hotels & Travel", "col_x": ""}, {"name": "David Brandt ", "city": "YANACANCHA", "company": "Vehement", "department": "accounting", "industry": "Technology", "col_x": ""}, {"name": "Ivan  Novak", "city": "Sheoganj", "company": "Stark Industries", "department": " Fin ", "industry": "Comms", "col_x": ""}], "clean": [{"name": "Lena Brandt", "city": "Saint-Constant", "company": "Stark Industries", "department": "Human Resources", "industry": "Financial Services"}, {"name": "Priya Lee", "city": "Sheoganj", "company": "Stark Industries", "department": "Operations", "industry": "Financial Services"}, {"name": "Mona Khan", "city": "Sheoganj", "company": "Acme Inc", "department": "Finance", "industry": "Financial Services"}, {"name": "Alice Moore", "city": "Yanacancha", "company": "Wonka Co", "department": "Operations", "industry": "Financial Services"}, {"name": "Grace Khan", "city": "Saint-Constant", "company": "Vehement", "department": "Finance", "industry": "Technology"}, {"name": "Lena Ali", "city": "Saint-Constant", "company": "Soylent Corp", "department": "Finance", "industry": "Hospitality"}, {"name": "Sara Smith", "city": "Sheoganj", "company": "Initech", "department": "Human Resources", "industry": "Hospitality"}, {"name": "David Moore", "city": "Saint-Constant", "company": "Globex", "department": "Human Resources", "industry": "Hospitality"}, {"name": "Bob Reyes", "city": "Saint-Constant", "company": "Wonka Co", "department": "Finance", "industry": "Hospitality"}, {"name": "Lena Fischer", "city": "Sheoganj", "company": "Wonka Co", "department": "Operations", "industry": "Telecommunications"}, {"name": "Grace Johnson", "city": "Sheoganj", "company": "Hooli", "department": "Operations", "industry": "Hospitality"}, {"name": "Judy Johnson", "city": "Yanacancha", "company": "Vehement", "department": "Human Resources", "industry": "Technology"}, {"name": "David Reyes", "city": "Sheoganj", "company": "Cyberdyne", "department": "Finance", "industry": "Financial Services"}, {"name": "Priya Moore", "city": "Saint-Constant", "company": "Vehement", "department": "Finance", "industry": "Telecommunications"}, {"name": "Mona Reyes", "city": "Sheoganj", "company": "Globex", "department": "Finance", "industry": "Telecommunications"}, {"name": "David Smith", "city": "Saint-Constant", "company": "Umbrella", "department": "Human Resources", "industry": "Financial Services"}, {"name": "Heidi Diaz", "city": "Sheoganj", "company": "Acme Inc", "department": "Human Resources", "industry": "Technology"}, {"name": "Karl Moore", "city": "Yanacancha", "company": "Wonka Co", "department": "Operations", "industry": "Financial Services"}, {"name": "Lena Lee", "city": "Saint-Constant", "company": "Soylent Corp", "department": "Finance", "industry": "Hospitality"}, {"name": "Bob Park", "city": "Saint-Constant", "company": "Umbrella", "department": "Finance", "industry": "Technology"}, {"name": "Omar Park", "city": "Sheoganj", "company": "Vehement", "department": "Finance", "industry": "Financial Services"}, {"name": "Heidi Fischer", "city": "Saint-Constant", "company": "Acme Inc", "department": "Operations", "industry": "Technology"}, {"name": "Bob Ali", "city": "Yanacancha", "company": "Soylent Corp", "department": "Operations", "industry": "Financial Services"}, {"name": "Ivan Brandt", "city": "Sheoganj", "company": "Cyberdyne", "department": "Finance", "industry": "Telecommunications"}, {"name": "Frank Ali", "city": "Saint-Constant", "company": "Soylent Corp", "department": "Operations", "industry": "Hospitality"}, {"name": "Carol Cruz", "city": "Saint-Constant", "company": "Cyberdyne", "department": "Operations", "industry": "Technology"}, {"name": "Omar Khan", "city": "Yanacancha", "company": "Wonka Co", "department": "Operations", "industry": "Technology"}, {"name": "David Park", "city": "Yanacancha", "company": "Initech", "department": "Operations", "industry": "Hospitality"}, {"name": "Judy Reyes", "city": "Sheoganj", "company": "Globex", "department": "Operations", "industry": "Technology"}, {"name": "Priya Johnson", "city": "Saint-Constant", "company": "Umbrella", "department": "Operations", "industry": "Financial Services"}, {"name": "Alice Park", "city": "Sheoganj", "company": "Umbrella", "department": "Operations", "industry": "Hospitality"}, {"name": "Grace Diaz", "city": "Sheoganj", "company": "Stark Industries", "department": "Human Resources", "industry": "Financial Services"}, {"name": "Judy Lee", "city": "Yanacancha", "company": "Umbrella", "department": "Finance", "industry": "Financial Services"}, {"name": "Priya Petrov", "city": "Yanacancha", "company": "Soylent Corp", "department": "Human Resources", "industry": "Hospitality"}, {"name": "Grace Brandt", "city": "Saint-Constant", "company": "Cyberdyne", "department": "Human Resources", "industry": "Technology"}, {"name": "Ivan Brandt", "city": "Sheoganj", "company": "Vehement", "department": "Human Resources", "industry": "Financial Services"}, {"name": "Ivan Novak", "city": "Sheoganj", "company": "Stark Industries", "department": "Finance", "industry": "Telecommunications"}, {"name": "Judy Cruz", "city": "Saint-Constant", "company": "Soylent Corp", "department": "Operations", "industry": "Financial Services"}, {"name": "Eve Ali", "city": "Yanacancha", "company": "Globex", "department": "Operations", "industry": "Hospitality"}, {"name": "Ivan Khan", "city": "Sheoganj", "company": "Wonka Co", "department": "Finance", "industry": "Hospitality"}, {"name": "David Brandt", "city": "Yanacancha", "company": "Vehement", "department": "Finance", "industry": "Technology"}], "dirty_cols": ["name", "city", "company", "department", "industry", "col_x"], "clean_cols": ["name", "city", "company", "department", "industry"], "plan": {"dataset_summary": "42 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sheoganj": "Sheoganj", "Sheoanj": "Sheoganj", "SaintConstant": "Saint-Constant", "sait-constant": "Saint-Constant", "SHEOGANJ": "Sheoganj", "Saint-Coonstant": "Saint-Constant", "saint-constant": "Saint-Constant", "Sheogaj": "Sheoganj", "Yanaqancha": "Yanacancha", "yanacancha": "Yanacancha", "SAINT-CONSTANT": "Saint-Constant", "Shoeganj": "Sheoganj", "YANACANCHA": "Yanacancha", "Yinacancha": "Yanacancha", "Suint-Constant": "Saint-Constant"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PEOPLE": "Human Resources", "Ops": "Operations", "accounting": "Finance", "Fin": "Finance", "HR": "Human Resources", "People Ops": "Human Resources", "people ops": "Human Resources", "FIN": "Finance", "fin": "Finance", "finacne": "Finance", "PEOPLE OPS": "Human Resources", "Pepole": "Human Resources", "Accounting": "Finance", "ops": "Operations", "operations": "Operations", "ACCOUNTING": "Finance", "OPERATIONS": "Operations", "hr": "Human Resources", "Fitance": "Finance", "Hr": "Human Resources"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Banking": "Financial Services", "FINSERV": "Financial Services", "FinServ": "Financial Services", "It": "Technology", "hospitality": "Hospitality", "travel": "Hospitality", "hoslitality": "Hospitality", "Cimms": "Telecommunications", "Turism": "Hospitality", "technology": "Technology", "comms": "Telecommunications", "Comms": "Telecommunications", "Fintech": "Financial Services", "Finance": "Financial Services", "it": "Technology", "Software": "Technology", "COMMS": "Telecommunications", "TECRNOLOGY": "Technology", "technoloogy": "Technology", "HOSPITALITY": "Hospitality", "tech": "Technology", "financial services": "Financial Services", "TRAVEL": "Hospitality", "BANKING": "Financial Services", "fintech": "Financial Services", "tceh": "Technology", "Hotels & Travel": "Hospitality"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "Y", "amount": "$2,715.62", "state": "NEW JELSEY", "company": " Wonka Co", "status": " Open ", "extra": ""}, {"is_active": "No", "amount": "$865.77", "state": "NEW JERSBY", "company": "Globex   ", "status": "open", "extra": ""}, {"is_active": "1", "amount": "1.889,08", "state": "NEW JERSEY  ", "company": "Wonka Co", "status": "In Progress ", "extra": ""}, {"is_active": "TRUE", "amount": "$4,270.37", "state": "MT", "company": " Wonka Co", "status": " NEW", "extra": ""}, {"is_active": "Y", "amount": "2.612,38", "state": "  Iowa", "company": "Wonka Co", "status": "ongoing  ", "extra": ""}, {"is_active": "N", "amount": "$6,941.54", "state": "Mt", "company": "  Vehement", "status": "in-progress", "extra": ""}, {"is_active": "Y", "amount": "$4,889.91", "state": "mt ", "company": "Cyberdyne", "status": "Oen", "extra": ""}, {"is_active": "T", "amount": "$6,716.16", "state": "nj", "company": "Vehement", "status": "in progress", "extra": ""}, {"is_active": "FALSE", "amount": "-100", "state": "  mt", "company": "Cyberdyne", "status": "OPEN", "extra": ""}, {"is_active": "true", "amount": "1.816,68", "state": "iowa", "company": "Wonka Co", "status": "WIP", "extra": ""}, {"is_active": "Y", "amount": "$229.63", "state": "IA", "company": "Hooli", "status": "in-progress", "extra": ""}, {"is_active": "T", "amount": "$2,695.77", "state": "Utah  ", "company": "Initech", "status": "New", "extra": ""}, {"is_active": "", "amount": "", "state": "", "company": "", "status": "", "extra": ""}], "clean": [{"is_active": true, "amount": 2715.62, "state": "New Jersey", "company": "Wonka Co", "status": "Open"}, {"is_active": false, "amount": 865.77, "state": "New Jersey", "company": "Globex", "status": "Open"}, {"is_active": true, "amount": 1889.08, "state": "New Jersey", "company": "Wonka Co", "status": "In Progress"}, {"is_active": true, "amount": 4270.37, "state": "Montana", "company": "Wonka Co", "status": "Open"}, {"is_active": true, "amount": 2612.38, "state": "Iowa", "company": "Wonka Co", "status": "In Progress"}, {"is_active": false, "amount": 6941.54, "state": "Montana", "company": "Vehement", "status": "In Progress"}, {"is_active": true, "amount": 4889.91, "state": "Montana", "company": "Cyberdyne", "status": "Open"}, {"is_active": true, "amount": 6716.16, "state": "New Jersey", "company": "Vehement", "status": "In Progress"}, {"is_active": false, "amount": -100.0, "state": "Montana", "company": "Cyberdyne", "status": "Open"}, {"is_active": true, "amount": 1816.68, "state": "Iowa", "company": "Wonka Co", "status": "In Progress"}, {"is_active": true, "amount": 229.63, "state": "Iowa", "company": "Hooli", "status": "In Progress"}, {"is_active": true, "amount": 2695.77, "state": "Utah", "company": "Initech", "status": "Open"}], "dirty_cols": ["is_active", "amount", "state", "company", "status", "extra"], "clean_cols": ["is_active", "amount", "state", "company", "status"], "plan": {"dataset_summary": "13 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NEW JELSEY": "New Jersey", "NEW JERSBY": "New Jersey", "NEW JERSEY": "New Jersey", "MT": "Montana", "Mt": "Montana", "mt": "Montana", "nj": "New Jersey", "iowa": "Iowa", "IA": "Iowa"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"open": "Open", "NEW": "Open", "ongoing": "In Progress", "in-progress": "In Progress", "Oen": "Open", "in progress": "In Progress", "OPEN": "Open", "WIP": "In Progress", "New": "Open"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"department": " FINANCE ", "status": "paid", "email": "ohmip@corp.io", "notes2": ""}, {"department": "Mktg", "status": "paid", "email": " ainanh@corp.io", "notes2": ""}, {"department": "people ops ", "status": "refunded", "email": "OCMFEK@TEST.ORG  ", "notes2": ""}, {"department": "Accounting", "status": "LATE", "email": "JAJDD@CORP.IO", "notes2": ""}, {"department": " Growth ", "status": "Refunded   ", "email": "aglh@example.com", "notes2": ""}, {"department": "MARKETING   ", "status": "unpaid", "email": "fieafm@corp.io   ", "notes2": ""}, {"department": "Fin", "status": " due ", "email": "BKFLP@TEST.ORG", "notes2": ""}, {"department": "People", "status": "overdue", "email": "bfmokc@test.org  ", "notes2": ""}, {"department": "GROWTH", "status": "Ovrdue", "email": "FJBLBFD@TEST.ORG", "notes2": ""}, {"department": " ACCOUNTING ", "status": "paid", "email": "cahejg@example.com", "notes2": ""}, {"department": "people", "status": "Overdue ", "email": " acmg@corp.io ", "notes2": ""}, {"department": "HR", "status": "overue", "email": "OFBCA@EXAMPLE.COM", "notes2": ""}, {"department": "  Mkt", "status": "paid", "email": " fjaobm@example.com ", "notes2": ""}, {"department": " ACCOUNTING ", "status": "pending", "email": "  KICKMD@TEST.ORG", "notes2": ""}, {"department": "Growth", "status": "late ", "email": " iinaa@test.org ", "notes2": ""}, {"department": "finance", "status": "paid", "email": "eljl@corp.io", "notes2": ""}, {"department": "fin", "status": "refund  ", "email": "fahlahc@corp.io", "notes2": ""}, {"department": "fin", "status": "late", "email": "LEFKHFL@MAIL.COM", "notes2": ""}, {"department": "   accounting", "status": "  paid", "email": "  khkl@example.com", "notes2": ""}, {"department": "finance", "status": "paid  ", "email": "  FFLIPI@EXAMPLE.COM", "notes2": ""}, {"department": "growht", "status": "Paid", "email": "   MEONLK@EXAMPLE.COM", "notes2": ""}, {"department": " Finance", "status": " refund ", "email": " AAOLJA@MAIL.COM ", "notes2": ""}, {"department": " Finance ", "status": "PAID   ", "email": "neojil@corp.io", "notes2": ""}, {"department": "People Ops", "status": "  Paid", "email": "hcjkdh@corp.io", "notes2": ""}, {"department": "accounting", "status": "padi", "email": " kgbmbgd@test.org", "notes2": ""}, {"department": " FINANCE ", "status": "paid", "email": "ohmip@corp.io", "notes2": ""}, {"department": "Accounting", "status": "LATE", "email": "JAJDD@CORP.IO", "notes2": ""}, {"department": "", "status": "", "email": "", "notes2": ""}], "clean": [{"department": "Finance", "status": "Paid", "email": "ohmip@corp.io"}, {"department": "Marketing", "status": "Paid", "email": "ainanh@corp.io"}, {"department": "Human Resources", "status": "Refunded", "email": "ocmfek@test.org"}, {"department": "Finance", "status": "Overdue", "email": "jajdd@corp.io"}, {"department": "Marketing", "status": "Refunded", "email": "aglh@example.com"}, {"department": "Marketing", "status": "Pending", "email": "fieafm@corp.io"}, {"department": "Finance", "status": "Pending", "email": "bkflp@test.org"}, {"department": "Human Resources", "status": "Overdue", "email": "bfmokc@test.org"}, {"department": "Marketing", "status": "Overdue", "email": "fjblbfd@test.org"}, {"department": "Finance", "status": "Paid", "email": "cahejg@example.com"}, {"department": "Human Resources", "status": "Overdue", "email": "acmg@corp.io"}, {"department": "Human Resources", "status": "Overdue", "email": "ofbca@example.com"}, {"department": "Marketing", "status": "Paid", "email": "fjaobm@example.com"}, {"department": "Finance", "status": "Pending", "email": "kickmd@test.org"}, {"department": "Marketing", "status": "Overdue", "email": "iinaa@test.org"}, {"department": "Finance", "status": "Paid", "email": "eljl@corp.io"}, {"department": "Finance", "status": "Refunded", "email": "fahlahc@corp.io"}, {"department": "Finance", "status": "Overdue", "email": "lefkhfl@mail.com"}, {"department": "Finance", "status": "Paid", "email": "khkl@example.com"}, {"department": "Finance", "status": "Paid", "email": "fflipi@example.com"}, {"department": "Marketing", "status": "Paid", "email": "meonlk@example.com"}, {"department": "Finance", "status": "Refunded", "email": "aaolja@mail.com"}, {"department": "Finance", "status": "Paid", "email": "neojil@corp.io"}, {"department": "Human Resources", "status": "Paid", "email": "hcjkdh@corp.io"}, {"department": "Finance", "status": "Paid", "email": "kgbmbgd@test.org"}], "dirty_cols": ["department", "status", "email", "notes2"], "clean_cols": ["department", "status", "email"], "plan": {"dataset_summary": "28 rows × 4 columns. 3 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"FINANCE": "Finance", "Mktg": "Marketing", "people ops": "Human Resources", "Accounting": "Finance", "Growth": "Marketing", "MARKETING": "Marketing", "Fin": "Finance", "People": "Human Resources", "GROWTH": "Marketing", "ACCOUNTING": "Finance", "people": "Human Resources", "HR": "Human Resources", "Mkt": "Marketing", "finance": "Finance", "fin": "Finance", "accounting": "Finance", "growht": "Marketing", "People Ops": "Human Resources"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paid": "Paid", "refunded": "Refunded", "LATE": "Overdue", "unpaid": "Pending", "due": "Pending", "overdue": "Overdue", "Ovrdue": "Overdue", "overue": "Overdue", "pending": "Pending", "late": "Overdue", "refund": "Refunded", "PAID": "Paid", "padi": "Paid"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"rate": "71.0%", "phone": "537-499-0686", "company": "Vehement", "state": "CO", "name": "Karl Khan", "job_title": "product manager", "signup_date": "2023-10-11", "currency": " Mvdol ", "extra": ""}, {"rate": "82.3%", "phone": "(577)014-6207", "company": "Acme Inc", "state": "pr", "name": "Grace Johnson  ", "job_title": " admin ", "signup_date": "45249", "currency": "burundi franc", "extra": ""}, {"rate": "61.9%", "phone": "528.106.4332", "company": "Acme Inc", "state": "PR", "name": "   Karl Wong", "job_title": "Product  Manager", "signup_date": "5/4/2023", "currency": "BZD", "extra": ""}, {"rate": "22.1%", "phone": "5437384502", "company": "   Vehement", "state": "Colorado ", "name": "Ivan Smith", "job_title": "  product maanger", "signup_date": "14 Mar 2023", "currency": "Mvdl   ", "extra": ""}, {"rate": "20.3%", "phone": "5302911893", "company": "Acme Inc", "state": "pr", "name": "David  Reyes", "job_title": "admin", "signup_date": "8/19/2023", "currency": " CAD ", "extra": ""}, {"rate": "35.3%", "phone": "505.817.1429", "company": "Cyberdyne", "state": "Co", "name": "Sara  Cruz", "job_title": "ADMIN ASSISTANT", "signup_date": "2023-08-01", "currency": "  cad", "extra": ""}, {"rate": "83.7%", "phone": "510-747-1204", "company": "Acme Inc", "state": "Colorado", "name": "Alice Brandt", "job_title": "product maanger", "signup_date": "8/12/2023", "currency": "belize dollar", "extra": ""}, {"rate": "5.6%", "phone": "582.155.5844", "company": "Soylent  Corp", "state": "CO", "name": "Lena  Park", "job_title": "Prod Mgr", "signup_date": "2023-11-19", "currency": "CAADIAN  DOLLAR", "extra": ""}, {"rate": "53.8%", "phone": "556-583-3641", "company": " Umbrella", "state": "colorado", "name": "Omar Wong", "job_title": "Admin Asst", "signup_date": "2/10/2023", "currency": "cad", "extra": ""}, {"rate": "87.4%", "phone": "578-866-0015", "company": "   Wonka Co", "state": "   CO", "name": "Grace Wong  ", "job_title": "Adin Assistant ", "signup_date": "45079", "currency": "Burundi Franc", "extra": ""}, {"rate": "54.2%", "phone": "5634830065", "company": "Soylent Corp", "state": " puerta rico", "name": "Omar  Adams", "job_title": "Product Manager", "signup_date": "21 Jun 2023", "currency": "Belize Dollar", "extra": ""}, {"rate": "79.4%", "phone": "(582)102-6995", "company": "Hooli", "state": "Colorado", "name": "Mona Diaz", "job_title": "admin asst", "signup_date": "9/9/2023", "currency": "BOV", "extra": ""}, {"rate": "25.8%", "phone": "5860824819", "company": "Soylent Corp", "state": "Pherto Rico", "name": "Omar Petrov", "job_title": "pm", "signup_date": "11/26/2023", "currency": "Mvdol", "extra": ""}, {"rate": "6.1%", "phone": "5228964464", "company": "Initech", "state": "pr", "name": "Priya Khan  ", "job_title": "admin", "signup_date": "2023-10-20", "currency": "CAD", "extra": ""}, {"rate": "28.8%", "phone": "(506)441-7678", "company": "Globex  ", "state": "COLORADO", "name": "  Eve Brandt", "job_title": "PROD MGR", "signup_date": "2023-09-18", "currency": "BELIZE DOLLRA", "extra": ""}, {"rate": "33.9%", "phone": "(585)272-6075", "company": "Stark Industries", "state": "colorado", "name": "  Ivan Ali", "job_title": "admin asst", "signup_date": "27 Nov 2023", "currency": "Belize Dollar  ", "extra": ""}, {"rate": "12.6%", "phone": "558.961.1335", "company": "Hooli", "state": "colorado", "name": "Judy Brandt   ", "job_title": "administrative assistant", "signup_date": "2023-07-24", "currency": "Mvdol", "extra": ""}, {"rate": "83.9%", "phone": "565.967.0103", "company": "  Umbrella", "state": "Colorado", "name": "Alice  Reyes", "job_title": "Admin Asst", "signup_date": "28 Feb 2023", "currency": "BOV", "extra": ""}, {"rate": "60.8%", "phone": "(566)644-9208", "company": "Hooli", "state": "co", "name": "Carol Adams", "job_title": "ADMIN", "signup_date": "10/9/2023", "currency": "Belize Dollar", "extra": ""}, {"rate": "47.1%", "phone": "5186968471", "company": "Hooli", "state": "Pr   ", "name": "   Judy Lee", "job_title": "admin assst", "signup_date": "44930", "currency": "BIF", "extra": ""}, {"rate": "55.4%", "phone": "564.646.3518", "company": "Acme Inc", "state": "puerto  rico", "name": "Omar Reyes", "job_title": "Product Manager", "signup_date": "2/8/2023", "currency": "bzd", "extra": ""}, {"rate": "48.0%", "phone": "5917877333", "company": "Globex   ", "state": "PR", "name": "  Eve Park", "job_title": "  admin assistant", "signup_date": "2023-08-20", "currency": "bif", "extra": ""}, {"rate": "64.5%", "phone": "5819808050", "company": "  Cyberdyne", "state": "Colorado", "name": " Mona Petrov", "job_title": "Prod Mgr", "signup_date": "2023-09-21", "currency": "BOV", "extra": ""}, {"rate": "46.6%", "phone": "(528)181-7874", "company": "Cyberdyne   ", "state": "colorado", "name": "Mona Petrov", "job_title": "ADMINISTRATIVE ASSISTANT", "signup_date": "45153", "currency": "Burundi Franc", "extra": ""}, {"rate": "15.6%", "phone": "597.785.4810", "company": "Vehement", "state": "puerto rico", "name": "Priya  Park", "job_title": "Prtduct Manager", "signup_date": "45281", "currency": "Canadian Dlolar", "extra": ""}, {"rate": "41.0%", "phone": "(585)623-1312", "company": " Umbrella", "state": "Colorado", "name": "Lena Ali", "job_title": " Prod Manager", "signup_date": "20 May 2023", "currency": "BZD", "extra": ""}, {"rate": "89.0%", "phone": "512.618.0785", "company": "Acme Inc", "state": "purto rico", "name": "Sara  Novak", "job_title": "Admin Asst", "signup_date": "11/22/2023", "currency": "CAD", "extra": ""}, {"rate": "52.0%", "phone": "5037844439", "company": "Umbrella", "state": "Colorado", "name": "Grace Diaz", "job_title": "Admin", "signup_date": "1/7/2023", "currency": "bzd", "extra": ""}, {"rate": "58.0%", "phone": "(561)367-3483", "company": "Globex", "state": "PR", "name": "Omar Novak", "job_title": "   PM", "signup_date": "13 Jun 2023", "currency": "burundi franc", "extra": ""}, {"rate": "93.3%", "phone": "571-382-4065", "company": " Vehement ", "state": "Puerto Rico", "name": "Ivan Fischer   ", "job_title": "Admin Assistant", "signup_date": "7 Apr 2023", "currency": "BOV", "extra": ""}, {"rate": "20.0%", "phone": "(543)343-8579", "company": "Cyberdyne ", "state": "   Colorado", "name": "   Carol Ali", "job_title": "PRODUCT MANAGER", "signup_date": "1 Aug 2023", "currency": "BOV", "extra": ""}, {"rate": "29.8%", "phone": "569.244.0201", "company": " Globex", "state": "Puerto Rico ", "name": "Karl Khan", "job_title": "  prod mgr", "signup_date": "2/3/2023", "currency": "Belize  Dollbr", "extra": ""}, {"rate": "50.2%", "phone": "(535)975-9917", "company": "  Acme Inc", "state": "Puerto Rico", "name": "Alice Moore", "job_title": "product  manager", "signup_date": "17 Feb 2023", "currency": "   Canadian Dollar", "extra": ""}, {"rate": "73.7%", "phone": "564-309-1439", "company": "Vehement", "state": "PR", "name": "Karl Fischer", "job_title": "administraative assistant", "signup_date": "8 Nov 2023", "currency": " mvdol", "extra": ""}, {"rate": "", "phone": "", "company": "", "state": "", "name": "", "job_title": "", "signup_date": "", "currency": "", "extra": ""}], "clean": [{"rate": 0.71, "phone": "(537) 499-0686", "company": "Vehement", "state": "Colorado", "name": "Karl Khan", "job_title": "Product Manager", "signup_date": "2023-10-11", "currency": "BOV"}, {"rate": 0.823, "phone": "(577) 014-6207", "company": "Acme Inc", "state": "Puerto Rico", "name": "Grace Johnson", "job_title": "Administrative Assistant", "signup_date": "2023-11-19", "currency": "BIF"}, {"rate": 0.619, "phone": "(528) 106-4332", "company": "Acme Inc", "state": "Puerto Rico", "name": "Karl Wong", "job_title": "Product Manager", "signup_date": "2023-05-04", "currency": "BZD"}, {"rate": 0.221, "phone": "(543) 738-4502", "company": "Vehement", "state": "Colorado", "name": "Ivan Smith", "job_title": "Product Manager", "signup_date": "2023-03-14", "currency": "BOV"}, {"rate": 0.203, "phone": "(530) 291-1893", "company": "Acme Inc", "state": "Puerto Rico", "name": "David Reyes", "job_title": "Administrative Assistant", "signup_date": "2023-08-19", "currency": "CAD"}, {"rate": 0.353, "phone": "(505) 817-1429", "company": "Cyberdyne", "state": "Colorado", "name": "Sara Cruz", "job_title": "Administrative Assistant", "signup_date": "2023-08-01", "currency": "CAD"}, {"rate": 0.8370000000000001, "phone": "(510) 747-1204", "company": "Acme Inc", "state": "Colorado", "name": "Alice Brandt", "job_title": "Product Manager", "signup_date": "2023-08-12", "currency": "BZD"}, {"rate": 0.055999999999999994, "phone": "(582) 155-5844", "company": "Soylent Corp", "state": "Colorado", "name": "Lena Park", "job_title": "Product Manager", "signup_date": "2023-11-19", "currency": "CAD"}, {"rate": 0.5379999999999999, "phone": "(556) 583-3641", "company": "Umbrella", "state": "Colorado", "name": "Omar Wong", "job_title": "Administrative Assistant", "signup_date": "2023-02-10", "currency": "CAD"}, {"rate": 0.8740000000000001, "phone": "(578) 866-0015", "company": "Wonka Co", "state": "Colorado", "name": "Grace Wong", "job_title": "Administrative Assistant", "signup_date": "2023-06-02", "currency": "BIF"}, {"rate": 0.542, "phone": "(563) 483-0065", "company": "Soylent Corp", "state": "Puerto Rico", "name": "Omar Adams", "job_title": "Product Manager", "signup_date": "2023-06-21", "currency": "BZD"}, {"rate": 0.794, "phone": "(582) 102-6995", "company": "Hooli", "state": "Colorado", "name": "Mona Diaz", "job_title": "Administrative Assistant", "signup_date": "2023-09-09", "currency": "BOV"}, {"rate": 0.258, "phone": "(586) 082-4819", "company": "Soylent Corp", "state": "Puerto Rico", "name": "Omar Petrov", "job_title": "Product Manager", "signup_date": "2023-11-26", "currency": "BOV"}, {"rate": 0.061, "phone": "(522) 896-4464", "company": "Initech", "state": "Puerto Rico", "name": "Priya Khan", "job_title": "Administrative Assistant", "signup_date": "2023-10-20", "currency": "CAD"}, {"rate": 0.28800000000000003, "phone": "(506) 441-7678", "company": "Globex", "state": "Colorado", "name": "Eve Brandt", "job_title": "Product Manager", "signup_date": "2023-09-18", "currency": "BZD"}, {"rate": 0.33899999999999997, "phone": "(585) 272-6075", "company": "Stark Industries", "state": "Colorado", "name": "Ivan Ali", "job_title": "Administrative Assistant", "signup_date": "2023-11-27", "currency": "BZD"}, {"rate": 0.126, "phone": "(558) 961-1335", "company": "Hooli", "state": "Colorado", "name": "Judy Brandt", "job_title": "Administrative Assistant", "signup_date": "2023-07-24", "currency": "BOV"}, {"rate": 0.8390000000000001, "phone": "(565) 967-0103", "company": "Umbrella", "state": "Colorado", "name": "Alice Reyes", "job_title": "Administrative Assistant", "signup_date": "2023-02-28", "currency": "BOV"}, {"rate": 0.608, "phone": "(566) 644-9208", "company": "Hooli", "state": "Colorado", "name": "Carol Adams", "job_title": "Administrative Assistant", "signup_date": "2023-10-09", "currency": "BZD"}, {"rate": 0.47100000000000003, "phone": "(518) 696-8471", "company": "Hooli", "state": "Puerto Rico", "name": "Judy Lee", "job_title": "Administrative Assistant", "signup_date": "2023-01-04", "currency": "BIF"}, {"rate": 0.5539999999999999, "phone": "(564) 646-3518", "company": "Acme Inc", "state": "Puerto Rico", "name": "Omar Reyes", "job_title": "Product Manager", "signup_date": "2023-02-08", "currency": "BZD"}, {"rate": 0.48, "phone": "(591) 787-7333", "company": "Globex", "state": "Puerto Rico", "name": "Eve Park", "job_title": "Administrative Assistant", "signup_date": "2023-08-20", "currency": "BIF"}, {"rate": 0.645, "phone": "(581) 980-8050", "company": "Cyberdyne", "state": "Colorado", "name": "Mona Petrov", "job_title": "Product Manager", "signup_date": "2023-09-21", "currency": "BOV"}, {"rate": 0.466, "phone": "(528) 181-7874", "company": "Cyberdyne", "state": "Colorado", "name": "Mona Petrov", "job_title": "Administrative Assistant", "signup_date": "2023-08-15", "currency": "BIF"}, {"rate": 0.156, "phone": "(597) 785-4810", "company": "Vehement", "state": "Puerto Rico", "name": "Priya Park", "job_title": "Product Manager", "signup_date": "2023-12-21", "currency": "CAD"}, {"rate": 0.41, "phone": "(585) 623-1312", "company": "Umbrella", "state": "Colorado", "name": "Lena Ali", "job_title": "Product Manager", "signup_date": "2023-05-20", "currency": "BZD"}, {"rate": 0.89, "phone": "(512) 618-0785", "company": "Acme Inc", "state": "Puerto Rico", "name": "Sara Novak", "job_title": "Administrative Assistant", "signup_date": "2023-11-22", "currency": "CAD"}, {"rate": 0.52, "phone": "(503) 784-4439", "company": "Umbrella", "state": "Colorado", "name": "Grace Diaz", "job_title": "Administrative Assistant", "signup_date": "2023-01-07", "currency": "BZD"}, {"rate": 0.58, "phone": "(561) 367-3483", "company": "Globex", "state": "Puerto Rico", "name": "Omar Novak", "job_title": "Product Manager", "signup_date": "2023-06-13", "currency": "BIF"}, {"rate": 0.9329999999999999, "phone": "(571) 382-4065", "company": "Vehement", "state": "Puerto Rico", "name": "Ivan Fischer", "job_title": "Administrative Assistant", "signup_date": "2023-04-07", "currency": "BOV"}, {"rate": 0.2, "phone": "(543) 343-8579", "company": "Cyberdyne", "state": "Colorado", "name": "Carol Ali", "job_title": "Product Manager", "signup_date": "2023-08-01", "currency": "BOV"}, {"rate": 0.298, "phone": "(569) 244-0201", "company": "Globex", "state": "Puerto Rico", "name": "Karl Khan", "job_title": "Product Manager", "signup_date": "2023-02-03", "currency": "BZD"}, {"rate": 0.502, "phone": "(535) 975-9917", "company": "Acme Inc", "state": "Puerto Rico", "name": "Alice Moore", "job_title": "Product Manager", "signup_date": "2023-02-17", "currency": "CAD"}, {"rate": 0.737, "phone": "(564) 309-1439", "company": "Vehement", "state": "Puerto Rico", "name": "Karl Fischer", "job_title": "Administrative Assistant", "signup_date": "2023-11-08", "currency": "BOV"}], "dirty_cols": ["rate", "phone", "company", "state", "name", "job_title", "signup_date", "currency", "extra"], "clean_cols": ["rate", "phone", "company", "state", "name", "job_title", "signup_date", "currency"], "plan": {"dataset_summary": "35 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CO": "Colorado", "pr": "Puerto Rico", "PR": "Puerto Rico", "Co": "Colorado", "colorado": "Colorado", "puerta rico": "Puerto Rico", "Pherto Rico": "Puerto Rico", "COLORADO": "Colorado", "co": "Colorado", "Pr": "Puerto Rico", "puerto rico": "Puerto Rico", "purto rico": "Puerto Rico"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"product manager": "Product Manager", "admin": "Administrative Assistant", "product maanger": "Product Manager", "ADMIN ASSISTANT": "Administrative Assistant", "Prod Mgr": "Product Manager", "Admin Asst": "Administrative Assistant", "Adin Assistant": "Administrative Assistant", "admin asst": "Administrative Assistant", "pm": "Product Manager", "PROD MGR": "Product Manager", "administrative assistant": "Administrative Assistant", "ADMIN": "Administrative Assistant", "admin assst": "Administrative Assistant", "admin assistant": "Administrative Assistant", "ADMINISTRATIVE ASSISTANT": "Administrative Assistant", "Prtduct Manager": "Product Manager", "Prod Manager": "Product Manager", "Admin": "Administrative Assistant", "PM": "Product Manager", "Admin Assistant": "Administrative Assistant", "PRODUCT MANAGER": "Product Manager", "prod mgr": "Product Manager", "administraative assistant": "Administrative Assistant"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mvdol": "BOV", "burundi franc": "BIF", "Mvdl": "BOV", "cad": "CAD", "belize dollar": "BZD", "CAADIAN DOLLAR": "CAD", "Burundi Franc": "BIF", "Belize Dollar": "BZD", "BELIZE DOLLRA": "BZD", "bzd": "BZD", "bif": "BIF", "Canadian Dlolar": "CAD", "Belize Dollbr": "BZD", "Canadian Dollar": "CAD", "mvdol": "BOV"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"company": "Hooli", "city": "Bellmore", "phone": "526-717-4721", "is_active": "No", "state": "west virginia", "currency": "Fiji  Dollar", "col_x": ""}, {"company": "Initech   ", "city": "itaocara", "phone": "520.132.0398", "is_active": "FALSE", "state": "Massachusetts", "currency": "Falkland Islands Pound", "col_x": ""}, {"company": "Initech", "city": " Yicheng ", "phone": "5199600609", "is_active": "N", "state": "Ct", "currency": "Fkp", "col_x": ""}, {"company": "Wonka  Co", "city": "Depew", "phone": "(514)278-8432", "is_active": "Yes", "state": "Connecticut", "currency": "falkland islands pound", "col_x": ""}, {"company": " Hooli", "city": "Daanbantayan", "phone": "596.789.3088", "is_active": "0", "state": "Connecticut", "currency": "Falkland Islands Pound", "col_x": ""}, {"company": "Acme Inc", "city": "Itaocara", "phone": "(588)798-2631", "is_active": "Y", "state": "CT", "currency": "fjd", "col_x": ""}, {"company": "Globex", "city": "yihceng", "phone": "5644510810", "is_active": "No", "state": "MA  ", "currency": " Fiji Dollar", "col_x": ""}, {"company": "Soylent Corp", "city": "Itaocara", "phone": "(570)301-6168", "is_active": "Y", "state": "CT", "currency": "Falkland Isalnds  Pound", "col_x": ""}, {"company": "Acme Inc", "city": "DAANBANTAYAN", "phone": "533.654.1602", "is_active": "false", "state": "massachusetts", "currency": "fjd", "col_x": ""}, {"company": "Soylent  Corp", "city": "Bellmore", "phone": "(535)437-5599", "is_active": "false", "state": "  MA", "currency": "fiji dollar", "col_x": ""}, {"company": "Umbrella", "city": "Yicfeng   ", "phone": "5281805973", "is_active": "TRUE", "state": "massachusetts", "currency": "Fiji Dollar", "col_x": ""}, {"company": "Initech", "city": "Depew", "phone": "573.952.9109", "is_active": "TRUE", "state": "ma", "currency": "FJD", "col_x": ""}, {"company": "Acme  Inc", "city": "itaocara", "phone": "5250645323", "is_active": "No", "state": "massachusetts", "currency": "  Fiji Dollar", "col_x": ""}, {"company": "Soylent Corp", "city": "Daanbantayan ", "phone": "508-539-2744", "is_active": "FALSE", "state": "MASSACHUSETTS", "currency": "FALKLAND ISLANDS POUND", "col_x": ""}, {"company": "Stark Industries", "city": "Daanbantayan", "phone": "5075109477", "is_active": "TRUE", "state": "MA", "currency": "FJD", "col_x": ""}, {"company": "Stark Industries", "city": "DTPEW", "phone": "591.709.4590", "is_active": "TRUE", "state": "united states minor outlying islands", "currency": "FKP", "col_x": ""}, {"company": "Cyberdyne", "city": " Daanbantayan", "phone": "5983294597", "is_active": "Y", "state": "Um", "currency": "fjd", "col_x": ""}, {"company": "Globex", "city": "Depew", "phone": "568.410.8979", "is_active": "F", "state": "massachusetts", "currency": "Falkland Islands Pound", "col_x": ""}, {"company": "Acme Inc", "city": "Depew", "phone": "(545)787-3646", "is_active": "true", "state": "MA", "currency": "fkp", "col_x": ""}, {"company": "Soylent Corp", "city": "DEPEEW", "phone": "5058936433", "is_active": "N", "state": "WV", "currency": "Falkland Islands Pound", "col_x": ""}, {"company": "Initech", "city": "yicheng", "phone": "522.574.0373", "is_active": "true", "state": "west virignia   ", "currency": " fkp ", "col_x": ""}, {"company": "Globex  ", "city": "Daanbantayan ", "phone": "(534)598-7492", "is_active": "Y", "state": "CT", "currency": "Fiji Dollar", "col_x": ""}, {"company": "Initech", "city": "depew", "phone": "5139161491", "is_active": "T", "state": " connecticut ", "currency": " Fjd", "col_x": ""}, {"company": "  Hooli", "city": "   Itaocara", "phone": "(571)056-7429", "is_active": "true", "state": "west virginia", "currency": "FJD", "col_x": ""}, {"company": " Vehement ", "city": "Yicheng", "phone": "5341453175", "is_active": "0", "state": "WV", "currency": "Fiji Dollar", "col_x": ""}, {"company": "Acme Inc", "city": "Itaocara", "phone": "(588)798-2631", "is_active": "Y", "state": "CT", "currency": "fjd", "col_x": ""}, {"company": "Acme  Inc", "city": "itaocara", "phone": "5250645323", "is_active": "No", "state": "massachusetts", "currency": "  Fiji Dollar", "col_x": ""}], "clean": [{"company": "Hooli", "city": "Bellmore", "phone": "(526) 717-4721", "is_active": false, "state": "West Virginia", "currency": "FJD"}, {"company": "Initech", "city": "Itaocara", "phone": "(520) 132-0398", "is_active": false, "state": "Massachusetts", "currency": "FKP"}, {"company": "Initech", "city": "Yicheng", "phone": "(519) 960-0609", "is_active": false, "state": "Connecticut", "currency": "FKP"}, {"company": "Wonka Co", "city": "Depew", "phone": "(514) 278-8432", "is_active": true, "state": "Connecticut", "currency": "FKP"}, {"company": "Hooli", "city": "Daanbantayan", "phone": "(596) 789-3088", "is_active": false, "state": "Connecticut", "currency": "FKP"}, {"company": "Acme Inc", "city": "Itaocara", "phone": "(588) 798-2631", "is_active": true, "state": "Connecticut", "currency": "FJD"}, {"company": "Globex", "city": "Yicheng", "phone": "(564) 451-0810", "is_active": false, "state": "Massachusetts", "currency": "FJD"}, {"company": "Soylent Corp", "city": "Itaocara", "phone": "(570) 301-6168", "is_active": true, "state": "Connecticut", "currency": "FKP"}, {"company": "Acme Inc", "city": "Daanbantayan", "phone": "(533) 654-1602", "is_active": false, "state": "Massachusetts", "currency": "FJD"}, {"company": "Soylent Corp", "city": "Bellmore", "phone": "(535) 437-5599", "is_active": false, "state": "Massachusetts", "currency": "FJD"}, {"company": "Umbrella", "city": "Yicheng", "phone": "(528) 180-5973", "is_active": true, "state": "Massachusetts", "currency": "FJD"}, {"company": "Initech", "city": "Depew", "phone": "(573) 952-9109", "is_active": true, "state": "Massachusetts", "currency": "FJD"}, {"company": "Acme Inc", "city": "Itaocara", "phone": "(525) 064-5323", "is_active": false, "state": "Massachusetts", "currency": "FJD"}, {"company": "Soylent Corp", "city": "Daanbantayan", "phone": "(508) 539-2744", "is_active": false, "state": "Massachusetts", "currency": "FKP"}, {"company": "Stark Industries", "city": "Daanbantayan", "phone": "(507) 510-9477", "is_active": true, "state": "Massachusetts", "currency": "FJD"}, {"company": "Stark Industries", "city": "Depew", "phone": "(591) 709-4590", "is_active": true, "state": "United States Minor Outlying Islands", "currency": "FKP"}, {"company": "Cyberdyne", "city": "Daanbantayan", "phone": "(598) 329-4597", "is_active": true, "state": "United States Minor Outlying Islands", "currency": "FJD"}, {"company": "Globex", "city": "Depew", "phone": "(568) 410-8979", "is_active": false, "state": "Massachusetts", "currency": "FKP"}, {"company": "Acme Inc", "city": "Depew", "phone": "(545) 787-3646", "is_active": true, "state": "Massachusetts", "currency": "FKP"}, {"company": "Soylent Corp", "city": "Depew", "phone": "(505) 893-6433", "is_active": false, "state": "West Virginia", "currency": "FKP"}, {"company": "Initech", "city": "Yicheng", "phone": "(522) 574-0373", "is_active": true, "state": "West Virginia", "currency": "FKP"}, {"company": "Globex", "city": "Daanbantayan", "phone": "(534) 598-7492", "is_active": true, "state": "Connecticut", "currency": "FJD"}, {"company": "Initech", "city": "Depew", "phone": "(513) 916-1491", "is_active": true, "state": "Connecticut", "currency": "FJD"}, {"company": "Hooli", "city": "Itaocara", "phone": "(571) 056-7429", "is_active": true, "state": "West Virginia", "currency": "FJD"}, {"company": "Vehement", "city": "Yicheng", "phone": "(534) 145-3175", "is_active": false, "state": "West Virginia", "currency": "FJD"}], "dirty_cols": ["company", "city", "phone", "is_active", "state", "currency", "col_x"], "clean_cols": ["company", "city", "phone", "is_active", "state", "currency"], "plan": {"dataset_summary": "27 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"itaocara": "Itaocara", "yihceng": "Yicheng", "DAANBANTAYAN": "Daanbantayan", "Yicfeng": "Yicheng", "DTPEW": "Depew", "DEPEEW": "Depew", "yicheng": "Yicheng", "depew": "Depew"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"west virginia": "West Virginia", "Ct": "Connecticut", "CT": "Connecticut", "MA": "Massachusetts", "massachusetts": "Massachusetts", "ma": "Massachusetts", "MASSACHUSETTS": "Massachusetts", "united states minor outlying islands": "United States Minor Outlying Islands", "Um": "United States Minor Outlying Islands", "WV": "West Virginia", "west virignia": "West Virginia", "connecticut": "Connecticut"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fiji Dollar": "FJD", "Falkland Islands Pound": "FKP", "Fkp": "FKP", "falkland islands pound": "FKP", "fjd": "FJD", "Falkland Isalnds Pound": "FKP", "fiji dollar": "FJD", "FALKLAND ISLANDS POUND": "FKP", "fkp": "FKP", "Fjd": "FJD"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "Utah", "rate": "80.0%", "country": "Niue", "amount": "$213.42", "industry": "Ed", "col_x": ""}, {"state": "UT", "rate": "99.5%", "country": "UNITED STATES MINOR OUTLYING ISLANDS", "amount": "1.468,35", "industry": " Education ", "col_x": ""}, {"state": " UT ", "rate": "4.9%", "country": "NU", "amount": "3.072,78", "industry": "E-commerce", "col_x": ""}, {"state": "in", "rate": "21.4%", "country": "umi", "amount": "837,55", "industry": "retail", "col_x": ""}, {"state": " UT ", "rate": "4.7%", "country": "REPUBLIC OF HAITI  ", "amount": "4.172,22", "industry": "Education", "col_x": ""}, {"state": "Utah", "rate": "4.2%", "country": "HTI", "amount": "$5,279.87", "industry": "   edtech", "col_x": ""}, {"state": "Washington", "rate": "76.9%", "country": "umi", "amount": "$4,871.24", "industry": " E-commerce ", "col_x": ""}, {"state": "Washington", "rate": "3.9%", "country": "Um", "amount": "4.590,32", "industry": "retail & e-commerce", "col_x": ""}, {"state": "WA", "rate": "85.5%", "country": "UM", "amount": "1.255,11", "industry": "   Reoail & E-commerce", "col_x": ""}, {"state": "utah", "rate": "37.5%", "country": "Nu", "amount": "823,56", "industry": "edtech", "col_x": ""}, {"state": "IN", "rate": "89.7%", "country": "HAIPI", "amount": "$580.00", "industry": "   retail & e-commerce", "col_x": ""}, {"state": "wa", "rate": "73.7%", "country": "Haiti", "amount": "7.392,98", "industry": "education", "col_x": ""}, {"state": "Washington", "rate": "28.2%", "country": "umi", "amount": "2.180,85", "industry": "Ed", "col_x": ""}, {"state": "Kansas", "rate": "76.7%", "country": "umi", "amount": "$1,432.49", "industry": "Educagion", "col_x": ""}, {"state": "utah", "rate": "32.0%", "country": "Haiti", "amount": "8.999,08", "industry": " Retail", "col_x": ""}, {"state": "UT", "rate": "79.7%", "country": "um", "amount": "$4,989.00", "industry": "ed", "col_x": ""}, {"state": "washington", "rate": "30.1%", "country": "NIUE", "amount": "2.672,54", "industry": "EDTECH", "col_x": ""}, {"state": "UT", "rate": "79.7%", "country": "um", "amount": "$4,989.00", "industry": "ed", "col_x": ""}, {"state": " UT ", "rate": "4.7%", "country": "REPUBLIC OF HAITI  ", "amount": "4.172,22", "industry": "Education", "col_x": ""}, {"state": "", "rate": "", "country": "", "amount": "", "industry": "", "col_x": ""}], "clean": [{"state": "Utah", "rate": 0.8, "country": "Niue", "amount": 213.42, "industry": "Education"}, {"state": "Utah", "rate": 0.995, "country": "United States Minor Outlying Islands", "amount": 1468.35, "industry": "Education"}, {"state": "Utah", "rate": 0.049, "country": "Niue", "amount": 3072.78, "industry": "Retail"}, {"state": "Indiana", "rate": 0.214, "country": "United States Minor Outlying Islands", "amount": 837.55, "industry": "Retail"}, {"state": "Utah", "rate": 0.047, "country": "Haiti", "amount": 4172.22, "industry": "Education"}, {"state": "Utah", "rate": 0.042, "country": "Haiti", "amount": 5279.87, "industry": "Education"}, {"state": "Washington", "rate": 0.769, "country": "United States Minor Outlying Islands", "amount": 4871.24, "industry": "Retail"}, {"state": "Washington", "rate": 0.039, "country": "United States Minor Outlying Islands", "amount": 4590.32, "industry": "Retail"}, {"state": "Washington", "rate": 0.855, "country": "United States Minor Outlying Islands", "amount": 1255.11, "industry": "Retail"}, {"state": "Utah", "rate": 0.375, "country": "Niue", "amount": 823.56, "industry": "Education"}, {"state": "Indiana", "rate": 0.897, "country": "Haiti", "amount": 580.0, "industry": "Retail"}, {"state": "Washington", "rate": 0.737, "country": "Haiti", "amount": 7392.98, "industry": "Education"}, {"state": "Washington", "rate": 0.282, "country": "United States Minor Outlying Islands", "amount": 2180.85, "industry": "Education"}, {"state": "Kansas", "rate": 0.767, "country": "United States Minor Outlying Islands", "amount": 1432.49, "industry": "Education"}, {"state": "Utah", "rate": 0.32, "country": "Haiti", "amount": 8999.08, "industry": "Retail"}, {"state": "Utah", "rate": 0.797, "country": "United States Minor Outlying Islands", "amount": 4989.0, "industry": "Education"}, {"state": "Washington", "rate": 0.301, "country": "Niue", "amount": 2672.54, "industry": "Education"}], "dirty_cols": ["state", "rate", "country", "amount", "industry", "col_x"], "clean_cols": ["state", "rate", "country", "amount", "industry"], "plan": {"dataset_summary": "20 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"UT": "Utah", "in": "Indiana", "WA": "Washington", "utah": "Utah", "IN": "Indiana", "wa": "Washington", "washington": "Washington"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"UNITED STATES MINOR OUTLYING ISLANDS": "United States Minor Outlying Islands", "NU": "Niue", "umi": "United States Minor Outlying Islands", "REPUBLIC OF HAITI": "Haiti", "HTI": "Haiti", "Um": "United States Minor Outlying Islands", "UM": "United States Minor Outlying Islands", "Nu": "Niue", "HAIPI": "Haiti", "um": "United States Minor Outlying Islands", "NIUE": "Niue"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ed": "Education", "E-commerce": "Retail", "retail": "Retail", "edtech": "Education", "retail & e-commerce": "Retail", "Reoail & E-commerce": "Retail", "education": "Education", "Educagion": "Education", "ed": "Education", "EDTECH": "Education"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "55.5%", "signup_date": "8/17/2023", "industry": "  tech", "amount": "1.596,58", "state": "HI", "phone": "583-246-1491", "is_active": "F", "city": "tapejara"}, {"rate": "37.1%", "signup_date": "14 Mar 2023", "industry": "Medical", "amount": "$6,480.84", "state": "MI", "phone": "594.602.2094", "is_active": "Yes", "city": "Soyo"}, {"rate": "71.2%", "signup_date": "2023-02-18", "industry": "medical", "amount": "7.072,87", "state": " HI", "phone": "5726756042", "is_active": "0", "city": "prague"}, {"rate": "43.0%", "signup_date": "8/4/2023", "industry": "property", "amount": "$3,439.43", "state": "  Ny", "phone": "5399113895", "is_active": "0", "city": "pameungpeuk"}, {"rate": "1.4%", "signup_date": "2023-07-04", "industry": " Healthcare ", "amount": "-100", "state": " ny ", "phone": "(587)616-1536", "is_active": "true", "city": "Pameungpeuk"}, {"rate": "41.8%", "signup_date": "1/8/2023", "industry": "health", "amount": "8.917,53", "state": "  ny", "phone": "(539)776-7135", "is_active": "N", "city": "Tapejara"}, {"rate": "33.5%", "signup_date": "2023-02-26", "industry": "Software", "amount": "$6,351.62", "state": "michigan", "phone": "503-174-8620", "is_active": "Yes", "city": "Tapejara"}, {"rate": "93.9%", "signup_date": "45221", "industry": " healtchare ", "amount": "8.345,78", "state": "HI", "phone": "529.274.0388", "is_active": "true", "city": " Tapejara "}, {"rate": "46.4%", "signup_date": "12/15/2023", "industry": "Real Estate", "amount": "3.131,29", "state": "Michigan", "phone": "5641535388", "is_active": "F", "city": "pamengpeuk"}, {"rate": "36.6%", "signup_date": "45275", "industry": "tech", "amount": "$2,230.75", "state": "   mi", "phone": "(556)763-6110", "is_active": "Y", "city": "soco"}, {"rate": "9.7%", "signup_date": "2 Sep 2023", "industry": "RealEstate", "amount": "$1,517.13", "state": "ny", "phone": "5008913555", "is_active": "0", "city": "Syo"}, {"rate": "26.1%", "signup_date": "8/7/2023", "industry": "Software", "amount": "8.018,28", "state": "michigan", "phone": "509-859-8017", "is_active": "Y", "city": "Tapejara"}, {"rate": "83.7%", "signup_date": "45078", "industry": "Mmdical", "amount": "$7,261.42", "state": "michigan", "phone": "(505)252-7151", "is_active": "No", "city": "Praggue"}, {"rate": "54.8%", "signup_date": "2023-07-17", "industry": " HEALTH ", "amount": "--", "state": "new york", "phone": "581.624.8865", "is_active": "Yes", "city": "soyo"}, {"rate": "42.4%", "signup_date": "3 Oct 2023", "industry": "Health  Care", "amount": "--", "state": "   Vermont", "phone": "(558)656-0121", "is_active": "Yes", "city": "pameungpeuk"}, {"rate": "69.0%", "signup_date": "45184", "industry": "realestate", "amount": "1.415,86", "state": "Michigan", "phone": "565.996.6903", "is_active": "No", "city": "tapejara"}, {"rate": "22.7%", "signup_date": "21 Jul 2023", "industry": "RealEstate", "amount": "4.585,42", "state": "Hwaii", "phone": "500.343.9187", "is_active": "0", "city": "Tapejara"}, {"rate": "39.2%", "signup_date": "1/4/2023", "industry": "Real Estate", "amount": "$5,782.53", "state": "HI", "phone": "5062335737", "is_active": "T", "city": " tapejara "}, {"rate": "20.0%", "signup_date": "2023-01-26", "industry": "Health  Care", "amount": "$6,542.67", "state": "new york", "phone": "563.028.5371", "is_active": "N", "city": "Prague"}, {"rate": "41.4%", "signup_date": "3/11/2023", "industry": "Medical", "amount": "None", "state": "Hi", "phone": "566-774-7169", "is_active": "true", "city": "Tapejara "}, {"rate": "92.5%", "signup_date": "2023-03-14", "industry": "tech   ", "amount": "na", "state": "hi", "phone": "512.643.8280", "is_active": "true", "city": "tapejara"}, {"rate": "74.1%", "signup_date": "5/23/2023", "industry": "IT", "amount": "1.043,59", "state": "Vermont ", "phone": "576.817.6510", "is_active": "No", "city": "PAMEUNGPEUK"}, {"rate": "55.7%", "signup_date": "2023-07-23", "industry": "Medical", "amount": "-", "state": "  Micihgan", "phone": "5538647732", "is_active": "0", "city": "PRAGUE"}, {"rate": "22.0%", "signup_date": "2023-10-10", "industry": "Healthcare", "amount": "4.073,23", "state": "Ne wYork", "phone": "5050413639", "is_active": "Yes", "city": "Soyo"}, {"rate": "74.8%", "signup_date": "2023-05-01", "industry": " Technology ", "amount": "7.683,18", "state": "VT", "phone": "5137276130", "is_active": "Y", "city": "soyo"}, {"rate": "57.1%", "signup_date": "11/20/2023", "industry": "Medical", "amount": "3.073,27", "state": "NY", "phone": "569.253.6987", "is_active": "N", "city": "Soyo"}, {"rate": "27.4%", "signup_date": "16 Nov 2023", "industry": "Health Care", "amount": "3.200,10", "state": " hi", "phone": "501-383-2259", "is_active": "0", "city": "Tapejara"}, {"rate": "84.8%", "signup_date": "2/25/2023", "industry": "PropTech", "amount": "$134.52", "state": "New York", "phone": "510-212-3012", "is_active": "0", "city": "tapejara"}, {"rate": "55.0%", "signup_date": "12/27/2023", "industry": "Medical", "amount": "1.486,92", "state": "Vermont", "phone": "(589)125-1368", "is_active": "0", "city": " Soyo "}, {"rate": "15.4%", "signup_date": "8/14/2023", "industry": "health care", "amount": "6.965,64", "state": "New  York", "phone": "581.377.5269", "is_active": "0", "city": "Tapejara"}, {"rate": "91.4%", "signup_date": "11/23/2023", "industry": "real estate ", "amount": "163,88", "state": "hi", "phone": "(501)467-2857", "is_active": "F", "city": "prague"}, {"rate": "40.9%", "signup_date": "2023-02-17", "industry": "Healthcare ", "amount": "$1,906.14", "state": "HAAII", "phone": "(526)331-7348", "is_active": "TRUE", "city": "PAMEUNGPEUK"}, {"rate": "26.0%", "signup_date": "6 Mar 2023", "industry": "real estate", "amount": "1.855,21", "state": "ny", "phone": "5961017336", "is_active": "F", "city": "SOOY"}, {"rate": "64.3%", "signup_date": "20 Nov 2023", "industry": "IT", "amount": "5.160,05", "state": "Hawaii", "phone": "(533)921-5816", "is_active": "F", "city": "Prjgue"}, {"rate": "99.1%", "signup_date": "2/11/2023", "industry": "real etate", "amount": "na", "state": "ny", "phone": "5514124363", "is_active": "FALSE", "city": "pameungpeuk"}, {"rate": "3.5%", "signup_date": "1/16/2023", "industry": "Tech", "amount": "1.704,06", "state": "NY", "phone": "541-305-4677", "is_active": "TRUE", "city": " soyo "}, {"rate": "61.1%", "signup_date": "9/16/2023", "industry": "PROPTECH", "amount": "$5,263.99", "state": "michigan", "phone": "5339072893", "is_active": "false", "city": "   Tapeara"}, {"rate": "43.4%", "signup_date": "2023-11-16", "industry": "PropTech", "amount": "-", "state": "new york", "phone": "502.587.5248", "is_active": "1", "city": "pameungpeuk"}, {"rate": "43.4%", "signup_date": "2023-01-12", "industry": "tech", "amount": "7.071,21", "state": "VT", "phone": "588-624-6744", "is_active": "false", "city": "tapejara"}, {"rate": "80.5%", "signup_date": "2023-07-02", "industry": "PROPTECH", "amount": "965,18", "state": "Vermont ", "phone": "5080829247", "is_active": "Yes", "city": "Sqyo"}, {"rate": "72.2%", "signup_date": "24 Apr 2023", "industry": "Tehc", "amount": "218,06", "state": "mi", "phone": "5602263119", "is_active": "No", "city": "Tapejara"}, {"rate": "18.4%", "signup_date": "44932", "industry": "Health Cre", "amount": "3.322,56", "state": "ny", "phone": "532-426-4895", "is_active": "TRUE", "city": "pameungpeuk"}, {"rate": "18.5%", "signup_date": "45071", "industry": " health ", "amount": "3.089,89", "state": "Michigan", "phone": "5284466921", "is_active": "Yes", "city": "pameungpeuk"}, {"rate": "37.3%", "signup_date": "2023-07-03", "industry": " tech ", "amount": "$350.10", "state": "mihigan", "phone": "5853944490", "is_active": "TRUE", "city": " Soyo"}, {"rate": "96.5%", "signup_date": "10 Aug 2023", "industry": "Healthcare", "amount": "6.747,34", "state": "HAWAII", "phone": "5385495772", "is_active": "Yes", "city": "PAMEUNGPEKU"}, {"rate": "74.3%", "signup_date": "11/9/2023", "industry": "PropTech", "amount": "6.186,75", "state": "NY", "phone": "(541)891-7402", "is_active": "Yes", "city": "Soyo"}, {"rate": "58.5%", "signup_date": "11/8/2023", "industry": " tech ", "amount": "945,96", "state": "HI", "phone": "532-486-5926", "is_active": "T", "city": "Prague"}, {"rate": "17.3%", "signup_date": "45105", "industry": "Healthcare", "amount": "$3,405.92", "state": "michigan   ", "phone": "5244736426", "is_active": "1", "city": "soyo"}, {"rate": "14.7%", "signup_date": "2 Mar 2023", "industry": "SOFTWARE", "amount": "8.881,09", "state": "Michigan", "phone": "(529)412-3703", "is_active": "0", "city": "Soyo   "}, {"rate": "91.4%", "signup_date": "21 Feb 2023", "industry": "Tech", "amount": "8.280,79", "state": "   Michigan", "phone": "5986577213", "is_active": "N", "city": "Prague"}, {"rate": "64.3%", "signup_date": "20 Nov 2023", "industry": "IT", "amount": "5.160,05", "state": "Hawaii", "phone": "(533)921-5816", "is_active": "F", "city": "Prjgue"}, {"rate": "", "signup_date": "", "industry": "", "amount": "", "state": "", "phone": "", "is_active": "", "city": ""}], "clean": [{"rate": 0.555, "signup_date": "2023-08-17", "industry": "Technology", "amount": 1596.58, "state": "Hawaii", "phone": "(583) 246-1491", "is_active": false, "city": "Tapejara"}, {"rate": 0.371, "signup_date": "2023-03-14", "industry": "Healthcare", "amount": 6480.84, "state": "Michigan", "phone": "(594) 602-2094", "is_active": true, "city": "Soyo"}, {"rate": 0.7120000000000001, "signup_date": "2023-02-18", "industry": "Healthcare", "amount": 7072.87, "state": "Hawaii", "phone": "(572) 675-6042", "is_active": false, "city": "Prague"}, {"rate": 0.43, "signup_date": "2023-08-04", "industry": "Real Estate", "amount": 3439.43, "state": "New York", "phone": "(539) 911-3895", "is_active": false, "city": "Pameungpeuk"}, {"rate": 0.013999999999999999, "signup_date": "2023-07-04", "industry": "Healthcare", "amount": -100.0, "state": "New York", "phone": "(587) 616-1536", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.418, "signup_date": "2023-01-08", "industry": "Healthcare", "amount": 8917.53, "state": "New York", "phone": "(539) 776-7135", "is_active": false, "city": "Tapejara"}, {"rate": 0.335, "signup_date": "2023-02-26", "industry": "Technology", "amount": 6351.62, "state": "Michigan", "phone": "(503) 174-8620", "is_active": true, "city": "Tapejara"}, {"rate": 0.9390000000000001, "signup_date": "2023-10-22", "industry": "Healthcare", "amount": 8345.78, "state": "Hawaii", "phone": "(529) 274-0388", "is_active": true, "city": "Tapejara"}, {"rate": 0.46399999999999997, "signup_date": "2023-12-15", "industry": "Real Estate", "amount": 3131.29, "state": "Michigan", "phone": "(564) 153-5388", "is_active": false, "city": "Pameungpeuk"}, {"rate": 0.366, "signup_date": "2023-12-15", "industry": "Technology", "amount": 2230.75, "state": "Michigan", "phone": "(556) 763-6110", "is_active": true, "city": "Soyo"}, {"rate": 0.09699999999999999, "signup_date": "2023-09-02", "industry": "Real Estate", "amount": 1517.13, "state": "New York", "phone": "(500) 891-3555", "is_active": false, "city": "Soyo"}, {"rate": 0.261, "signup_date": "2023-08-07", "industry": "Technology", "amount": 8018.28, "state": "Michigan", "phone": "(509) 859-8017", "is_active": true, "city": "Tapejara"}, {"rate": 0.8370000000000001, "signup_date": "2023-06-01", "industry": "Healthcare", "amount": 7261.42, "state": "Michigan", "phone": "(505) 252-7151", "is_active": false, "city": "Prague"}, {"rate": 0.5479999999999999, "signup_date": "2023-07-17", "industry": "Healthcare", "amount": NaN, "state": "New York", "phone": "(581) 624-8865", "is_active": true, "city": "Soyo"}, {"rate": 0.424, "signup_date": "2023-10-03", "industry": "Healthcare", "amount": NaN, "state": "Vermont", "phone": "(558) 656-0121", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.69, "signup_date": "2023-09-15", "industry": "Real Estate", "amount": 1415.86, "state": "Michigan", "phone": "(565) 996-6903", "is_active": false, "city": "Tapejara"}, {"rate": 0.22699999999999998, "signup_date": "2023-07-21", "industry": "Real Estate", "amount": 4585.42, "state": "Hawaii", "phone": "(500) 343-9187", "is_active": false, "city": "Tapejara"}, {"rate": 0.392, "signup_date": "2023-01-04", "industry": "Real Estate", "amount": 5782.53, "state": "Hawaii", "phone": "(506) 233-5737", "is_active": true, "city": "Tapejara"}, {"rate": 0.2, "signup_date": "2023-01-26", "industry": "Healthcare", "amount": 6542.67, "state": "New York", "phone": "(563) 028-5371", "is_active": false, "city": "Prague"}, {"rate": 0.414, "signup_date": "2023-03-11", "industry": "Healthcare", "amount": NaN, "state": "Hawaii", "phone": "(566) 774-7169", "is_active": true, "city": "Tapejara"}, {"rate": 0.925, "signup_date": "2023-03-14", "industry": "Technology", "amount": NaN, "state": "Hawaii", "phone": "(512) 643-8280", "is_active": true, "city": "Tapejara"}, {"rate": 0.741, "signup_date": "2023-05-23", "industry": "Technology", "amount": 1043.59, "state": "Vermont", "phone": "(576) 817-6510", "is_active": false, "city": "Pameungpeuk"}, {"rate": 0.557, "signup_date": "2023-07-23", "industry": "Healthcare", "amount": NaN, "state": "Michigan", "phone": "(553) 864-7732", "is_active": false, "city": "Prague"}, {"rate": 0.22, "signup_date": "2023-10-10", "industry": "Healthcare", "amount": 4073.23, "state": "New York", "phone": "(505) 041-3639", "is_active": true, "city": "Soyo"}, {"rate": 0.748, "signup_date": "2023-05-01", "industry": "Technology", "amount": 7683.18, "state": "Vermont", "phone": "(513) 727-6130", "is_active": true, "city": "Soyo"}, {"rate": 0.5710000000000001, "signup_date": "2023-11-20", "industry": "Healthcare", "amount": 3073.27, "state": "New York", "phone": "(569) 253-6987", "is_active": false, "city": "Soyo"}, {"rate": 0.27399999999999997, "signup_date": "2023-11-16", "industry": "Healthcare", "amount": 3200.1, "state": "Hawaii", "phone": "(501) 383-2259", "is_active": false, "city": "Tapejara"}, {"rate": 0.848, "signup_date": "2023-02-25", "industry": "Real Estate", "amount": 134.52, "state": "New York", "phone": "(510) 212-3012", "is_active": false, "city": "Tapejara"}, {"rate": 0.55, "signup_date": "2023-12-27", "industry": "Healthcare", "amount": 1486.92, "state": "Vermont", "phone": "(589) 125-1368", "is_active": false, "city": "Soyo"}, {"rate": 0.154, "signup_date": "2023-08-14", "industry": "Healthcare", "amount": 6965.64, "state": "New York", "phone": "(581) 377-5269", "is_active": false, "city": "Tapejara"}, {"rate": 0.914, "signup_date": "2023-11-23", "industry": "Real Estate", "amount": 163.88, "state": "Hawaii", "phone": "(501) 467-2857", "is_active": false, "city": "Prague"}, {"rate": 0.409, "signup_date": "2023-02-17", "industry": "Healthcare", "amount": 1906.14, "state": "Hawaii", "phone": "(526) 331-7348", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.26, "signup_date": "2023-03-06", "industry": "Real Estate", "amount": 1855.21, "state": "New York", "phone": "(596) 101-7336", "is_active": false, "city": "Soyo"}, {"rate": 0.643, "signup_date": "2023-11-20", "industry": "Technology", "amount": 5160.05, "state": "Hawaii", "phone": "(533) 921-5816", "is_active": false, "city": "Prague"}, {"rate": 0.991, "signup_date": "2023-02-11", "industry": "Real Estate", "amount": NaN, "state": "New York", "phone": "(551) 412-4363", "is_active": false, "city": "Pameungpeuk"}, {"rate": 0.035, "signup_date": "2023-01-16", "industry": "Technology", "amount": 1704.06, "state": "New York", "phone": "(541) 305-4677", "is_active": true, "city": "Soyo"}, {"rate": 0.611, "signup_date": "2023-09-16", "industry": "Real Estate", "amount": 5263.99, "state": "Michigan", "phone": "(533) 907-2893", "is_active": false, "city": "Tapejara"}, {"rate": 0.434, "signup_date": "2023-11-16", "industry": "Real Estate", "amount": NaN, "state": "New York", "phone": "(502) 587-5248", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.434, "signup_date": "2023-01-12", "industry": "Technology", "amount": 7071.21, "state": "Vermont", "phone": "(588) 624-6744", "is_active": false, "city": "Tapejara"}, {"rate": 0.805, "signup_date": "2023-07-02", "industry": "Real Estate", "amount": 965.18, "state": "Vermont", "phone": "(508) 082-9247", "is_active": true, "city": "Soyo"}, {"rate": 0.722, "signup_date": "2023-04-24", "industry": "Technology", "amount": 218.06, "state": "Michigan", "phone": "(560) 226-3119", "is_active": false, "city": "Tapejara"}, {"rate": 0.184, "signup_date": "2023-01-06", "industry": "Healthcare", "amount": 3322.56, "state": "New York", "phone": "(532) 426-4895", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.185, "signup_date": "2023-05-25", "industry": "Healthcare", "amount": 3089.89, "state": "Michigan", "phone": "(528) 446-6921", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.373, "signup_date": "2023-07-03", "industry": "Technology", "amount": 350.1, "state": "Michigan", "phone": "(585) 394-4490", "is_active": true, "city": "Soyo"}, {"rate": 0.965, "signup_date": "2023-08-10", "industry": "Healthcare", "amount": 6747.34, "state": "Hawaii", "phone": "(538) 549-5772", "is_active": true, "city": "Pameungpeuk"}, {"rate": 0.743, "signup_date": "2023-11-09", "industry": "Real Estate", "amount": 6186.75, "state": "New York", "phone": "(541) 891-7402", "is_active": true, "city": "Soyo"}, {"rate": 0.585, "signup_date": "2023-11-08", "industry": "Technology", "amount": 945.96, "state": "Hawaii", "phone": "(532) 486-5926", "is_active": true, "city": "Prague"}, {"rate": 0.17300000000000001, "signup_date": "2023-06-28", "industry": "Healthcare", "amount": 3405.92, "state": "Michigan", "phone": "(524) 473-6426", "is_active": true, "city": "Soyo"}, {"rate": 0.147, "signup_date": "2023-03-02", "industry": "Technology", "amount": 8881.09, "state": "Michigan", "phone": "(529) 412-3703", "is_active": false, "city": "Soyo"}, {"rate": 0.914, "signup_date": "2023-02-21", "industry": "Technology", "amount": 8280.79, "state": "Michigan", "phone": "(598) 657-7213", "is_active": false, "city": "Prague"}], "dirty_cols": ["rate", "signup_date", "industry", "amount", "state", "phone", "is_active", "city"], "clean_cols": ["rate", "signup_date", "industry", "amount", "state", "phone", "is_active", "city"], "plan": {"dataset_summary": "52 rows × 8 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"tech": "Technology", "Medical": "Healthcare", "medical": "Healthcare", "property": "Real Estate", "health": "Healthcare", "Software": "Technology", "healtchare": "Healthcare", "RealEstate": "Real Estate", "Mmdical": "Healthcare", "HEALTH": "Healthcare", "Health Care": "Healthcare", "realestate": "Real Estate", "Health  Care": "Healthcare", "IT": "Technology", "PropTech": "Real Estate", "health care": "Healthcare", "real estate": "Real Estate", "real etate": "Real Estate", "Tech": "Technology", "PROPTECH": "Real Estate", "Tehc": "Technology", "Health Cre": "Healthcare", "SOFTWARE": "Technology"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HI": "Hawaii", "MI": "Michigan", "Ny": "New York", "ny": "New York", "michigan": "Michigan", "mi": "Michigan", "new york": "New York", "Hwaii": "Hawaii", "Hi": "Hawaii", "hi": "Hawaii", "Micihgan": "Michigan", "Ne wYork": "New York", "VT": "Vermont", "NY": "New York", "HAAII": "Hawaii", "mihigan": "Michigan", "HAWAII": "Hawaii"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"tapejara": "Tapejara", "prague": "Prague", "pameungpeuk": "Pameungpeuk", "pamengpeuk": "Pameungpeuk", "soco": "Soyo", "Syo": "Soyo", "Praggue": "Prague", "soyo": "Soyo", "PAMEUNGPEUK": "Pameungpeuk", "PRAGUE": "Prague", "SOOY": "Soyo", "Prjgue": "Prague", "Tapeara": "Tapejara", "Sqyo": "Soyo", "PAMEUNGPEKU": "Pameungpeuk"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"company": "  Hooli", "country": "United States Mino Outlying  Islands", "department": "Marketing", "city": "   Chaiyaphum", "rate": "3.6%", "signup_date": "45165", "name": " Heidi Petrov", "notes2": ""}, {"company": "Wonka Co", "country": "Kingdom of Morocco", "department": "R&D", "city": " pekin ", "rate": "29.0%", "signup_date": "5/23/2023", "name": "  Grace Adams", "notes2": ""}, {"company": "  Wonka Co", "country": "ZM", "department": "Engineering", "city": "Nsoatre", "rate": "74.4%", "signup_date": "2023-02-02", "name": "Karl Petrov", "notes2": ""}, {"company": "Umbrella   ", "country": " KN ", "department": "   engineering", "city": "Sundbyberg", "rate": "51.3%", "signup_date": "45280", "name": "Bob Diaz", "notes2": ""}, {"company": "Acme Inc   ", "country": "zm", "department": "Customer Support", "city": "SUNDBYBERG", "rate": "31.9%", "signup_date": "45175", "name": "Priya Moore", "notes2": ""}, {"company": "Stark Industries", "country": "um", "department": "eng.", "city": "Nsatre", "rate": "13.6%", "signup_date": "2023-07-14", "name": "Heidi Smith   ", "notes2": ""}, {"company": "   Vehement", "country": "MOROCCO", "department": "Mkt", "city": "Chaiyaphum", "rate": "44.6%", "signup_date": "8 Sep 2023", "name": "Grace Khan", "notes2": ""}, {"company": "  Cyberdyne", "country": "zmb", "department": "Customer Support", "city": "Sundbyberg", "rate": "89.8%", "signup_date": "2 Apr 2023", "name": "   David Adams", "notes2": ""}, {"company": "   Wonka Co", "country": "MAR   ", "department": "Prod", "city": "Nsoatre", "rate": "11.0%", "signup_date": "44928", "name": "Omar  Lee", "notes2": ""}, {"company": " Stark Industries", "country": "saint  kitts and nevis", "department": "Mktg", "city": "   CHAIYAPHUM", "rate": "82.2%", "signup_date": "14 Jan 2023", "name": "Grace Khan", "notes2": ""}, {"company": "  Stark Industries", "country": "United States Minor Outlying Islands", "department": "cust  support", "city": "  nsoatre", "rate": "70.0%", "signup_date": "2023-02-26", "name": "Eve  Johnson", "notes2": ""}, {"company": " Globex ", "country": "Saint Kitts and Nevis", "department": "growth", "city": "Sundbyberg", "rate": "48.6%", "signup_date": "2023-07-25", "name": "Ivan Khan", "notes2": ""}, {"company": "  Hooli", "country": "Kingdom Of Morocco", "department": "prod", "city": "sundbyberg", "rate": "18.6%", "signup_date": "45172", "name": "  Bob Reyes", "notes2": ""}, {"company": "Initech", "country": "um", "department": "prod", "city": "Nsoatre", "rate": "42.4%", "signup_date": "2023-11-09", "name": "Judy Khan", "notes2": ""}, {"company": "Cyberdyne", "country": "zmb", "department": "Eng", "city": "  Chaiyaphum", "rate": "94.6%", "signup_date": "14 Aug 2023", "name": "Heidi  Adams", "notes2": ""}, {"company": "Stark Industries", "country": "UM", "department": "Eng.", "city": "   Chaiyaphum", "rate": "28.2%", "signup_date": "2023-06-03", "name": "Judy Reyes   ", "notes2": ""}, {"company": " Hooli", "country": "Republic Of Zambia", "department": "pm   ", "city": "sundbyberg   ", "rate": "53.5%", "signup_date": "45105", "name": "Mona Johnson  ", "notes2": ""}, {"company": "Stark Industries", "country": "umi", "department": "Dev", "city": "Al Khurmah", "rate": "96.2%", "signup_date": "2023-05-16", "name": "Heidi Cruz", "notes2": ""}, {"company": "Globex", "country": "um", "department": "Makreting", "city": "al khurmah", "rate": "99.4%", "signup_date": "24 Mar 2023", "name": "Priya Novak", "notes2": ""}, {"company": "Globex", "country": "ma", "department": "   DEV", "city": "Al Khurmah", "rate": "87.4%", "signup_date": "2023-04-17", "name": "Karl Reyes", "notes2": ""}, {"company": " Vehement ", "country": "UMI   ", "department": " SUPPORT ", "city": "peekin", "rate": "1.5%", "signup_date": "2023-11-03", "name": "  David Reyes", "notes2": ""}, {"company": "Umbrella   ", "country": "saint kitts and nevis", "department": "dev", "city": "CHAIYAPHUM", "rate": "12.4%", "signup_date": "11/23/2023", "name": "Judy Diaz ", "notes2": ""}, {"company": "Umbrella", "country": "KNA", "department": "MARKETING", "city": "nsoate", "rate": "42.6%", "signup_date": "45151", "name": "Lena Petrov", "notes2": ""}, {"company": "Stark Industries", "country": "UM", "department": "MARKETIGN", "city": "nsoatre", "rate": "33.3%", "signup_date": "2023-08-04", "name": "Grace  Ali", "notes2": ""}, {"company": "Cyberdyne", "country": "United States Minor Outlying Islands", "department": "Marketing", "city": " chaiyaphum ", "rate": "24.8%", "signup_date": "2023-02-11", "name": "Bob Brandt  ", "notes2": ""}, {"company": "  Stark Industries", "country": "Zavbia", "department": "Growth", "city": " Sundbyberg", "rate": "34.4%", "signup_date": "2023-06-28", "name": "Ivan Wong", "notes2": ""}, {"company": "Cyberdyne", "country": "Saint Kitts And Nevis", "department": "R&D", "city": "Pekin", "rate": "45.3%", "signup_date": "16 Nov 2023", "name": "Priya Brandt", "notes2": ""}, {"company": " Acme Inc", "country": "  Kn", "department": "Eng.", "city": "CHAIYAPHUM", "rate": "29.7%", "signup_date": "11/2/2023", "name": "Mona  Cruz", "notes2": ""}, {"company": "Acme Inc", "country": "  kna", "department": "Prod", "city": "Nsoatre", "rate": "13.7%", "signup_date": "2/26/2023", "name": "Heidi Moore", "notes2": ""}, {"company": "Initech", "country": "ZM", "department": "Mkt", "city": "chaiyaphum", "rate": "81.6%", "signup_date": "7/16/2023", "name": "Omar  Lee", "notes2": ""}, {"company": "  Vehement", "country": "ZM  ", "department": "cs", "city": "Al Khurmah", "rate": "5.3%", "signup_date": "45184", "name": "Bob Moore  ", "notes2": ""}, {"company": "Wonka Co", "country": "ZMB", "department": " Product", "city": "AL KHURMAH", "rate": "94.3%", "signup_date": "11/1/2023", "name": "Eve Brandt", "notes2": ""}, {"company": "   Vehement", "country": "KNA  ", "department": "  cust support", "city": "NSOATRE", "rate": "11.3%", "signup_date": "45104", "name": "Alice Adams  ", "notes2": ""}, {"company": "Soylent Corp", "country": "United States Minor Outlying Islands", "department": "product", "city": "   Sundbyberg", "rate": "31.6%", "signup_date": "45246", "name": "Frank Brandt", "notes2": ""}, {"company": "  Hooli", "country": "United States Mino Outlying  Islands", "department": "Marketing", "city": "   Chaiyaphum", "rate": "3.6%", "signup_date": "45165", "name": " Heidi Petrov", "notes2": ""}, {"company": "Stark Industries", "country": "UM", "department": "Eng.", "city": "   Chaiyaphum", "rate": "28.2%", "signup_date": "2023-06-03", "name": "Judy Reyes   ", "notes2": ""}], "clean": [{"company": "Hooli", "country": "United States Minor Outlying Islands", "department": "Marketing", "city": "Chaiyaphum", "rate": 0.036000000000000004, "signup_date": "2023-08-27", "name": "Heidi Petrov"}, {"company": "Wonka Co", "country": "Morocco", "department": "Engineering", "city": "Pekin", "rate": 0.29, "signup_date": "2023-05-23", "name": "Grace Adams"}, {"company": "Wonka Co", "country": "Zambia", "department": "Engineering", "city": "Nsoatre", "rate": 0.7440000000000001, "signup_date": "2023-02-02", "name": "Karl Petrov"}, {"company": "Umbrella", "country": "Saint Kitts and Nevis", "department": "Engineering", "city": "Sundbyberg", "rate": 0.513, "signup_date": "2023-12-20", "name": "Bob Diaz"}, {"company": "Acme Inc", "country": "Zambia", "department": "Customer Support", "city": "Sundbyberg", "rate": 0.319, "signup_date": "2023-09-06", "name": "Priya Moore"}, {"company": "Stark Industries", "country": "United States Minor Outlying Islands", "department": "Engineering", "city": "Nsoatre", "rate": 0.136, "signup_date": "2023-07-14", "name": "Heidi Smith"}, {"company": "Vehement", "country": "Morocco", "department": "Marketing", "city": "Chaiyaphum", "rate": 0.446, "signup_date": "2023-09-08", "name": "Grace Khan"}, {"company": "Cyberdyne", "country": "Zambia", "department": "Customer Support", "city": "Sundbyberg", "rate": 0.898, "signup_date": "2023-04-02", "name": "David Adams"}, {"company": "Wonka Co", "country": "Morocco", "department": "Product", "city": "Nsoatre", "rate": 0.11, "signup_date": "2023-01-02", "name": "Omar Lee"}, {"company": "Stark Industries", "country": "Saint Kitts and Nevis", "department": "Marketing", "city": "Chaiyaphum", "rate": 0.8220000000000001, "signup_date": "2023-01-14", "name": "Grace Khan"}, {"company": "Stark Industries", "country": "United States Minor Outlying Islands", "department": "Customer Support", "city": "Nsoatre", "rate": 0.7, "signup_date": "2023-02-26", "name": "Eve Johnson"}, {"company": "Globex", "country": "Saint Kitts and Nevis", "department": "Marketing", "city": "Sundbyberg", "rate": 0.486, "signup_date": "2023-07-25", "name": "Ivan Khan"}, {"company": "Hooli", "country": "Morocco", "department": "Product", "city": "Sundbyberg", "rate": 0.18600000000000003, "signup_date": "2023-09-03", "name": "Bob Reyes"}, {"company": "Initech", "country": "United States Minor Outlying Islands", "department": "Product", "city": "Nsoatre", "rate": 0.424, "signup_date": "2023-11-09", "name": "Judy Khan"}, {"company": "Cyberdyne", "country": "Zambia", "department": "Engineering", "city": "Chaiyaphum", "rate": 0.946, "signup_date": "2023-08-14", "name": "Heidi Adams"}, {"company": "Stark Industries", "country": "United States Minor Outlying Islands", "department": "Engineering", "city": "Chaiyaphum", "rate": 0.282, "signup_date": "2023-06-03", "name": "Judy Reyes"}, {"company": "Hooli", "country": "Zambia", "department": "Product", "city": "Sundbyberg", "rate": 0.535, "signup_date": "2023-06-28", "name": "Mona Johnson"}, {"company": "Stark Industries", "country": "United States Minor Outlying Islands", "department": "Engineering", "city": "Al Khurmah", "rate": 0.9620000000000001, "signup_date": "2023-05-16", "name": "Heidi Cruz"}, {"company": "Globex", "country": "United States Minor Outlying Islands", "department": "Marketing", "city": "Al Khurmah", "rate": 0.9940000000000001, "signup_date": "2023-03-24", "name": "Priya Novak"}, {"company": "Globex", "country": "Morocco", "department": "Engineering", "city": "Al Khurmah", "rate": 0.8740000000000001, "signup_date": "2023-04-17", "name": "Karl Reyes"}, {"company": "Vehement", "country": "United States Minor Outlying Islands", "department": "Customer Support", "city": "Pekin", "rate": 0.015, "signup_date": "2023-11-03", "name": "David Reyes"}, {"company": "Umbrella", "country": "Saint Kitts and Nevis", "department": "Engineering", "city": "Chaiyaphum", "rate": 0.124, "signup_date": "2023-11-23", "name": "Judy Diaz"}, {"company": "Umbrella", "country": "Saint Kitts and Nevis", "department": "Marketing", "city": "Nsoatre", "rate": 0.426, "signup_date": "2023-08-13", "name": "Lena Petrov"}, {"company": "Stark Industries", "country": "United States Minor Outlying Islands", "department": "Marketing", "city": "Nsoatre", "rate": 0.33299999999999996, "signup_date": "2023-08-04", "name": "Grace Ali"}, {"company": "Cyberdyne", "country": "United States Minor Outlying Islands", "department": "Marketing", "city": "Chaiyaphum", "rate": 0.248, "signup_date": "2023-02-11", "name": "Bob Brandt"}, {"company": "Stark Industries", "country": "Zambia", "department": "Marketing", "city": "Sundbyberg", "rate": 0.344, "signup_date": "2023-06-28", "name": "Ivan Wong"}, {"company": "Cyberdyne", "country": "Saint Kitts and Nevis", "department": "Engineering", "city": "Pekin", "rate": 0.45299999999999996, "signup_date": "2023-11-16", "name": "Priya Brandt"}, {"company": "Acme Inc", "country": "Saint Kitts and Nevis", "department": "Engineering", "city": "Chaiyaphum", "rate": 0.297, "signup_date": "2023-11-02", "name": "Mona Cruz"}, {"company": "Acme Inc", "country": "Saint Kitts and Nevis", "department": "Product", "city": "Nsoatre", "rate": 0.13699999999999998, "signup_date": "2023-02-26", "name": "Heidi Moore"}, {"company": "Initech", "country": "Zambia", "department": "Marketing", "city": "Chaiyaphum", "rate": 0.816, "signup_date": "2023-07-16", "name": "Omar Lee"}, {"company": "Vehement", "country": "Zambia", "department": "Customer Support", "city": "Al Khurmah", "rate": 0.053, "signup_date": "2023-09-15", "name": "Bob Moore"}, {"company": "Wonka Co", "country": "Zambia", "department": "Product", "city": "Al Khurmah", "rate": 0.943, "signup_date": "2023-11-01", "name": "Eve Brandt"}, {"company": "Vehement", "country": "Saint Kitts and Nevis", "department": "Customer Support", "city": "Nsoatre", "rate": 0.113, "signup_date": "2023-06-27", "name": "Alice Adams"}, {"company": "Soylent Corp", "country": "United States Minor Outlying Islands", "department": "Product", "city": "Sundbyberg", "rate": 0.316, "signup_date": "2023-11-16", "name": "Frank Brandt"}], "dirty_cols": ["company", "country", "department", "city", "rate", "signup_date", "name", "notes2"], "clean_cols": ["company", "country", "department", "city", "rate", "signup_date", "name"], "plan": {"dataset_summary": "36 rows × 8 columns. 7 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"United States Mino Outlying Islands": "United States Minor Outlying Islands", "Kingdom of Morocco": "Morocco", "ZM": "Zambia", "KN": "Saint Kitts and Nevis", "zm": "Zambia", "um": "United States Minor Outlying Islands", "MOROCCO": "Morocco", "zmb": "Zambia", "MAR": "Morocco", "saint kitts and nevis": "Saint Kitts and Nevis", "Kingdom Of Morocco": "Morocco", "UM": "United States Minor Outlying Islands", "Republic Of Zambia": "Zambia", "umi": "United States Minor Outlying Islands", "ma": "Morocco", "UMI": "United States Minor Outlying Islands", "KNA": "Saint Kitts and Nevis", "Zavbia": "Zambia", "Saint Kitts And Nevis": "Saint Kitts and Nevis", "Kn": "Saint Kitts and Nevis", "kna": "Saint Kitts and Nevis", "ZMB": "Zambia"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"R&D": "Engineering", "engineering": "Engineering", "eng.": "Engineering", "Mkt": "Marketing", "Prod": "Product", "Mktg": "Marketing", "cust support": "Customer Support", "growth": "Marketing", "prod": "Product", "Eng": "Engineering", "Eng.": "Engineering", "pm": "Product", "Dev": "Engineering", "Makreting": "Marketing", "DEV": "Engineering", "SUPPORT": "Customer Support", "dev": "Engineering", "MARKETING": "Marketing", "MARKETIGN": "Marketing", "Growth": "Marketing", "cs": "Customer Support", "product": "Product"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pekin": "Pekin", "SUNDBYBERG": "Sundbyberg", "Nsatre": "Nsoatre", "CHAIYAPHUM": "Chaiyaphum", "nsoatre": "Nsoatre", "sundbyberg": "Sundbyberg", "al khurmah": "Al Khurmah", "peekin": "Pekin", "nsoate": "Nsoatre", "chaiyaphum": "Chaiyaphum", "AL KHURMAH": "Al Khurmah", "NSOATRE": "Nsoatre"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"status": " overdue", "industry": "IT", "company": "  Initech", "city": "Pekin", "email": "FJBKAM@CORP.IO", "job_title": "CEO  "}, {"status": "PID", "industry": "MANUFACTURING", "company": " Vehement ", "city": "Adam", "email": "DKIMH@MAIL.COM   ", "job_title": "Rep"}, {"status": "OVERDUE", "industry": "IT", "company": "   Globex", "city": "Kirguli", "email": " EPMOP@EXAMPLE.COM ", "job_title": "engineering mgr"}, {"status": "refunded", "industry": "oil & gas", "company": "Globex ", "city": "Bombay", "email": "OHEAAG@EXAMPLE.COM", "job_title": "Engineering Manager"}, {"status": "paid", "industry": "   Mfg", "company": "Umbrella", "city": "  mumbai", "email": "fedfhlk@corp.io", "job_title": "Engineering Manager"}, {"status": "REUND", "industry": "hotlls & travel", "company": "Acme  Inc", "city": "Mumbai", "email": "knkec@corp.io", "job_title": "Chief Executive Officer"}, {"status": "DUE", "industry": "MANUFACTURING", "company": "Wonka  Co", "city": "MUMBAI", "email": "cemn@example.com", "job_title": "Sales Representative"}, {"status": "UNPAID", "industry": "utilities", "company": "Acme Inc", "city": "Amsterdam", "email": "GIIGO@EXAMPLE.COM  ", "job_title": "CTO"}, {"status": "Padi", "industry": "  hotels & travel", "company": " Vehement", "city": "KIRGULI", "email": "OAGB@EXAMPLE.COM", "job_title": "Rep"}, {"status": "REFUNDED", "industry": "HOTELS & TRAVEL", "company": " Hooli ", "city": "Yautepec", "email": "DPKJOD@CORP.IO", "job_title": "C.T.O."}, {"status": "Refund", "industry": "OIL & GAS", "company": "Wonka Co", "city": "ADAM", "email": "FBFLM@TEST.ORG", "job_title": "c.e.o."}, {"status": "Pending", "industry": "TOURISM", "company": "Vehement", "city": "pekin", "email": "kpdaoe@test.org", "job_title": " EngrMgr "}, {"status": "PENDING", "industry": "SOFTAWRE", "company": "Stark Industries", "city": "Mumbai", "email": "PKEJ@CORP.IO", "job_title": "sales rep"}, {"status": "refunded", "industry": "HOSPITALOTY", "company": "Stark  Industries", "city": "Kirguli", "email": "NEIACD@EXAMPLE.COM   ", "job_title": "C.T.O."}, {"status": " OVERDUE ", "industry": "Oil & Gas", "company": "Globex", "city": "YAUTEPEC   ", "email": "JNDCALB@TEST.ORG", "job_title": "CEO"}, {"status": "refund ", "industry": " Manufacturing ", "company": "Cyberdyne", "city": " Kiguli", "email": "mpmch@corp.io", "job_title": "C.T.O."}, {"status": "LATE", "industry": "Mfg   ", "company": "Wonka Co", "city": "Kirgui", "email": "HCHMGN@TEST.ORG", "job_title": "Engineering Mgr"}, {"status": "", "industry": "", "company": "", "city": "", "email": "", "job_title": ""}], "clean": [{"status": "Overdue", "industry": "Technology", "company": "Initech", "city": "Pekin", "email": "fjbkam@corp.io", "job_title": "Chief Executive Officer"}, {"status": "Paid", "industry": "Manufacturing", "company": "Vehement", "city": "Amsterdam", "email": "dkimh@mail.com", "job_title": "Sales Representative"}, {"status": "Overdue", "industry": "Technology", "company": "Globex", "city": "Kirguli", "email": "epmop@example.com", "job_title": "Engineering Manager"}, {"status": "Refunded", "industry": "Energy", "company": "Globex", "city": "Mumbai", "email": "oheaag@example.com", "job_title": "Engineering Manager"}, {"status": "Paid", "industry": "Manufacturing", "company": "Umbrella", "city": "Mumbai", "email": "fedfhlk@corp.io", "job_title": "Engineering Manager"}, {"status": "Refunded", "industry": "Hospitality", "company": "Acme Inc", "city": "Mumbai", "email": "knkec@corp.io", "job_title": "Chief Executive Officer"}, {"status": "Pending", "industry": "Manufacturing", "company": "Wonka Co", "city": "Mumbai", "email": "cemn@example.com", "job_title": "Sales Representative"}, {"status": "Pending", "industry": "Energy", "company": "Acme Inc", "city": "Amsterdam", "email": "giigo@example.com", "job_title": "Chief Technology Officer"}, {"status": "Paid", "industry": "Hospitality", "company": "Vehement", "city": "Kirguli", "email": "oagb@example.com", "job_title": "Sales Representative"}, {"status": "Refunded", "industry": "Hospitality", "company": "Hooli", "city": "Yautepec", "email": "dpkjod@corp.io", "job_title": "Chief Technology Officer"}, {"status": "Refunded", "industry": "Energy", "company": "Wonka Co", "city": "Amsterdam", "email": "fbflm@test.org", "job_title": "Chief Executive Officer"}, {"status": "Pending", "industry": "Hospitality", "company": "Vehement", "city": "Pekin", "email": "kpdaoe@test.org", "job_title": "Engineering Manager"}, {"status": "Pending", "industry": "Technology", "company": "Stark Industries", "city": "Mumbai", "email": "pkej@corp.io", "job_title": "Sales Representative"}, {"status": "Refunded", "industry": "Hospitality", "company": "Stark Industries", "city": "Kirguli", "email": "neiacd@example.com", "job_title": "Chief Technology Officer"}, {"status": "Overdue", "industry": "Energy", "company": "Globex", "city": "Yautepec", "email": "jndcalb@test.org", "job_title": "Chief Executive Officer"}, {"status": "Refunded", "industry": "Manufacturing", "company": "Cyberdyne", "city": "Kirguli", "email": "mpmch@corp.io", "job_title": "Chief Technology Officer"}, {"status": "Overdue", "industry": "Manufacturing", "company": "Wonka Co", "city": "Kirguli", "email": "hchmgn@test.org", "job_title": "Engineering Manager"}], "dirty_cols": ["status", "industry", "company", "city", "email", "job_title"], "clean_cols": ["status", "industry", "company", "city", "email", "job_title"], "plan": {"dataset_summary": "18 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"overdue": "Overdue", "PID": "Paid", "OVERDUE": "Overdue", "refunded": "Refunded", "paid": "Paid", "REUND": "Refunded", "DUE": "Pending", "UNPAID": "Pending", "Padi": "Paid", "REFUNDED": "Refunded", "Refund": "Refunded", "PENDING": "Pending", "refund": "Refunded", "LATE": "Overdue"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Technology", "MANUFACTURING": "Manufacturing", "oil & gas": "Energy", "Mfg": "Manufacturing", "hotlls & travel": "Hospitality", "utilities": "Energy", "hotels & travel": "Hospitality", "HOTELS & TRAVEL": "Hospitality", "OIL & GAS": "Energy", "TOURISM": "Hospitality", "SOFTAWRE": "Technology", "HOSPITALOTY": "Hospitality", "Oil & Gas": "Energy"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Adam": "Amsterdam", "Bombay": "Mumbai", "mumbai": "Mumbai", "MUMBAI": "Mumbai", "KIRGULI": "Kirguli", "ADAM": "Amsterdam", "pekin": "Pekin", "YAUTEPEC": "Yautepec", "Kiguli": "Kirguli", "Kirgui": "Kirguli"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CEO": "Chief Executive Officer", "Rep": "Sales Representative", "engineering mgr": "Engineering Manager", "CTO": "Chief Technology Officer", "C.T.O.": "Chief Technology Officer", "c.e.o.": "Chief Executive Officer", "EngrMgr": "Engineering Manager", "sales rep": "Sales Representative", "Engineering Mgr": "Engineering Manager"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": "Mkt", "company": " Vehement", "status": "M", "country": "Guadeloupe", "signup_date": "22 Sep 2023", "amount": "?", "city": "saguor", "industry": "Tech", "notes2": ""}, {"department": " cs ", "company": "Globex", "status": "L", "country": "GP", "signup_date": "14 Jan 2023", "amount": "#N/A", "city": "  Saugor", "industry": "Software", "notes2": ""}, {"department": "People Ops", "company": "Hooli", "status": "h", "country": "guadeloupe", "signup_date": "10 Dec 2023", "amount": "126,21", "city": " neuillysurseine ", "industry": "   Mfg", "notes2": ""}, {"department": "Growth", "company": " Umbrella ", "status": "Low", "country": "Guadeloupe", "signup_date": "2023-05-01", "amount": "$4,416.32", "city": "NeuillysurSeine", "industry": "manufacturing", "notes2": ""}, {"department": " PEOPLE OPS", "company": "Soylent Corp", "status": "P3", "country": "Jamaica", "signup_date": "4/15/2023", "amount": "$5,366.67", "city": "KWINANA", "industry": "TECH", "notes2": ""}, {"department": "Human  Resources", "company": "Vehement", "status": " H ", "country": "JAM", "signup_date": "45044", "amount": "4.979,67", "city": "ifrane", "industry": "Technology", "notes2": ""}, {"department": "SUPPORT", "company": " Cyberdyne ", "status": "HIH", "country": "Guadeloupe", "signup_date": "2023-10-01", "amount": "None", "city": " Kwinana ", "industry": "SOFTWARE", "notes2": ""}, {"department": "people ops", "company": "Hooli", "status": "L", "country": "Guadeloupe", "signup_date": "2/16/2023", "amount": "406,03", "city": "SIOCON", "industry": "HALTH", "notes2": ""}, {"department": "customer  support", "company": "   Acme Inc", "status": "MED", "country": "GLP", "signup_date": "10/16/2023", "amount": "8.078,03", "city": "Kwinana", "industry": "   Tech", "notes2": ""}, {"department": "mkt", "company": " Wonka Co", "status": "HIGH", "country": "jm", "signup_date": "9/16/2023", "amount": "8.488,40", "city": "ifrane", "industry": "Industrial", "notes2": ""}, {"department": "mktg", "company": "Globex  ", "status": "  H", "country": "JAM  ", "signup_date": "2023-03-17", "amount": "-100", "city": "Saugor", "industry": "medical", "notes2": ""}, {"department": "cs", "company": "Acme Inc", "status": "medium", "country": "JAM", "signup_date": "2023-04-03", "amount": "$8,908.82", "city": "neuilly-sur-seine", "industry": "IT", "notes2": ""}, {"department": "People", "company": "Wonka  Co", "status": "med", "country": "   gp", "signup_date": "1/5/2023", "amount": "8.206,98", "city": "Saugor", "industry": "indutsrial", "notes2": ""}, {"department": "people ops", "company": " Initech ", "status": "M", "country": "jam", "signup_date": "2023-01-02", "amount": "4.089,98", "city": "siocon", "industry": "Manufacturing", "notes2": ""}, {"department": "mkt", "company": "Stark Industries", "status": "   p3", "country": "Jamaica", "signup_date": "45238", "amount": "2.543,11", "city": " Saugor ", "industry": "Health Care", "notes2": ""}, {"department": "Human Resources", "company": "Hooli", "status": " High ", "country": "gp", "signup_date": "45111", "amount": "-", "city": "  KWINANA", "industry": " IT ", "notes2": ""}, {"department": "Cust Support", "company": "Umbrella", "status": "medium", "country": "JM", "signup_date": "28 Nov 2023", "amount": "5.833,81", "city": "Kwiana", "industry": "Health Care", "notes2": ""}, {"department": "Customer Support", "company": "   Umbrella", "status": "LOW", "country": "jamaica", "signup_date": "2023-03-13", "amount": "$6,444.97", "city": "kwinana", "industry": "SOFTWAE", "notes2": ""}, {"department": "Human Resources", "company": "Hooli", "status": "Low", "country": "   Jm", "signup_date": "2023-06-13", "amount": "2.742,73", "city": "Neuilly-sur-Seine", "industry": "Mfg", "notes2": ""}, {"department": "Human Resources", "company": "Cyberdyne", "status": "P2", "country": "JM", "signup_date": "3/1/2023", "amount": "5.905,57", "city": "Irfane", "industry": "manufacturing", "notes2": ""}, {"department": "Peole", "company": "Stark Industries", "status": "high", "country": "guadeloupe", "signup_date": "2023-11-22", "amount": "4.840,75", "city": "Siocon", "industry": "Health Care ", "notes2": ""}, {"department": "Growht", "company": "  Globex", "status": "M", "country": "guadeloupe", "signup_date": "2023-03-23", "amount": "209,92", "city": "Sugor", "industry": "Healthare", "notes2": ""}, {"department": "CS", "company": " Acme Inc", "status": "High", "country": "Glp", "signup_date": "16 Jan 2023", "amount": "$5,640.44", "city": "Kwinana  ", "industry": "Software", "notes2": ""}, {"department": "hr", "company": "Globex", "status": "HIGH  ", "country": "  glp", "signup_date": "12 Dec 2023", "amount": "7.348,78", "city": "neuillysurseine", "industry": "Manufacturing", "notes2": ""}, {"department": "Growth", "company": "Acme Inc", "status": "P1", "country": "GP", "signup_date": "2023-02-04", "amount": "92,28", "city": "saugor", "industry": "heacth   ", "notes2": ""}, {"department": "   CUSTOMER SUPPORT", "company": " Acme Inc", "status": "Medium", "country": "guadeloupe", "signup_date": "6 Sep 2023", "amount": "na", "city": "Saugor", "industry": "Mfg", "notes2": ""}, {"department": "cust support   ", "company": "Soylent Corp", "status": " P3 ", "country": "Jamaica", "signup_date": "2023-03-07", "amount": "$6,045.40", "city": "Saugor", "industry": "teph", "notes2": ""}, {"department": "hr", "company": "Globex", "status": "HIGH  ", "country": "  glp", "signup_date": "12 Dec 2023", "amount": "7.348,78", "city": "neuillysurseine", "industry": "Manufacturing", "notes2": ""}], "clean": [{"department": "Marketing", "company": "Vehement", "status": "Medium", "country": "Guadeloupe", "signup_date": "2023-09-22", "amount": NaN, "city": "Saugor", "industry": "Technology"}, {"department": "Customer Support", "company": "Globex", "status": "Low", "country": "Guadeloupe", "signup_date": "2023-01-14", "amount": NaN, "city": "Saugor", "industry": "Technology"}, {"department": "Human Resources", "company": "Hooli", "status": "High", "country": "Guadeloupe", "signup_date": "2023-12-10", "amount": 126.21, "city": "Neuilly-sur-Seine", "industry": "Manufacturing"}, {"department": "Marketing", "company": "Umbrella", "status": "Low", "country": "Guadeloupe", "signup_date": "2023-05-01", "amount": 4416.32, "city": "Neuilly-sur-Seine", "industry": "Manufacturing"}, {"department": "Human Resources", "company": "Soylent Corp", "status": "Low", "country": "Jamaica", "signup_date": "2023-04-15", "amount": 5366.67, "city": "Kwinana", "industry": "Technology"}, {"department": "Human Resources", "company": "Vehement", "status": "High", "country": "Jamaica", "signup_date": "2023-04-28", "amount": 4979.67, "city": "Ifrane", "industry": "Technology"}, {"department": "Customer Support", "company": "Cyberdyne", "status": "High", "country": "Guadeloupe", "signup_date": "2023-10-01", "amount": NaN, "city": "Kwinana", "industry": "Technology"}, {"department": "Human Resources", "company": "Hooli", "status": "Low", "country": "Guadeloupe", "signup_date": "2023-02-16", "amount": 406.03, "city": "Siocon", "industry": "Healthcare"}, {"department": "Customer Support", "company": "Acme Inc", "status": "Medium", "country": "Guadeloupe", "signup_date": "2023-10-16", "amount": 8078.03, "city": "Kwinana", "industry": "Technology"}, {"department": "Marketing", "company": "Wonka Co", "status": "High", "country": "Jamaica", "signup_date": "2023-09-16", "amount": 8488.4, "city": "Ifrane", "industry": "Manufacturing"}, {"department": "Marketing", "company": "Globex", "status": "High", "country": "Jamaica", "signup_date": "2023-03-17", "amount": -100.0, "city": "Saugor", "industry": "Healthcare"}, {"department": "Customer Support", "company": "Acme Inc", "status": "Medium", "country": "Jamaica", "signup_date": "2023-04-03", "amount": 8908.82, "city": "Neuilly-sur-Seine", "industry": "Technology"}, {"department": "Human Resources", "company": "Wonka Co", "status": "Medium", "country": "Guadeloupe", "signup_date": "2023-01-05", "amount": 8206.98, "city": "Saugor", "industry": "Manufacturing"}, {"department": "Human Resources", "company": "Initech", "status": "Medium", "country": "Jamaica", "signup_date": "2023-01-02", "amount": 4089.98, "city": "Siocon", "industry": "Manufacturing"}, {"department": "Marketing", "company": "Stark Industries", "status": "Low", "country": "Jamaica", "signup_date": "2023-11-08", "amount": 2543.11, "city": "Saugor", "industry": "Healthcare"}, {"department": "Human Resources", "company": "Hooli", "status": "High", "country": "Guadeloupe", "signup_date": "2023-07-04", "amount": NaN, "city": "Kwinana", "industry": "Technology"}, {"department": "Customer Support", "company": "Umbrella", "status": "Medium", "country": "Jamaica", "signup_date": "2023-11-28", "amount": 5833.81, "city": "Kwinana", "industry": "Healthcare"}, {"department": "Customer Support", "company": "Umbrella", "status": "Low", "country": "Jamaica", "signup_date": "2023-03-13", "amount": 6444.97, "city": "Kwinana", "industry": "Technology"}, {"department": "Human Resources", "company": "Hooli", "status": "Low", "country": "Jamaica", "signup_date": "2023-06-13", "amount": 2742.73, "city": "Neuilly-sur-Seine", "industry": "Manufacturing"}, {"department": "Human Resources", "company": "Cyberdyne", "status": "Medium", "country": "Jamaica", "signup_date": "2023-03-01", "amount": 5905.57, "city": "Ifrane", "industry": "Manufacturing"}, {"department": "Human Resources", "company": "Stark Industries", "status": "High", "country": "Guadeloupe", "signup_date": "2023-11-22", "amount": 4840.75, "city": "Siocon", "industry": "Healthcare"}, {"department": "Marketing", "company": "Globex", "status": "Medium", "country": "Guadeloupe", "signup_date": "2023-03-23", "amount": 209.92, "city": "Saugor", "industry": "Healthcare"}, {"department": "Customer Support", "company": "Acme Inc", "status": "High", "country": "Guadeloupe", "signup_date": "2023-01-16", "amount": 5640.44, "city": "Kwinana", "industry": "Technology"}, {"department": "Human Resources", "company": "Globex", "status": "High", "country": "Guadeloupe", "signup_date": "2023-12-12", "amount": 7348.78, "city": "Neuilly-sur-Seine", "industry": "Manufacturing"}, {"department": "Marketing", "company": "Acme Inc", "status": "High", "country": "Guadeloupe", "signup_date": "2023-02-04", "amount": 92.28, "city": "Saugor", "industry": "Healthcare"}, {"department": "Customer Support", "company": "Acme Inc", "status": "Medium", "country": "Guadeloupe", "signup_date": "2023-09-06", "amount": NaN, "city": "Saugor", "industry": "Manufacturing"}, {"department": "Customer Support", "company": "Soylent Corp", "status": "Low", "country": "Jamaica", "signup_date": "2023-03-07", "amount": 6045.4, "city": "Saugor", "industry": "Technology"}], "dirty_cols": ["department", "company", "status", "country", "signup_date", "amount", "city", "industry", "notes2"], "clean_cols": ["department", "company", "status", "country", "signup_date", "amount", "city", "industry"], "plan": {"dataset_summary": "28 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mkt": "Marketing", "cs": "Customer Support", "People Ops": "Human Resources", "Growth": "Marketing", "PEOPLE OPS": "Human Resources", "SUPPORT": "Customer Support", "people ops": "Human Resources", "customer support": "Customer Support", "mkt": "Marketing", "mktg": "Marketing", "People": "Human Resources", "Cust Support": "Customer Support", "Peole": "Human Resources", "Growht": "Marketing", "CS": "Customer Support", "hr": "Human Resources", "CUSTOMER SUPPORT": "Customer Support", "cust support": "Customer Support"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"M": "Medium", "L": "Low", "h": "High", "P3": "Low", "H": "High", "HIH": "High", "MED": "Medium", "HIGH": "High", "medium": "Medium", "med": "Medium", "p3": "Low", "LOW": "Low", "P2": "Medium", "high": "High", "P1": "High"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GP": "Guadeloupe", "guadeloupe": "Guadeloupe", "JAM": "Jamaica", "GLP": "Guadeloupe", "jm": "Jamaica", "gp": "Guadeloupe", "jam": "Jamaica", "JM": "Jamaica", "jamaica": "Jamaica", "Jm": "Jamaica", "Glp": "Guadeloupe", "glp": "Guadeloupe"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"saguor": "Saugor", "neuillysurseine": "Neuilly-sur-Seine", "NeuillysurSeine": "Neuilly-sur-Seine", "KWINANA": "Kwinana", "ifrane": "Ifrane", "SIOCON": "Siocon", "neuilly-sur-seine": "Neuilly-sur-Seine", "siocon": "Siocon", "Kwiana": "Kwinana", "kwinana": "Kwinana", "Irfane": "Ifrane", "Sugor": "Saugor", "saugor": "Saugor"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Tech": "Technology", "Software": "Technology", "Mfg": "Manufacturing", "manufacturing": "Manufacturing", "TECH": "Technology", "SOFTWARE": "Technology", "HALTH": "Healthcare", "Industrial": "Manufacturing", "medical": "Healthcare", "IT": "Technology", "indutsrial": "Manufacturing", "Health Care": "Healthcare", "SOFTWAE": "Technology", "Healthare": "Healthcare", "heacth": "Healthcare", "teph": "Technology"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"rate": "11.1%", "currency": "Boliviano", "name": " Judy Khan", "email": "kljc@mail.com", "industry": "industrial", "unit": "in", "phone": "5933908280", "signup_date": "6/19/2023", "extra": ""}, {"rate": "60.9%", "currency": "BOLIVIANO", "name": "  Carol Johnson", "email": "FIFBIK@MAIL.COM", "industry": "Financial Services", "unit": " CENTIMETER ", "phone": "(549)650-3292", "signup_date": "3 Jun 2023", "extra": ""}, {"rate": "42.5%", "currency": " BRL ", "name": "Grace Khan", "email": "adbpp@test.org", "industry": "Telecom", "unit": "IN", "phone": "5396681308", "signup_date": "45193", "extra": ""}, {"rate": "66.4%", "currency": "BRL", "name": "Lena Smith", "email": " HELJ@CORP.IO ", "industry": "  mfg", "unit": "centimeter  ", "phone": "570-069-9570", "signup_date": "2023-08-02", "extra": ""}, {"rate": "14.6%", "currency": "BRAZILIAN REAL", "name": "   Frank Khan", "email": "ebcjdj@test.org", "industry": " Fintech ", "unit": "INCH  ", "phone": "5627077140", "signup_date": "45266", "extra": ""}, {"rate": "74.6%", "currency": "   bob", "name": "Priya Park", "email": "PBIFN@EXAMPLE.COM", "industry": "Industrial", "unit": "centimeter", "phone": "(572)342-2657", "signup_date": "1/6/2023", "extra": ""}, {"rate": "23.5%", "currency": "BOB", "name": "Frank Wong", "email": "mcbp@mail.com", "industry": "telco   ", "unit": "   inbhes", "phone": "503-458-4897", "signup_date": "5/17/2023", "extra": ""}, {"rate": "3.2%", "currency": "Boliviano", "name": "   Lena Lee", "email": "agbak@example.com", "industry": "Telco", "unit": "inches", "phone": "(548)880-3998", "signup_date": "2023-01-21", "extra": ""}, {"rate": "22.9%", "currency": "BOB", "name": "  Judy Khan", "email": "  chojbb@mail.com", "industry": "Telecom", "unit": "ich", "phone": "5773180856", "signup_date": "2023-01-01", "extra": ""}, {"rate": "19.2%", "currency": "BRL", "name": "  Eve Moore", "email": "lgba@corp.io", "industry": "Retail & E-commerce", "unit": "INCH ", "phone": "502-829-7849", "signup_date": "6/2/2023", "extra": ""}, {"rate": "53.8%", "currency": "brl", "name": "   Lena Petrov", "email": "mfaic@mail.com", "industry": "Mfg", "unit": "Cm", "phone": "5883310349", "signup_date": "16 Jan 2023", "extra": ""}, {"rate": "70.7%", "currency": "Bob", "name": "Priya Reyes", "email": "caeg@example.com", "industry": "Comms", "unit": "centimeters", "phone": "543-150-4515", "signup_date": "45209", "extra": ""}, {"rate": "33.8%", "currency": "BRAZILIAN REAL", "name": "Heidi Novak ", "email": " gdlh@mail.com", "industry": "financial services", "unit": "Cm", "phone": "5061507266", "signup_date": "1/7/2023", "extra": ""}, {"rate": "8.4%", "currency": "Boliviano", "name": "Grace Smith", "email": "JNAGCEB@TEST.ORG", "industry": "   Retail & E-commerce", "unit": "centimmeters", "phone": "508-052-8737", "signup_date": "45216", "extra": ""}, {"rate": "2.2%", "currency": "BRL", "name": "Bob Diaz   ", "email": " aioom@corp.io", "industry": "Finance", "unit": "inch", "phone": "505-910-1970", "signup_date": "8/1/2023", "extra": ""}, {"rate": "71.5%", "currency": "BOB", "name": "Sara Adams ", "email": "jjkm@mail.com", "industry": "Mfg", "unit": " CM ", "phone": "5358392450", "signup_date": "5/21/2023", "extra": ""}, {"rate": "53.8%", "currency": "bob", "name": "Carol Ali", "email": "eglc@mail.com", "industry": "Manufacturing", "unit": "Inches", "phone": "529-495-9494", "signup_date": "14 Mar 2023", "extra": ""}, {"rate": "50.0%", "currency": "braziliann real", "name": "Frank Ali", "email": "nhng@mail.com", "industry": "Retail", "unit": "centimeter ", "phone": "5280854848", "signup_date": "2023-02-09", "extra": ""}, {"rate": "77.6%", "currency": " BOB ", "name": "Mona Lee", "email": "oghbp@corp.io", "industry": "retail", "unit": "in", "phone": "5823387225", "signup_date": "2023-10-18", "extra": ""}, {"rate": "10.2%", "currency": "boliviano", "name": "Bob  Wong", "email": " MAJPG@CORP.IO ", "industry": "Manufacturing", "unit": "Inch", "phone": "578.718.7045", "signup_date": "10 Apr 2023", "extra": ""}, {"rate": "92.7%", "currency": "brazilian real", "name": "David Cruz", "email": "CHDPDA@TEST.ORG", "industry": " Telco ", "unit": "   centimeters", "phone": "506-657-1678", "signup_date": "12/13/2023", "extra": ""}, {"rate": "68.8%", "currency": "Bolviano", "name": "Mona Johnson", "email": "  IKDF@TEST.ORG", "industry": "Retail", "unit": "Inches", "phone": "574-528-4954", "signup_date": "18 Apr 2023", "extra": ""}, {"rate": "98.4%", "currency": " BRAZILIAN REAL", "name": "Omar Wong", "email": "HGKCND@TEST.ORG", "industry": "Comms", "unit": "\"", "phone": "(547)741-5531", "signup_date": "3 Jun 2023", "extra": ""}, {"rate": "5.8%", "currency": "BOB", "name": "Frank Ali", "email": "GKBJEM@TEST.ORG", "industry": "Manufayturing", "unit": "inch", "phone": "547.135.8464", "signup_date": "8 Feb 2023", "extra": ""}, {"rate": "0.2%", "currency": "BOB", "name": " Lena Petrov", "email": "boconjl@example.com  ", "industry": "industrial", "unit": "cm", "phone": "5683875513", "signup_date": "27 Feb 2023", "extra": ""}, {"rate": "94.8%", "currency": "Brazilian Real", "name": "Karl  Johnson", "email": "ICEM@EXAMPLE.COM", "industry": "finance", "unit": "CENTIMETERS", "phone": "571.789.3338", "signup_date": "5/13/2023", "extra": ""}, {"rate": "60.9%", "currency": "BOB", "name": " David Adams", "email": "OLKOCDH@TEST.ORG", "industry": "retail  ", "unit": "incch", "phone": "(553)983-3439", "signup_date": "8 Sep 2023", "extra": ""}, {"rate": "15.9%", "currency": "BOB   ", "name": "  Heidi Adams", "email": " pmjmi@example.com ", "industry": "Fintech", "unit": " Cm ", "phone": "598.033.1088", "signup_date": "25 Apr 2023", "extra": ""}, {"rate": "80.4%", "currency": "Brazilian Real", "name": "Judy Lee", "email": "dmfeg@corp.io", "industry": "Financial Services", "unit": "centimeter", "phone": "(580)142-5257", "signup_date": "22 Feb 2023", "extra": ""}, {"rate": "14.5%", "currency": "brl", "name": " Karl Novak", "email": "gfkf@test.org", "industry": "Comms", "unit": "cm ", "phone": "567.767.4500", "signup_date": "19 May 2023", "extra": ""}, {"rate": "26.0%", "currency": "braziilian real", "name": "Karl Moore", "email": "KIIKAJD@EXAMPLE.COM", "industry": "teleocmmunications", "unit": "cntimeters", "phone": "(500)991-1115", "signup_date": "5/1/2023", "extra": ""}, {"rate": "71.7%", "currency": "Boliviano", "name": "Omar Cruz", "email": "   ACLB@MAIL.COM", "industry": "Fintech", "unit": " centmieters ", "phone": "5147643925", "signup_date": "2 Feb 2023", "extra": ""}, {"rate": "51.4%", "currency": "BOOLIVIANO ", "name": "Sara Wong", "email": "ibofap@test.org", "industry": "INDUSTRIAL", "unit": "centimeters", "phone": "576-963-6819", "signup_date": "27 Jun 2023", "extra": ""}, {"rate": "61.3%", "currency": "Brazilian Real", "name": "Judy Novak   ", "email": "  KMLMKJM@CORP.IO", "industry": "COMSM", "unit": "cm   ", "phone": "5150630242", "signup_date": "2023-10-28", "extra": ""}, {"rate": "78.8%", "currency": " bob ", "name": "Omar Brandt", "email": "IAOADIP@MAIL.COM", "industry": "Telecommunications", "unit": "cm", "phone": "554-733-1425", "signup_date": "45111", "extra": ""}, {"rate": "74.7%", "currency": "brazilian real", "name": " David Reyes", "email": "mjdm@mail.com", "industry": "financial services", "unit": "centimeters", "phone": "5144830692", "signup_date": "2023-11-12", "extra": ""}, {"rate": "98.7%", "currency": "Boliviano", "name": "Omar Park", "email": "NJID@MAIL.COM", "industry": "teecom", "unit": "Centimeter", "phone": "(561)964-8359", "signup_date": "2023-09-01", "extra": ""}, {"rate": "98.3%", "currency": " BOLIVIANO", "name": "Karl Johnson", "email": "   ihdbe@test.org", "industry": "Banking", "unit": " cm", "phone": "517-987-4851", "signup_date": "2023-01-14", "extra": ""}, {"rate": "70.5%", "currency": " Boliviano ", "name": "Alice Fischer", "email": "ABKMBOM@MAIL.COM ", "industry": "Industrial", "unit": "inch", "phone": "(516)498-8222", "signup_date": "8 Jan 2023", "extra": ""}, {"rate": "81.6%", "currency": "Brl", "name": "Priya Diaz", "email": "CEPBF@CORP.IO", "industry": "manufacturing", "unit": "in", "phone": "505.435.5168", "signup_date": "2 May 2023", "extra": ""}, {"rate": "63.8%", "currency": "Boliviano", "name": "Karl Reyes", "email": "pejkjfb@corp.io", "industry": "TECLO", "unit": "  in", "phone": "(536)071-4600", "signup_date": "45255", "extra": ""}, {"rate": "67.4%", "currency": "Brazilian  Real", "name": "Frank Adams  ", "email": " HJNILME@TEST.ORG ", "industry": "MANUAFCTURING", "unit": "\"", "phone": "(596)774-6477", "signup_date": "2023-11-19", "extra": ""}, {"rate": "19.8%", "currency": "BOB", "name": "Bob Park", "email": "  ffkjcf@mail.com", "industry": "Chmms", "unit": "cm", "phone": "594.668.4001", "signup_date": "45133", "extra": ""}, {"rate": "49.6%", "currency": " BRL", "name": "Karl Fischer", "email": "DIDOAC@CORP.IO", "industry": "Manufacturing", "unit": "In", "phone": "599.234.8001", "signup_date": "2023-06-21", "extra": ""}, {"rate": "61.6%", "currency": "  BOB", "name": "  Grace Diaz", "email": " IMPFA@EXAMPLE.COM ", "industry": "manufacturing", "unit": "in", "phone": "5205989709", "signup_date": "45135", "extra": ""}, {"rate": "77.0%", "currency": "boliviano", "name": "   Frank Reyes", "email": "hfcpffb@corp.io", "industry": "FiinServ", "unit": "inch", "phone": "554.844.6901", "signup_date": "44961", "extra": ""}, {"rate": "9.0%", "currency": "BRL", "name": "Bob Khan", "email": "  HNFL@MAIL.COM", "industry": "Telecom", "unit": "centimeters", "phone": "5030977002", "signup_date": "15 Jul 2023", "extra": ""}, {"rate": "37.1%", "currency": "Boliviano", "name": "   Lena Diaz", "email": "CEKL@EXAMPLE.COM", "industry": "telecommunications  ", "unit": "\"", "phone": "(501)190-7586", "signup_date": "9 Aug 2023", "extra": ""}, {"rate": "34.9%", "currency": "brazilian real", "name": "  Omar Brandt", "email": "AFBEEJE@CORP.IO  ", "industry": "manufacturing", "unit": "CENTIMETERS", "phone": "547.776.7906", "signup_date": "5 Oct 2023", "extra": ""}, {"rate": "96.4%", "currency": "Boliviano", "name": "Karl Petrov", "email": "GBDF@TEST.ORG   ", "industry": "FINANCIAL SERVICES", "unit": "centimeters", "phone": "(582)817-5272", "signup_date": "7 Sep 2023", "extra": ""}, {"rate": "72.6%", "currency": "Brazilian Real", "name": "Frank  Petrov", "email": "oejopdl@corp.io", "industry": "e-commerce  ", "unit": "centimeters", "phone": "589.638.2499", "signup_date": "10/10/2023", "extra": ""}, {"rate": "36.1%", "currency": "brazilian  real", "name": "Judy Lee", "email": "ignobp@test.org", "industry": "manufacturing", "unit": "CM", "phone": "(577)625-0848", "signup_date": "5/24/2023", "extra": ""}, {"rate": "36.3%", "currency": "Brazilian Real", "name": "Mona  Cruz", "email": " GAPG@MAIL.COM", "industry": "   Finanke", "unit": "centimeters", "phone": "577-012-1803", "signup_date": "11/11/2023", "extra": ""}, {"rate": "77.6%", "currency": " BOB ", "name": "Mona Lee", "email": "oghbp@corp.io", "industry": "retail", "unit": "in", "phone": "5823387225", "signup_date": "2023-10-18", "extra": ""}, {"rate": "", "currency": "", "name": "", "email": "", "industry": "", "unit": "", "phone": "", "signup_date": "", "extra": ""}], "clean": [{"rate": 0.111, "currency": "BOB", "name": "Judy Khan", "email": "kljc@mail.com", "industry": "Manufacturing", "unit": "in", "phone": "(593) 390-8280", "signup_date": "2023-06-19"}, {"rate": 0.609, "currency": "BOB", "name": "Carol Johnson", "email": "fifbik@mail.com", "industry": "Financial Services", "unit": "cm", "phone": "(549) 650-3292", "signup_date": "2023-06-03"}, {"rate": 0.425, "currency": "BRL", "name": "Grace Khan", "email": "adbpp@test.org", "industry": "Telecommunications", "unit": "in", "phone": "(539) 668-1308", "signup_date": "2023-09-24"}, {"rate": 0.664, "currency": "BRL", "name": "Lena Smith", "email": "helj@corp.io", "industry": "Manufacturing", "unit": "cm", "phone": "(570) 069-9570", "signup_date": "2023-08-02"}, {"rate": 0.146, "currency": "BRL", "name": "Frank Khan", "email": "ebcjdj@test.org", "industry": "Financial Services", "unit": "in", "phone": "(562) 707-7140", "signup_date": "2023-12-06"}, {"rate": 0.746, "currency": "BOB", "name": "Priya Park", "email": "pbifn@example.com", "industry": "Manufacturing", "unit": "cm", "phone": "(572) 342-2657", "signup_date": "2023-01-06"}, {"rate": 0.235, "currency": "BOB", "name": "Frank Wong", "email": "mcbp@mail.com", "industry": "Telecommunications", "unit": "in", "phone": "(503) 458-4897", "signup_date": "2023-05-17"}, {"rate": 0.032, "currency": "BOB", "name": "Lena Lee", "email": "agbak@example.com", "industry": "Telecommunications", "unit": "in", "phone": "(548) 880-3998", "signup_date": "2023-01-21"}, {"rate": 0.22899999999999998, "currency": "BOB", "name": "Judy Khan", "email": "chojbb@mail.com", "industry": "Telecommunications", "unit": "in", "phone": "(577) 318-0856", "signup_date": "2023-01-01"}, {"rate": 0.192, "currency": "BRL", "name": "Eve Moore", "email": "lgba@corp.io", "industry": "Retail", "unit": "in", "phone": "(502) 829-7849", "signup_date": "2023-06-02"}, {"rate": 0.5379999999999999, "currency": "BRL", "name": "Lena Petrov", "email": "mfaic@mail.com", "industry": "Manufacturing", "unit": "cm", "phone": "(588) 331-0349", "signup_date": "2023-01-16"}, {"rate": 0.7070000000000001, "currency": "BOB", "name": "Priya Reyes", "email": "caeg@example.com", "industry": "Telecommunications", "unit": "cm", "phone": "(543) 150-4515", "signup_date": "2023-10-10"}, {"rate": 0.33799999999999997, "currency": "BRL", "name": "Heidi Novak", "email": "gdlh@mail.com", "industry": "Financial Services", "unit": "cm", "phone": "(506) 150-7266", "signup_date": "2023-01-07"}, {"rate": 0.084, "currency": "BOB", "name": "Grace Smith", "email": "jnagceb@test.org", "industry": "Retail", "unit": "cm", "phone": "(508) 052-8737", "signup_date": "2023-10-17"}, {"rate": 0.022000000000000002, "currency": "BRL", "name": "Bob Diaz", "email": "aioom@corp.io", "industry": "Financial Services", "unit": "in", "phone": "(505) 910-1970", "signup_date": "2023-08-01"}, {"rate": 0.715, "currency": "BOB", "name": "Sara Adams", "email": "jjkm@mail.com", "industry": "Manufacturing", "unit": "cm", "phone": "(535) 839-2450", "signup_date": "2023-05-21"}, {"rate": 0.5379999999999999, "currency": "BOB", "name": "Carol Ali", "email": "eglc@mail.com", "industry": "Manufacturing", "unit": "in", "phone": "(529) 495-9494", "signup_date": "2023-03-14"}, {"rate": 0.5, "currency": "BRL", "name": "Frank Ali", "email": "nhng@mail.com", "industry": "Retail", "unit": "cm", "phone": "(528) 085-4848", "signup_date": "2023-02-09"}, {"rate": 0.7759999999999999, "currency": "BOB", "name": "Mona Lee", "email": "oghbp@corp.io", "industry": "Retail", "unit": "in", "phone": "(582) 338-7225", "signup_date": "2023-10-18"}, {"rate": 0.102, "currency": "BOB", "name": "Bob Wong", "email": "majpg@corp.io", "industry": "Manufacturing", "unit": "in", "phone": "(578) 718-7045", "signup_date": "2023-04-10"}, {"rate": 0.927, "currency": "BRL", "name": "David Cruz", "email": "chdpda@test.org", "industry": "Telecommunications", "unit": "cm", "phone": "(506) 657-1678", "signup_date": "2023-12-13"}, {"rate": 0.688, "currency": "BOB", "name": "Mona Johnson", "email": "ikdf@test.org", "industry": "Retail", "unit": "in", "phone": "(574) 528-4954", "signup_date": "2023-04-18"}, {"rate": 0.9840000000000001, "currency": "BRL", "name": "Omar Wong", "email": "hgkcnd@test.org", "industry": "Telecommunications", "unit": "in", "phone": "(547) 741-5531", "signup_date": "2023-06-03"}, {"rate": 0.057999999999999996, "currency": "BOB", "name": "Frank Ali", "email": "gkbjem@test.org", "industry": "Manufacturing", "unit": "in", "phone": "(547) 135-8464", "signup_date": "2023-02-08"}, {"rate": 0.002, "currency": "BOB", "name": "Lena Petrov", "email": "boconjl@example.com", "industry": "Manufacturing", "unit": "cm", "phone": "(568) 387-5513", "signup_date": "2023-02-27"}, {"rate": 0.948, "currency": "BRL", "name": "Karl Johnson", "email": "icem@example.com", "industry": "Financial Services", "unit": "cm", "phone": "(571) 789-3338", "signup_date": "2023-05-13"}, {"rate": 0.609, "currency": "BOB", "name": "David Adams", "email": "olkocdh@test.org", "industry": "Retail", "unit": "in", "phone": "(553) 983-3439", "signup_date": "2023-09-08"}, {"rate": 0.159, "currency": "BOB", "name": "Heidi Adams", "email": "pmjmi@example.com", "industry": "Financial Services", "unit": "cm", "phone": "(598) 033-1088", "signup_date": "2023-04-25"}, {"rate": 0.804, "currency": "BRL", "name": "Judy Lee", "email": "dmfeg@corp.io", "industry": "Financial Services", "unit": "cm", "phone": "(580) 142-5257", "signup_date": "2023-02-22"}, {"rate": 0.145, "currency": "BRL", "name": "Karl Novak", "email": "gfkf@test.org", "industry": "Telecommunications", "unit": "cm", "phone": "(567) 767-4500", "signup_date": "2023-05-19"}, {"rate": 0.26, "currency": "BRL", "name": "Karl Moore", "email": "kiikajd@example.com", "industry": "Telecommunications", "unit": "cm", "phone": "(500) 991-1115", "signup_date": "2023-05-01"}, {"rate": 0.7170000000000001, "currency": "BOB", "name": "Omar Cruz", "email": "aclb@mail.com", "industry": "Financial Services", "unit": "cm", "phone": "(514) 764-3925", "signup_date": "2023-02-02"}, {"rate": 0.514, "currency": "BOB", "name": "Sara Wong", "email": "ibofap@test.org", "industry": "Manufacturing", "unit": "cm", "phone": "(576) 963-6819", "signup_date": "2023-06-27"}, {"rate": 0.613, "currency": "BRL", "name": "Judy Novak", "email": "kmlmkjm@corp.io", "industry": "Telecommunications", "unit": "cm", "phone": "(515) 063-0242", "signup_date": "2023-10-28"}, {"rate": 0.7879999999999999, "currency": "BOB", "name": "Omar Brandt", "email": "iaoadip@mail.com", "industry": "Telecommunications", "unit": "cm", "phone": "(554) 733-1425", "signup_date": "2023-07-04"}, {"rate": 0.747, "currency": "BRL", "name": "David Reyes", "email": "mjdm@mail.com", "industry": "Financial Services", "unit": "cm", "phone": "(514) 483-0692", "signup_date": "2023-11-12"}, {"rate": 0.987, "currency": "BOB", "name": "Omar Park", "email": "njid@mail.com", "industry": "Telecommunications", "unit": "cm", "phone": "(561) 964-8359", "signup_date": "2023-09-01"}, {"rate": 0.983, "currency": "BOB", "name": "Karl Johnson", "email": "ihdbe@test.org", "industry": "Financial Services", "unit": "cm", "phone": "(517) 987-4851", "signup_date": "2023-01-14"}, {"rate": 0.705, "currency": "BOB", "name": "Alice Fischer", "email": "abkmbom@mail.com", "industry": "Manufacturing", "unit": "in", "phone": "(516) 498-8222", "signup_date": "2023-01-08"}, {"rate": 0.816, "currency": "BRL", "name": "Priya Diaz", "email": "cepbf@corp.io", "industry": "Manufacturing", "unit": "in", "phone": "(505) 435-5168", "signup_date": "2023-05-02"}, {"rate": 0.638, "currency": "BOB", "name": "Karl Reyes", "email": "pejkjfb@corp.io", "industry": "Telecommunications", "unit": "in", "phone": "(536) 071-4600", "signup_date": "2023-11-25"}, {"rate": 0.674, "currency": "BRL", "name": "Frank Adams", "email": "hjnilme@test.org", "industry": "Manufacturing", "unit": "in", "phone": "(596) 774-6477", "signup_date": "2023-11-19"}, {"rate": 0.198, "currency": "BOB", "name": "Bob Park", "email": "ffkjcf@mail.com", "industry": "Telecommunications", "unit": "cm", "phone": "(594) 668-4001", "signup_date": "2023-07-26"}, {"rate": 0.496, "currency": "BRL", "name": "Karl Fischer", "email": "didoac@corp.io", "industry": "Manufacturing", "unit": "in", "phone": "(599) 234-8001", "signup_date": "2023-06-21"}, {"rate": 0.616, "currency": "BOB", "name": "Grace Diaz", "email": "impfa@example.com", "industry": "Manufacturing", "unit": "in", "phone": "(520) 598-9709", "signup_date": "2023-07-28"}, {"rate": 0.77, "currency": "BOB", "name": "Frank Reyes", "email": "hfcpffb@corp.io", "industry": "Financial Services", "unit": "in", "phone": "(554) 844-6901", "signup_date": "2023-02-04"}, {"rate": 0.09, "currency": "BRL", "name": "Bob Khan", "email": "hnfl@mail.com", "industry": "Telecommunications", "unit": "cm", "phone": "(503) 097-7002", "signup_date": "2023-07-15"}, {"rate": 0.371, "currency": "BOB", "name": "Lena Diaz", "email": "cekl@example.com", "industry": "Telecommunications", "unit": "in", "phone": "(501) 190-7586", "signup_date": "2023-08-09"}, {"rate": 0.349, "currency": "BRL", "name": "Omar Brandt", "email": "afbeeje@corp.io", "industry": "Manufacturing", "unit": "cm", "phone": "(547) 776-7906", "signup_date": "2023-10-05"}, {"rate": 0.9640000000000001, "currency": "BOB", "name": "Karl Petrov", "email": "gbdf@test.org", "industry": "Financial Services", "unit": "cm", "phone": "(582) 817-5272", "signup_date": "2023-09-07"}, {"rate": 0.726, "currency": "BRL", "name": "Frank Petrov", "email": "oejopdl@corp.io", "industry": "Retail", "unit": "cm", "phone": "(589) 638-2499", "signup_date": "2023-10-10"}, {"rate": 0.361, "currency": "BRL", "name": "Judy Lee", "email": "ignobp@test.org", "industry": "Manufacturing", "unit": "cm", "phone": "(577) 625-0848", "signup_date": "2023-05-24"}, {"rate": 0.363, "currency": "BRL", "name": "Mona Cruz", "email": "gapg@mail.com", "industry": "Financial Services", "unit": "cm", "phone": "(577) 012-1803", "signup_date": "2023-11-11"}], "dirty_cols": ["rate", "currency", "name", "email", "industry", "unit", "phone", "signup_date", "extra"], "clean_cols": ["rate", "currency", "name", "email", "industry", "unit", "phone", "signup_date"], "plan": {"dataset_summary": "55 rows × 9 columns. 8 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Boliviano": "BOB", "BOLIVIANO": "BOB", "BRAZILIAN REAL": "BRL", "bob": "BOB", "brl": "BRL", "Bob": "BOB", "braziliann real": "BRL", "boliviano": "BOB", "brazilian real": "BRL", "Bolviano": "BOB", "Brazilian Real": "BRL", "braziilian real": "BRL", "BOOLIVIANO": "BOB", "Brl": "BRL"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"industrial": "Manufacturing", "Telecom": "Telecommunications", "mfg": "Manufacturing", "Fintech": "Financial Services", "Industrial": "Manufacturing", "telco": "Telecommunications", "Telco": "Telecommunications", "Retail & E-commerce": "Retail", "Mfg": "Manufacturing", "Comms": "Telecommunications", "financial services": "Financial Services", "Finance": "Financial Services", "retail": "Retail", "Manufayturing": "Manufacturing", "finance": "Financial Services", "teleocmmunications": "Telecommunications", "INDUSTRIAL": "Manufacturing", "COMSM": "Telecommunications", "teecom": "Telecommunications", "Banking": "Financial Services", "manufacturing": "Manufacturing", "TECLO": "Telecommunications", "MANUAFCTURING": "Manufacturing", "Chmms": "Telecommunications", "FiinServ": "Financial Services", "telecommunications": "Telecommunications", "FINANCIAL SERVICES": "Financial Services", "e-commerce": "Retail", "Finanke": "Financial Services"}, "rationale": "Unified 29 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CENTIMETER": "cm", "IN": "in", "centimeter": "cm", "INCH": "in", "inbhes": "in", "inches": "in", "ich": "in", "Cm": "cm", "centimeters": "cm", "centimmeters": "cm", "inch": "in", "CM": "cm", "Inches": "in", "Inch": "in", "\"": "in", "CENTIMETERS": "cm", "incch": "in", "cntimeters": "cm", "centmieters": "cm", "Centimeter": "cm", "In": "in"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"status": "UNPAID", "currency": "   all", "email": "   gcjieh@mail.com", "phone": "5070682881"}, {"status": "refunded", "currency": "falkland islands pound", "email": " cpam@corp.io ", "phone": "5859567688"}, {"status": "due", "currency": "lek", "email": "GKAPE@EXAMPLE.COM", "phone": "(550)166-7900"}, {"status": "Pending", "currency": "falkland islands pound", "email": "POKL@CORP.IO   ", "phone": "5717398589"}, {"status": "paid", "currency": "  Lek", "email": "cpeaaho@test.org", "phone": "5154840412"}, {"status": " DUE ", "currency": "   Lek", "email": "degfnlk@corp.io", "phone": "547-647-6197"}, {"status": "Paid", "currency": "falkland islands pound", "email": " elbompc@mail.com ", "phone": "555.456.8998"}, {"status": "Pending", "currency": "falkland islands pound", "email": "PDLPNLO@CORP.IO", "phone": "582-113-0233"}, {"status": "paid", "currency": "Falkland Islands Pound", "email": "jjohn@corp.io", "phone": "517-377-4457"}, {"status": "refunded", "currency": "Lek", "email": "mbpc@test.org", "phone": "527.340.8369"}, {"status": "Refund", "currency": "Lek", "email": "enhlf@corp.io", "phone": "519-451-4206"}, {"status": "due   ", "currency": "Falkland Islands Pound", "email": "iaobd@corp.io", "phone": "5412625124"}, {"status": "Refunded", "currency": "Lek", "email": " bgpede@corp.io", "phone": "593-852-3044"}, {"status": "Paid", "currency": "LEK", "email": "COGGA@EXAMPLE.COM", "phone": "547.959.3177"}, {"status": " Refunded ", "currency": "FKP ", "email": "HEFL@CORP.IO", "phone": "(586)790-3096"}, {"status": "  Paid", "currency": "falkland islands pound", "email": "FCCNE@CORP.IO", "phone": "(519)900-6870"}, {"status": " refund ", "currency": "ALL", "email": "PHMM@CORP.IO ", "phone": "5588454041"}, {"status": "refund", "currency": "FKP", "email": "   bbie@corp.io", "phone": "(566)927-3403"}, {"status": "Refunded", "currency": "FKP", "email": "NAKJ@TEST.ORG   ", "phone": "577-800-1195"}, {"status": "unpaid", "currency": "FALKLAND ISLANDS POUND", "email": " dinlnoc@test.org ", "phone": "538-524-9430"}, {"status": "refund", "currency": "LEK ", "email": "kpkk@corp.io", "phone": "(540)332-1610"}, {"status": "paid", "currency": "FKP", "email": "  ndcih@corp.io", "phone": "523.433.1417"}, {"status": "paid", "currency": "  Lek", "email": "cpeaaho@test.org", "phone": "5154840412"}, {"status": "Refunded", "currency": "FKP", "email": "NAKJ@TEST.ORG   ", "phone": "577-800-1195"}, {"status": "", "currency": "", "email": "", "phone": ""}], "clean": [{"status": "Pending", "currency": "ALL", "email": "gcjieh@mail.com", "phone": "(507) 068-2881"}, {"status": "Refunded", "currency": "FKP", "email": "cpam@corp.io", "phone": "(585) 956-7688"}, {"status": "Pending", "currency": "ALL", "email": "gkape@example.com", "phone": "(550) 166-7900"}, {"status": "Pending", "currency": "FKP", "email": "pokl@corp.io", "phone": "(571) 739-8589"}, {"status": "Paid", "currency": "ALL", "email": "cpeaaho@test.org", "phone": "(515) 484-0412"}, {"status": "Pending", "currency": "ALL", "email": "degfnlk@corp.io", "phone": "(547) 647-6197"}, {"status": "Paid", "currency": "FKP", "email": "elbompc@mail.com", "phone": "(555) 456-8998"}, {"status": "Pending", "currency": "FKP", "email": "pdlpnlo@corp.io", "phone": "(582) 113-0233"}, {"status": "Paid", "currency": "FKP", "email": "jjohn@corp.io", "phone": "(517) 377-4457"}, {"status": "Refunded", "currency": "ALL", "email": "mbpc@test.org", "phone": "(527) 340-8369"}, {"status": "Refunded", "currency": "ALL", "email": "enhlf@corp.io", "phone": "(519) 451-4206"}, {"status": "Pending", "currency": "FKP", "email": "iaobd@corp.io", "phone": "(541) 262-5124"}, {"status": "Refunded", "currency": "ALL", "email": "bgpede@corp.io", "phone": "(593) 852-3044"}, {"status": "Paid", "currency": "ALL", "email": "cogga@example.com", "phone": "(547) 959-3177"}, {"status": "Refunded", "currency": "FKP", "email": "hefl@corp.io", "phone": "(586) 790-3096"}, {"status": "Paid", "currency": "FKP", "email": "fccne@corp.io", "phone": "(519) 900-6870"}, {"status": "Refunded", "currency": "ALL", "email": "phmm@corp.io", "phone": "(558) 845-4041"}, {"status": "Refunded", "currency": "FKP", "email": "bbie@corp.io", "phone": "(566) 927-3403"}, {"status": "Refunded", "currency": "FKP", "email": "nakj@test.org", "phone": "(577) 800-1195"}, {"status": "Pending", "currency": "FKP", "email": "dinlnoc@test.org", "phone": "(538) 524-9430"}, {"status": "Refunded", "currency": "ALL", "email": "kpkk@corp.io", "phone": "(540) 332-1610"}, {"status": "Paid", "currency": "FKP", "email": "ndcih@corp.io", "phone": "(523) 433-1417"}], "dirty_cols": ["status", "currency", "email", "phone"], "clean_cols": ["status", "currency", "email", "phone"], "plan": {"dataset_summary": "25 rows × 4 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"UNPAID": "Pending", "refunded": "Refunded", "due": "Pending", "paid": "Paid", "DUE": "Pending", "Refund": "Refunded", "refund": "Refunded", "unpaid": "Pending"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"all": "ALL", "falkland islands pound": "FKP", "lek": "ALL", "Lek": "ALL", "Falkland Islands Pound": "FKP", "LEK": "ALL", "FALKLAND ISLANDS POUND": "FKP"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"amount": "N/A", "company": "   Vehement", "state": "RI", "job_title": "Dir", "is_active": "No"}, {"amount": "2.397,96", "company": "Soylent Corp", "state": " Rhode Ilsand", "job_title": "Director", "is_active": "N"}, {"amount": "$1,565.50", "company": "Cyberdyne  ", "state": "Rhode Island", "job_title": "ADMIN", "is_active": "0"}, {"amount": "4.120,24", "company": "  Globex", "state": "tennessee", "job_title": "Chief Technology Officer", "is_active": "N"}, {"amount": "TBD", "company": "Wonka Co", "state": "CA", "job_title": "administrative assistant", "is_active": "N"}, {"amount": "5.027,72", "company": "Umbrella   ", "state": "ut", "job_title": "Admin", "is_active": "true"}, {"amount": "3.342,03", "company": "Wonka Co", "state": "rhode island", "job_title": "Admin Assistant", "is_active": "0"}, {"amount": "$7,661.30", "company": " Vehement ", "state": "mn", "job_title": "Chief Technology Officer", "is_active": "T"}, {"amount": "7.788,88", "company": " Globex ", "state": "Rhode Issand ", "job_title": "CTO", "is_active": "F"}, {"amount": "N/A", "company": "Globex", "state": "ca", "job_title": "Admin Asst", "is_active": "N"}, {"amount": "$5,382.83", "company": "Stark Industries", "state": "mn ", "job_title": "admin   ", "is_active": "No"}, {"amount": "2.317,29", "company": "Hooli", "state": "mn   ", "job_title": "admin", "is_active": "N"}, {"amount": "5.960,67", "company": "Globex", "state": " california ", "job_title": "chief technology officer", "is_active": "No"}, {"amount": "TBD", "company": "Cyberdyne  ", "state": "MN", "job_title": "CTO", "is_active": "No"}, {"amount": "None", "company": "Cyberdyne  ", "state": "Minnesota", "job_title": "dir   ", "is_active": "TRUE"}, {"amount": "4.272,45", "company": " Cyberdyne ", "state": "Mn", "job_title": "Dir", "is_active": "Y"}, {"amount": "na", "company": "   Initech", "state": "ca", "job_title": "Admin Asst", "is_active": "0"}, {"amount": "$58.47", "company": "Acme Inc   ", "state": "Tennessee", "job_title": "director", "is_active": "false"}, {"amount": "$2,752.85", "company": "   Globex", "state": "MINNESOTA", "job_title": "Chief Technology Officer ", "is_active": "Y"}, {"amount": "$3,354.16", "company": "Vehement", "state": "MINNESOOTA", "job_title": "  Admin Assistant", "is_active": "Yes"}, {"amount": "4.154,24", "company": "Hooli  ", "state": "minnesota", "job_title": "AdminAssistant", "is_active": "T"}, {"amount": "659,24", "company": "Stark Industries  ", "state": "tennessee", "job_title": "CTO", "is_active": "1"}, {"amount": "$4,839.55", "company": "  Stark Industries", "state": "Minnesota", "job_title": "  chief technology officer", "is_active": "1"}, {"amount": "7.563,59", "company": "Hooli", "state": "California", "job_title": "  Admiin", "is_active": "1"}, {"amount": "1.959,43", "company": "Wonka  Co", "state": "ut", "job_title": "Cto", "is_active": "N"}, {"amount": "1.943,35", "company": "Vehement", "state": "MN", "job_title": " dir. ", "is_active": "Y"}, {"amount": "2.718,22", "company": "Acme Inc", "state": "CA", "job_title": "c.t.o.", "is_active": "N"}, {"amount": "1.171,13", "company": "Vehement   ", "state": "UT", "job_title": "Administrative Assistant", "is_active": "false"}, {"amount": "$4,218.74", "company": "Vehement", "state": "TN", "job_title": "Admin", "is_active": "T"}, {"amount": "551,39", "company": "Stark Industries", "state": "california", "job_title": "admin  ", "is_active": "N"}, {"amount": "", "company": "", "state": "", "job_title": "", "is_active": ""}], "clean": [{"amount": NaN, "company": "Vehement", "state": "Rhode Island", "job_title": "Director", "is_active": false}, {"amount": 2397.96, "company": "Soylent Corp", "state": "Rhode Island", "job_title": "Director", "is_active": false}, {"amount": 1565.5, "company": "Cyberdyne", "state": "Rhode Island", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 4120.24, "company": "Globex", "state": "Tennessee", "job_title": "Chief Technology Officer", "is_active": false}, {"amount": NaN, "company": "Wonka Co", "state": "California", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 5027.72, "company": "Umbrella", "state": "Utah", "job_title": "Administrative Assistant", "is_active": true}, {"amount": 3342.03, "company": "Wonka Co", "state": "Rhode Island", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 7661.3, "company": "Vehement", "state": "Minnesota", "job_title": "Chief Technology Officer", "is_active": true}, {"amount": 7788.88, "company": "Globex", "state": "Rhode Island", "job_title": "Chief Technology Officer", "is_active": false}, {"amount": NaN, "company": "Globex", "state": "California", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 5382.83, "company": "Stark Industries", "state": "Minnesota", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 2317.29, "company": "Hooli", "state": "Minnesota", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 5960.67, "company": "Globex", "state": "California", "job_title": "Chief Technology Officer", "is_active": false}, {"amount": NaN, "company": "Cyberdyne", "state": "Minnesota", "job_title": "Chief Technology Officer", "is_active": false}, {"amount": NaN, "company": "Cyberdyne", "state": "Minnesota", "job_title": "Director", "is_active": true}, {"amount": 4272.45, "company": "Cyberdyne", "state": "Minnesota", "job_title": "Director", "is_active": true}, {"amount": NaN, "company": "Initech", "state": "California", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 58.47, "company": "Acme Inc", "state": "Tennessee", "job_title": "Director", "is_active": false}, {"amount": 2752.85, "company": "Globex", "state": "Minnesota", "job_title": "Chief Technology Officer", "is_active": true}, {"amount": 3354.16, "company": "Vehement", "state": "Minnesota", "job_title": "Administrative Assistant", "is_active": true}, {"amount": 4154.24, "company": "Hooli", "state": "Minnesota", "job_title": "Administrative Assistant", "is_active": true}, {"amount": 659.24, "company": "Stark Industries", "state": "Tennessee", "job_title": "Chief Technology Officer", "is_active": true}, {"amount": 4839.55, "company": "Stark Industries", "state": "Minnesota", "job_title": "Chief Technology Officer", "is_active": true}, {"amount": 7563.59, "company": "Hooli", "state": "California", "job_title": "Administrative Assistant", "is_active": true}, {"amount": 1959.43, "company": "Wonka Co", "state": "Utah", "job_title": "Chief Technology Officer", "is_active": false}, {"amount": 1943.35, "company": "Vehement", "state": "Minnesota", "job_title": "Director", "is_active": true}, {"amount": 2718.22, "company": "Acme Inc", "state": "California", "job_title": "Chief Technology Officer", "is_active": false}, {"amount": 1171.13, "company": "Vehement", "state": "Utah", "job_title": "Administrative Assistant", "is_active": false}, {"amount": 4218.74, "company": "Vehement", "state": "Tennessee", "job_title": "Administrative Assistant", "is_active": true}, {"amount": 551.39, "company": "Stark Industries", "state": "California", "job_title": "Administrative Assistant", "is_active": false}], "dirty_cols": ["amount", "company", "state", "job_title", "is_active"], "clean_cols": ["amount", "company", "state", "job_title", "is_active"], "plan": {"dataset_summary": "31 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"RI": "Rhode Island", "Rhode Ilsand": "Rhode Island", "tennessee": "Tennessee", "CA": "California", "ut": "Utah", "rhode island": "Rhode Island", "mn": "Minnesota", "Rhode Issand": "Rhode Island", "ca": "California", "california": "California", "MN": "Minnesota", "Mn": "Minnesota", "MINNESOTA": "Minnesota", "MINNESOOTA": "Minnesota", "minnesota": "Minnesota", "UT": "Utah", "TN": "Tennessee"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dir": "Director", "ADMIN": "Administrative Assistant", "administrative assistant": "Administrative Assistant", "Admin": "Administrative Assistant", "Admin Assistant": "Administrative Assistant", "CTO": "Chief Technology Officer", "Admin Asst": "Administrative Assistant", "admin": "Administrative Assistant", "chief technology officer": "Chief Technology Officer", "dir": "Director", "director": "Director", "AdminAssistant": "Administrative Assistant", "Admiin": "Administrative Assistant", "Cto": "Chief Technology Officer", "dir.": "Director", "c.t.o.": "Chief Technology Officer"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"currency": "BMD", "amount": "0", "city": "Chkclayo", "country": "ICELAND ", "job_title": "Engineering Mgr", "company": "Vehement", "state": "UM", "phone": "509.307.9390"}, {"currency": "Egyptian Pound", "amount": "None", "city": "BLLHEAD CITY", "country": "isl", "job_title": "Dbr.", "company": " Globex", "state": "Hi", "phone": "(590)207-0329"}, {"currency": "brunei dollar", "amount": "$766.95", "city": "Bullhead City", "country": " is ", "job_title": "Enginering Manager", "company": "Umbrella", "state": "Hawaii", "phone": "545-209-4617"}, {"currency": "BND  ", "amount": "3.284,50", "city": " Bullhead City", "country": "CAN", "job_title": "sr.  engineer", "company": "Soylent Corp", "state": "HI", "phone": "572.047.5810"}, {"currency": "egyptian pound", "amount": "$8,740.26", "city": "Bullhead City", "country": "CAN", "job_title": "dir.", "company": "Initech", "state": "ar", "phone": "517.990.4014"}, {"currency": "Egyptian Pound", "amount": "$5,568.46", "city": "Chilcayo", "country": "Iceland", "job_title": "sr. engineer", "company": " Stark Industries", "state": "HI  ", "phone": "582-418-7760"}, {"currency": "BMD  ", "amount": "$1,117.02", "city": "bulllhead city", "country": "Iceland", "job_title": "Eng Manager", "company": "  Hooli", "state": "HI", "phone": "506.649.7898"}, {"currency": "egyptian pound", "amount": "929,63", "city": "chiclayo", "country": "Republic of Iceland", "job_title": "eng manager", "company": "Wonka Co", "state": "um", "phone": "(548)690-9992"}, {"currency": "egp", "amount": "$212.51", "city": "Bullhead City", "country": "Canada  ", "job_title": "Engineering Manager", "company": "  Stark Industries", "state": "United States Minor  Outyling Islands", "phone": "535-974-0008"}, {"currency": "Brunei Dollar", "amount": "8.076,32", "city": "BULLHEAD  CITY", "country": "CAN", "job_title": "Dir", "company": "Acme Inc", "state": "akansas", "phone": "524-191-8987"}, {"currency": "EGP", "amount": "N/A", "city": "BULLHEAD CITY", "country": "Canaa", "job_title": "Sr Engineer", "company": "Stark Industries ", "state": "HI", "phone": "5040578213"}, {"currency": "Bermudian Dollar", "amount": "7.361,02", "city": "Chiclayo", "country": " is ", "job_title": "Eng  Manager", "company": " Wonka Co", "state": "Um", "phone": "583-244-4994"}, {"currency": "EGP", "amount": "7.925,85", "city": "bullhead city", "country": "   canada", "job_title": "Sr Etgineer", "company": "Vehement", "state": "UM", "phone": "570-622-9184"}, {"currency": "Egyptian Pound", "amount": "3.084,53", "city": "Chiclayo", "country": "CA", "job_title": "   Eng Mnager", "company": "Hooli", "state": "HI", "phone": "514.646.7617"}, {"currency": "   egyptian pound", "amount": "1.415,65", "city": "chiclayo", "country": "IS", "job_title": "Dir", "company": "Soylent  Corp", "state": "hi", "phone": "5605114149"}, {"currency": "Bermudian Dollar", "amount": "$6,202.43", "city": "Bullhead City", "country": "isl", "job_title": "senior eng", "company": "Initech", "state": "United States Minor Outlying Islands", "phone": "(589)765-8696"}, {"currency": "egp", "amount": "736,09", "city": "CHICLAYO", "country": "IS", "job_title": "   senior eng", "company": "Acme Inc", "state": "AR", "phone": "582-627-8925"}, {"currency": "EGP", "amount": "$1,023.31", "city": "   Bullhead City", "country": "republic of iceland", "job_title": "ENGINEERING MANAGER", "company": "Hooli  ", "state": "UM ", "phone": "5849771478"}, {"currency": "BND", "amount": "8.485,30", "city": "Bullhead Uity ", "country": "ISL", "job_title": "SENIOR ENGINEER", "company": "  Stark Industries", "state": "AR", "phone": "(512)677-5250"}, {"currency": "Brunei Dollar", "amount": "$456.09", "city": "CHICLAYO", "country": "isl", "job_title": "Director", "company": " Umbrella", "state": "arkansas  ", "phone": "(572)183-7189"}, {"currency": "bermudian dollar", "amount": "5.247,16", "city": " bullhlad city", "country": "Is", "job_title": "dir", "company": "Initech", "state": "Arkdnsas ", "phone": "557.730.7034"}, {"currency": "bnd", "amount": "$4,478.95", "city": "Chiclayo", "country": "is", "job_title": "Snr Engineer", "company": " Cyberdyne ", "state": "United States Minor Outlying Islands", "phone": "5273035829"}, {"currency": "Egp", "amount": "$6,242.33", "city": "bullhead city", "country": "can", "job_title": "Eng Mgr", "company": "Hooli", "state": "arkasnas", "phone": "5270655343"}, {"currency": "bermudiaan dollar  ", "amount": "TBD", "city": "Chiclayo", "country": "CAN", "job_title": "Senior  Eng", "company": "Umbrella", "state": "  Hawaii", "phone": "(519)398-0830"}, {"currency": "bermudian dollar", "amount": "3.864,95", "city": "Chiclayo", "country": "canada", "job_title": "DIRECWOR", "company": "Hooli", "state": "United States Minor Outlying Islands", "phone": "552-168-0279"}, {"currency": "BND", "amount": "6.600,69", "city": "bullhead city", "country": "ISL", "job_title": "Eng Manager", "company": "Umbrella", "state": "hi", "phone": "510-350-4988"}, {"currency": "Brunei Dollar", "amount": "$1,703.32", "city": "chiclayo", "country": "Canada", "job_title": " engineering mgr", "company": "Vehement   ", "state": "UM", "phone": "(533)608-1236"}, {"currency": "Brunei Dollar", "amount": "$456.09", "city": "CHICLAYO", "country": "isl", "job_title": "Director", "company": " Umbrella", "state": "arkansas  ", "phone": "(572)183-7189"}, {"currency": "Egyptian Pound", "amount": "3.084,53", "city": "Chiclayo", "country": "CA", "job_title": "   Eng Mnager", "company": "Hooli", "state": "HI", "phone": "514.646.7617"}, {"currency": "", "amount": "", "city": "", "country": "", "job_title": "", "company": "", "state": "", "phone": ""}], "clean": [{"currency": "BMD", "amount": 0.0, "city": "Chiclayo", "country": "Iceland", "job_title": "Engineering Manager", "company": "Vehement", "state": "United States Minor Outlying Islands", "phone": "(509) 307-9390"}, {"currency": "EGP", "amount": NaN, "city": "Bullhead City", "country": "Iceland", "job_title": "Director", "company": "Globex", "state": "Hawaii", "phone": "(590) 207-0329"}, {"currency": "BND", "amount": 766.95, "city": "Bullhead City", "country": "Iceland", "job_title": "Engineering Manager", "company": "Umbrella", "state": "Hawaii", "phone": "(545) 209-4617"}, {"currency": "BND", "amount": 3284.5, "city": "Bullhead City", "country": "Canada", "job_title": "Senior Engineer", "company": "Soylent Corp", "state": "Hawaii", "phone": "(572) 047-5810"}, {"currency": "EGP", "amount": 8740.26, "city": "Bullhead City", "country": "Canada", "job_title": "Director", "company": "Initech", "state": "Arkansas", "phone": "(517) 990-4014"}, {"currency": "EGP", "amount": 5568.46, "city": "Chiclayo", "country": "Iceland", "job_title": "Senior Engineer", "company": "Stark Industries", "state": "Hawaii", "phone": "(582) 418-7760"}, {"currency": "BMD", "amount": 1117.02, "city": "Bullhead City", "country": "Iceland", "job_title": "Engineering Manager", "company": "Hooli", "state": "Hawaii", "phone": "(506) 649-7898"}, {"currency": "EGP", "amount": 929.63, "city": "Chiclayo", "country": "Iceland", "job_title": "Engineering Manager", "company": "Wonka Co", "state": "United States Minor Outlying Islands", "phone": "(548) 690-9992"}, {"currency": "EGP", "amount": 212.51, "city": "Bullhead City", "country": "Canada", "job_title": "Engineering Manager", "company": "Stark Industries", "state": "United States Minor Outlying Islands", "phone": "(535) 974-0008"}, {"currency": "BND", "amount": 8076.32, "city": "Bullhead City", "country": "Canada", "job_title": "Director", "company": "Acme Inc", "state": "Arkansas", "phone": "(524) 191-8987"}, {"currency": "EGP", "amount": NaN, "city": "Bullhead City", "country": "Canada", "job_title": "Senior Engineer", "company": "Stark Industries", "state": "Hawaii", "phone": "(504) 057-8213"}, {"currency": "BMD", "amount": 7361.02, "city": "Chiclayo", "country": "Iceland", "job_title": "Engineering Manager", "company": "Wonka Co", "state": "United States Minor Outlying Islands", "phone": "(583) 244-4994"}, {"currency": "EGP", "amount": 7925.85, "city": "Bullhead City", "country": "Canada", "job_title": "Senior Engineer", "company": "Vehement", "state": "United States Minor Outlying Islands", "phone": "(570) 622-9184"}, {"currency": "EGP", "amount": 3084.53, "city": "Chiclayo", "country": "Canada", "job_title": "Engineering Manager", "company": "Hooli", "state": "Hawaii", "phone": "(514) 646-7617"}, {"currency": "EGP", "amount": 1415.65, "city": "Chiclayo", "country": "Iceland", "job_title": "Director", "company": "Soylent Corp", "state": "Hawaii", "phone": "(560) 511-4149"}, {"currency": "BMD", "amount": 6202.43, "city": "Bullhead City", "country": "Iceland", "job_title": "Senior Engineer", "company": "Initech", "state": "United States Minor Outlying Islands", "phone": "(589) 765-8696"}, {"currency": "EGP", "amount": 736.09, "city": "Chiclayo", "country": "Iceland", "job_title": "Senior Engineer", "company": "Acme Inc", "state": "Arkansas", "phone": "(582) 627-8925"}, {"currency": "EGP", "amount": 1023.31, "city": "Bullhead City", "country": "Iceland", "job_title": "Engineering Manager", "company": "Hooli", "state": "United States Minor Outlying Islands", "phone": "(584) 977-1478"}, {"currency": "BND", "amount": 8485.3, "city": "Bullhead City", "country": "Iceland", "job_title": "Senior Engineer", "company": "Stark Industries", "state": "Arkansas", "phone": "(512) 677-5250"}, {"currency": "BND", "amount": 456.09, "city": "Chiclayo", "country": "Iceland", "job_title": "Director", "company": "Umbrella", "state": "Arkansas", "phone": "(572) 183-7189"}, {"currency": "BMD", "amount": 5247.16, "city": "Bullhead City", "country": "Iceland", "job_title": "Director", "company": "Initech", "state": "Arkansas", "phone": "(557) 730-7034"}, {"currency": "BND", "amount": 4478.95, "city": "Chiclayo", "country": "Iceland", "job_title": "Senior Engineer", "company": "Cyberdyne", "state": "United States Minor Outlying Islands", "phone": "(527) 303-5829"}, {"currency": "EGP", "amount": 6242.33, "city": "Bullhead City", "country": "Canada", "job_title": "Engineering Manager", "company": "Hooli", "state": "Arkansas", "phone": "(527) 065-5343"}, {"currency": "BMD", "amount": NaN, "city": "Chiclayo", "country": "Canada", "job_title": "Senior Engineer", "company": "Umbrella", "state": "Hawaii", "phone": "(519) 398-0830"}, {"currency": "BMD", "amount": 3864.95, "city": "Chiclayo", "country": "Canada", "job_title": "Director", "company": "Hooli", "state": "United States Minor Outlying Islands", "phone": "(552) 168-0279"}, {"currency": "BND", "amount": 6600.69, "city": "Bullhead City", "country": "Iceland", "job_title": "Engineering Manager", "company": "Umbrella", "state": "Hawaii", "phone": "(510) 350-4988"}, {"currency": "BND", "amount": 1703.32, "city": "Chiclayo", "country": "Canada", "job_title": "Engineering Manager", "company": "Vehement", "state": "United States Minor Outlying Islands", "phone": "(533) 608-1236"}], "dirty_cols": ["currency", "amount", "city", "country", "job_title", "company", "state", "phone"], "clean_cols": ["currency", "amount", "city", "country", "job_title", "company", "state", "phone"], "plan": {"dataset_summary": "30 rows × 8 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Egyptian Pound": "EGP", "brunei dollar": "BND", "egyptian pound": "EGP", "egp": "EGP", "Brunei Dollar": "BND", "Bermudian Dollar": "BMD", "bermudian dollar": "BMD", "bnd": "BND", "Egp": "EGP", "bermudiaan dollar": "BMD"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Chkclayo": "Chiclayo", "BLLHEAD CITY": "Bullhead City", "Chilcayo": "Chiclayo", "bulllhead city": "Bullhead City", "chiclayo": "Chiclayo", "BULLHEAD  CITY": "Bullhead City", "BULLHEAD CITY": "Bullhead City", "bullhead city": "Bullhead City", "CHICLAYO": "Chiclayo", "Bullhead Uity": "Bullhead City", "bullhlad city": "Bullhead City"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ICELAND": "Iceland", "isl": "Iceland", "is": "Iceland", "CAN": "Canada", "Republic of Iceland": "Iceland", "Canaa": "Canada", "canada": "Canada", "CA": "Canada", "IS": "Iceland", "republic of iceland": "Iceland", "ISL": "Iceland", "Is": "Iceland", "can": "Canada"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Engineering Mgr": "Engineering Manager", "Dbr.": "Director", "Enginering Manager": "Engineering Manager", "sr. engineer": "Senior Engineer", "dir.": "Director", "Eng Manager": "Engineering Manager", "eng manager": "Engineering Manager", "Dir": "Director", "Sr Engineer": "Senior Engineer", "Sr Etgineer": "Senior Engineer", "Eng Mnager": "Engineering Manager", "senior eng": "Senior Engineer", "ENGINEERING MANAGER": "Engineering Manager", "SENIOR ENGINEER": "Senior Engineer", "dir": "Director", "Snr Engineer": "Senior Engineer", "Eng Mgr": "Engineering Manager", "Senior Eng": "Senior Engineer", "DIRECWOR": "Director", "engineering mgr": "Engineering Manager"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"UM": "United States Minor Outlying Islands", "Hi": "Hawaii", "HI": "Hawaii", "ar": "Arkansas", "um": "United States Minor Outlying Islands", "United States Minor Outyling Islands": "United States Minor Outlying Islands", "akansas": "Arkansas", "Um": "United States Minor Outlying Islands", "hi": "Hawaii", "AR": "Arkansas", "arkansas": "Arkansas", "Arkdnsas": "Arkansas", "arkasnas": "Arkansas"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"signup_date": "2023-06-09", "city": "dayal", "name": "Ivan Brandt   ", "phone": "5805261671", "country": "French Polynesia", "notes2": ""}, {"signup_date": "45147", "city": "Tapejara", "name": "Eve  Khan", "phone": "5488864760", "country": "pf ", "notes2": ""}, {"signup_date": "44973", "city": "Tapejara", "name": "Heidi Park  ", "phone": "568.323.8273", "country": "Pf", "notes2": ""}, {"signup_date": "24 Apr 2023", "city": "Dayal", "name": "Karl Wong  ", "phone": "(563)571-5028", "country": "French Polynesia", "notes2": ""}, {"signup_date": "5 Mar 2023", "city": "Dayal", "name": "Alice  Khan", "phone": "511.301.6808", "country": "LV", "notes2": ""}, {"signup_date": "44998", "city": "   daayal", "name": "Priya  Petrov", "phone": "(580)924-9828", "country": "GT", "notes2": ""}, {"signup_date": "45209", "city": "TAPEJARA  ", "name": "Judy  Park", "phone": "(562)096-4323", "country": "pyf", "notes2": ""}, {"signup_date": "44986", "city": "Dayal", "name": "Ivan Smith", "phone": "5150348794", "country": "GT", "notes2": ""}, {"signup_date": "5 Apr 2023", "city": "Tapejara", "name": "Mona Lee", "phone": "594-117-9609", "country": "pyf", "notes2": ""}, {"signup_date": "45110", "city": "Tapejara", "name": "  Lena Fischer", "phone": "599.044.4547", "country": "French Polynesia ", "notes2": ""}, {"signup_date": "45003", "city": "Dayal", "name": "Carol Smith ", "phone": "(543)004-1212", "country": "VCT", "notes2": ""}, {"signup_date": "8/16/2023", "city": "dayal", "name": "  David Reyes", "phone": "(555)193-7176", "country": "saint vincent and the grenadines", "notes2": ""}, {"signup_date": "2023-08-19", "city": "DAYAL", "name": "Omar Adams", "phone": "(513)042-2053", "country": "VCT", "notes2": ""}, {"signup_date": "5 Mar 2023", "city": "Dayal", "name": "Alice  Khan", "phone": "511.301.6808", "country": "LV", "notes2": ""}], "clean": [{"signup_date": "2023-06-09", "city": "Dayal", "name": "Ivan Brandt", "phone": "(580) 526-1671", "country": "French Polynesia"}, {"signup_date": "2023-08-09", "city": "Tapejara", "name": "Eve Khan", "phone": "(548) 886-4760", "country": "French Polynesia"}, {"signup_date": "2023-02-16", "city": "Tapejara", "name": "Heidi Park", "phone": "(568) 323-8273", "country": "French Polynesia"}, {"signup_date": "2023-04-24", "city": "Dayal", "name": "Karl Wong", "phone": "(563) 571-5028", "country": "French Polynesia"}, {"signup_date": "2023-03-05", "city": "Dayal", "name": "Alice Khan", "phone": "(511) 301-6808", "country": "Latvia"}, {"signup_date": "2023-03-13", "city": "Dayal", "name": "Priya Petrov", "phone": "(580) 924-9828", "country": "Guatemala"}, {"signup_date": "2023-10-10", "city": "Tapejara", "name": "Judy Park", "phone": "(562) 096-4323", "country": "French Polynesia"}, {"signup_date": "2023-03-01", "city": "Dayal", "name": "Ivan Smith", "phone": "(515) 034-8794", "country": "Guatemala"}, {"signup_date": "2023-04-05", "city": "Tapejara", "name": "Mona Lee", "phone": "(594) 117-9609", "country": "French Polynesia"}, {"signup_date": "2023-07-03", "city": "Tapejara", "name": "Lena Fischer", "phone": "(599) 044-4547", "country": "French Polynesia"}, {"signup_date": "2023-03-18", "city": "Dayal", "name": "Carol Smith", "phone": "(543) 004-1212", "country": "Saint Vincent and the Grenadines"}, {"signup_date": "2023-08-16", "city": "Dayal", "name": "David Reyes", "phone": "(555) 193-7176", "country": "Saint Vincent and the Grenadines"}, {"signup_date": "2023-08-19", "city": "Dayal", "name": "Omar Adams", "phone": "(513) 042-2053", "country": "Saint Vincent and the Grenadines"}], "dirty_cols": ["signup_date", "city", "name", "phone", "country", "notes2"], "clean_cols": ["signup_date", "city", "name", "phone", "country"], "plan": {"dataset_summary": "14 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"dayal": "Dayal", "daayal": "Dayal", "TAPEJARA": "Tapejara", "DAYAL": "Dayal"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pf": "French Polynesia", "Pf": "French Polynesia", "LV": "Latvia", "GT": "Guatemala", "pyf": "French Polynesia", "VCT": "Saint Vincent and the Grenadines", "saint vincent and the grenadines": "Saint Vincent and the Grenadines"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"unit": "  kilograms", "email": "IIDLGL@TEST.ORG  ", "status": "Medium ", "state": "Minnesota", "city": "Prague", "amount": "1.532,79", "is_active": "Y", "country": "   republic of malawi", "extra": ""}, {"unit": "CM", "email": "LONEEFG@MAIL.COM", "status": "  m", "state": "VA", "city": "monterrey   ", "amount": "?", "is_active": "Yes", "country": "ALB", "extra": ""}, {"unit": "  centimeters", "email": "bddknb@example.com", "status": "m", "state": " Virginia ", "city": "Prague", "amount": "177,01", "is_active": "Y", "country": "CC", "extra": ""}, {"unit": "kg  ", "email": "EIEF@EXAMPLE.COM ", "status": "P2", "state": " mn ", "city": "Pont eVedra Beach", "amount": "7.647,59", "is_active": "T", "country": "malawi", "extra": ""}, {"unit": "kilograms", "email": "cdkfje@mail.com ", "status": "m   ", "state": "Colorado", "city": "Ormoc", "amount": "$5,783.00", "is_active": "F", "country": "NZ", "extra": ""}, {"unit": "KG", "email": "EAPJDFN@MAIL.COM", "status": "High", "state": "Connecticut", "city": "PRAGUE", "amount": "#N/A", "is_active": "FALSE", "country": "Bouvet Island", "extra": ""}, {"unit": "Ceentimeter", "email": "NGLJANG@EXAMPLE.COM", "status": "MED   ", "state": "Minnesota", "city": "ORMOC", "amount": "$6,545.33", "is_active": "false", "country": "MW", "extra": ""}, {"unit": " Kg ", "email": " CILHO@MAIL.COM", "status": "med", "state": " minnesota", "city": " Ormoc ", "amount": "$3,476.42", "is_active": "Y", "country": "Republic of Malawi", "extra": ""}, {"unit": "kilogrcms", "email": "HIJIB@TEST.ORG", "status": "p1   ", "state": "  co", "city": " monterrey ", "amount": "8,876", "is_active": "0", "country": "CC", "extra": ""}, {"unit": " KG ", "email": "JGKLP@CORP.IO", "status": "H", "state": "CO", "city": "Ormoc", "amount": "8.085,85", "is_active": "true", "country": "Cocos Keeling Islands", "extra": ""}, {"unit": " centimeter ", "email": "pkkpakp@mail.com", "status": "Medium", "state": "Wyoming  ", "city": "  monterrey", "amount": "$2,088.81", "is_active": "1", "country": "albvnia", "extra": ""}, {"unit": "Kg", "email": "mljiph@test.org", "status": "P1", "state": "Connecticut", "city": "   monterrey", "amount": "$7,023.50", "is_active": "T", "country": " MWI ", "extra": ""}, {"unit": "kilograms", "email": " AGEEEKA@CORP.IO", "status": "  HIGH", "state": "Minnesota", "city": "Ormoc", "amount": "4.468,51", "is_active": "No", "country": "REPUBLIC OF MALAWI", "extra": ""}, {"unit": "centimeters", "email": "nocbhj@example.com", "status": " H", "state": "VA", "city": "Prague", "amount": "1.475,19", "is_active": "T", "country": " BVT ", "extra": ""}, {"unit": "Kg ", "email": " oikonhg@mail.com ", "status": "high", "state": "Minnesota", "city": "ORMOC", "amount": "$1,734.35", "is_active": "0", "country": "BVT", "extra": ""}, {"unit": "kilograms", "email": "   MLLLFB@EXAMPLE.COM", "status": "medium", "state": "   Wy", "city": " ormoc", "amount": "5.225,60", "is_active": "F", "country": "bvt", "extra": ""}, {"unit": "CENTIMETER", "email": "   FIFLCA@EXAMPLE.COM", "status": "HIGH ", "state": "Minnesota", "city": "Monterrey", "amount": "$7,408.11", "is_active": "0", "country": "New Zealand", "extra": ""}, {"unit": "KILOGRAMS", "email": "  amdc@mail.com", "status": "med ", "state": "minnesota", "city": "Ponte Vedra Beach", "amount": "452,47", "is_active": "T", "country": "Albania", "extra": ""}, {"unit": "KILOGRAMS  ", "email": "OJLPIEP@CORP.IO", "status": "Medium", "state": "MN", "city": "Monterrey", "amount": "$7,262.33", "is_active": "1", "country": "Malawi", "extra": ""}, {"unit": "cm", "email": "OMHF@EXAMPLE.COM", "status": "HIGH", "state": "MINNESOTA", "city": "Monterrey", "amount": "TBD", "is_active": "No", "country": " BVT ", "extra": ""}, {"unit": "CENTIMETER ", "email": " jkgcic@test.org", "status": "P1", "state": "Virginia  ", "city": "ponte vedra beach ", "amount": "1.644,98", "is_active": "Y", "country": "republic of malawi", "extra": ""}, {"unit": "kg", "email": "egaie@corp.io", "status": " H ", "state": "virginia", "city": "Ponte Vedra Beach", "amount": "4.341,25", "is_active": "TRUE", "country": "CCK", "extra": ""}, {"unit": "centimeter", "email": "DEHDJ@MAIL.COM", "status": "high", "state": "   Colorado", "city": " Ormoc ", "amount": "TBD", "is_active": "FALSE", "country": "Malawi", "extra": ""}, {"unit": "cm", "email": " KHLEHK@TEST.ORG", "status": "H", "state": "MN", "city": " MONTERREY", "amount": "7.817,76", "is_active": "true", "country": "COCOS (KEELING) ISLANDS", "extra": ""}, {"unit": "KG", "email": "ciehka@corp.io", "status": "higgh", "state": "MN", "city": "ponte vedra beach", "amount": "$6,887.56", "is_active": "1", "country": "BVT ", "extra": ""}, {"unit": "centimeter", "email": "KPJKFBJ@CORP.IO", "status": "h", "state": "  Connecticut", "city": "Ormoc", "amount": "$2,122.02", "is_active": "TRUE", "country": "alb", "extra": ""}, {"unit": "kg", "email": "ialbdli@test.org  ", "status": "MEDIUM", "state": "ct", "city": "Prague", "amount": "2.458,64", "is_active": "1", "country": "CCK", "extra": ""}, {"unit": "Kg   ", "email": "blfima@mail.com", "status": "p2", "state": "co", "city": "PonteV edra Beach   ", "amount": "$1,652.27", "is_active": "Y", "country": "Republic of Albania", "extra": ""}, {"unit": "cm", "email": "   NHNBNO@MAIL.COM", "status": "P1", "state": "Co", "city": "ormoc", "amount": "4.044,22", "is_active": "No", "country": "Bv", "extra": ""}, {"unit": "kg", "email": "emhhpob@test.org", "status": "High", "state": "Wyoming", "city": "Prague", "amount": "$6,468.26", "is_active": "FALSE", "country": "bv", "extra": ""}, {"unit": "centimeter", "email": "lofnk@mail.com", "status": "High", "state": "WY", "city": "Ormoc", "amount": "2.211,49", "is_active": "FALSE", "country": "Nwe Zealand", "extra": ""}, {"unit": "centimeers", "email": "FCONFCB@MAIL.COM", "status": "h", "state": "  ct", "city": "Ponte Vedra Beach  ", "amount": "?", "is_active": "0", "country": "republic of malawi", "extra": ""}, {"unit": "kg", "email": " CMJF@MAIL.COM ", "status": "P1", "state": "MINNESOTA", "city": "ponte vedra beach", "amount": "$8,275.98", "is_active": "true", "country": "  CCK", "extra": ""}, {"unit": "centimater", "email": "geecj@mail.com ", "status": "Medium   ", "state": "va", "city": "Ponte Vedra Beach", "amount": "3.057,70", "is_active": "T", "country": "NZ", "extra": ""}, {"unit": "KILOGRAM", "email": "aplfje@example.com", "status": "P1   ", "state": "WY", "city": "Ormoc", "amount": "7.256,77", "is_active": "FALSE", "country": "Bouvet  Island", "extra": ""}, {"unit": " Cm ", "email": "FMBLI@EXAMPLE.COM", "status": "med", "state": "Colorado", "city": "Munterrey", "amount": "$102.73", "is_active": "Yes", "country": "Malwi", "extra": ""}, {"unit": "kg", "email": " nignc@mail.com ", "status": " higgh ", "state": "VA", "city": "Prague", "amount": "2.726,09", "is_active": "N", "country": " NZ ", "extra": ""}, {"unit": "KG", "email": "MKKIN@TEST.ORG", "status": "medium", "state": "Minnesota", "city": "   ormooc", "amount": "null", "is_active": "No", "country": "AL", "extra": ""}, {"unit": "cm", "email": "   LBAE@TEST.ORG", "status": "high", "state": "Wyoming", "city": "praguue ", "amount": "-", "is_active": "true", "country": "MW", "extra": ""}, {"unit": "kg", "email": "   JJCE@CORP.IO", "status": "P1", "state": " Connecticut", "city": "Ponte Vedra Beach", "amount": "$8,415.44", "is_active": "N", "country": "malawi", "extra": ""}, {"unit": "cntimeters", "email": "  bkfccnl@mail.com", "status": "HIGH", "state": "Virginia", "city": "Monterrey", "amount": "$1,880.70", "is_active": "false", "country": "Al", "extra": ""}, {"unit": " CM", "email": " NNHEMML@TEST.ORG ", "status": " Medium", "state": "VA", "city": "Ormoc   ", "amount": "8.759,69", "is_active": "TRUE", "country": "AL", "extra": ""}, {"unit": " centimeters ", "email": " DKDDK@CORP.IO ", "status": "P1", "state": "VA", "city": "Prague", "amount": "2.995,23", "is_active": "false", "country": "republi cof malawi", "extra": ""}, {"unit": "Cm", "email": "jcfbp@corp.io", "status": "P1", "state": "Connecticut", "city": "PRAGUE   ", "amount": "8.052,09", "is_active": "0", "country": "NZ", "extra": ""}, {"unit": "Centimeter   ", "email": "   lgocg@mail.com", "status": "HIGH", "state": "Wyoming", "city": "Prague", "amount": "N/A", "is_active": "No", "country": "Republic of aMlawi", "extra": ""}, {"unit": "kilogram", "email": " bffpcef@corp.io ", "status": "M", "state": "mn", "city": "Monterrey", "amount": "$4,399.93", "is_active": "FALSE", "country": "NZ  ", "extra": ""}, {"unit": "KG", "email": "coio@test.org", "status": "h", "state": "co", "city": "Ormoc", "amount": "None", "is_active": "Y", "country": "albania", "extra": ""}, {"unit": "centimeter", "email": " dhfke@mail.com ", "status": "med", "state": "connecticut", "city": "monnterrey", "amount": "8.202,10", "is_active": "No", "country": "   BV", "extra": ""}, {"unit": " centiceter ", "email": "ALFAPHN@EXAMPLE.COM", "status": " HIGH ", "state": "ct", "city": "Monterrey", "amount": "$7,122.87", "is_active": "Yes", "country": "bvt", "extra": ""}, {"unit": "centimeter", "email": "KIAKEF@CORP.IO", "status": "M", "state": "VA   ", "city": "ormoc", "amount": "6.257,99", "is_active": "F", "country": "AL", "extra": ""}, {"unit": "Cm", "email": "IIDDFP@TEST.ORG", "status": "Medum", "state": "WY", "city": "Prague", "amount": "7.307,29", "is_active": "F", "country": "BVT", "extra": ""}, {"unit": "centieeter", "email": "ncedj@corp.io", "status": "medium", "state": "WY", "city": " monterrey ", "amount": "na", "is_active": "Yes", "country": "BV", "extra": ""}, {"unit": "Kilogram", "email": " DCHPPNM@MAIL.COM ", "status": "   m", "state": "Wy   ", "city": " prague", "amount": "$6,772.59", "is_active": "FALSE", "country": "MALAWI", "extra": ""}, {"unit": "Kg", "email": "PNMK@CORP.IO", "status": "high", "state": "CO", "city": "monterrey", "amount": "$7,866.25", "is_active": "FALSE", "country": "cck  ", "extra": ""}, {"unit": "", "email": "", "status": "", "state": "", "city": "", "amount": "", "is_active": "", "country": "", "extra": ""}], "clean": [{"unit": "kg", "email": "iidlgl@test.org", "status": "Medium", "state": "Minnesota", "city": "Prague", "amount": 1532.79, "is_active": true, "country": "Malawi"}, {"unit": "cm", "email": "loneefg@mail.com", "status": "Medium", "state": "Virginia", "city": "Monterrey", "amount": NaN, "is_active": true, "country": "Albania"}, {"unit": "cm", "email": "bddknb@example.com", "status": "Medium", "state": "Virginia", "city": "Prague", "amount": 177.01, "is_active": true, "country": "Cocos (Keeling) Islands"}, {"unit": "kg", "email": "eief@example.com", "status": "Medium", "state": "Minnesota", "city": "Ponte Vedra Beach", "amount": 7647.59, "is_active": true, "country": "Malawi"}, {"unit": "kg", "email": "cdkfje@mail.com", "status": "Medium", "state": "Colorado", "city": "Ormoc", "amount": 5783.0, "is_active": false, "country": "New Zealand"}, {"unit": "kg", "email": "eapjdfn@mail.com", "status": "High", "state": "Connecticut", "city": "Prague", "amount": NaN, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "ngljang@example.com", "status": "Medium", "state": "Minnesota", "city": "Ormoc", "amount": 6545.33, "is_active": false, "country": "Malawi"}, {"unit": "kg", "email": "cilho@mail.com", "status": "Medium", "state": "Minnesota", "city": "Ormoc", "amount": 3476.42, "is_active": true, "country": "Malawi"}, {"unit": "kg", "email": "hijib@test.org", "status": "High", "state": "Colorado", "city": "Monterrey", "amount": 8876.0, "is_active": false, "country": "Cocos (Keeling) Islands"}, {"unit": "kg", "email": "jgklp@corp.io", "status": "High", "state": "Colorado", "city": "Ormoc", "amount": 8085.85, "is_active": true, "country": "Cocos (Keeling) Islands"}, {"unit": "cm", "email": "pkkpakp@mail.com", "status": "Medium", "state": "Wyoming", "city": "Monterrey", "amount": 2088.81, "is_active": true, "country": "Albania"}, {"unit": "kg", "email": "mljiph@test.org", "status": "High", "state": "Connecticut", "city": "Monterrey", "amount": 7023.5, "is_active": true, "country": "Malawi"}, {"unit": "kg", "email": "ageeeka@corp.io", "status": "High", "state": "Minnesota", "city": "Ormoc", "amount": 4468.51, "is_active": false, "country": "Malawi"}, {"unit": "cm", "email": "nocbhj@example.com", "status": "High", "state": "Virginia", "city": "Prague", "amount": 1475.19, "is_active": true, "country": "Bouvet Island"}, {"unit": "kg", "email": "oikonhg@mail.com", "status": "High", "state": "Minnesota", "city": "Ormoc", "amount": 1734.35, "is_active": false, "country": "Bouvet Island"}, {"unit": "kg", "email": "mlllfb@example.com", "status": "Medium", "state": "Wyoming", "city": "Ormoc", "amount": 5225.6, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "fiflca@example.com", "status": "High", "state": "Minnesota", "city": "Monterrey", "amount": 7408.11, "is_active": false, "country": "New Zealand"}, {"unit": "kg", "email": "amdc@mail.com", "status": "Medium", "state": "Minnesota", "city": "Ponte Vedra Beach", "amount": 452.47, "is_active": true, "country": "Albania"}, {"unit": "kg", "email": "ojlpiep@corp.io", "status": "Medium", "state": "Minnesota", "city": "Monterrey", "amount": 7262.33, "is_active": true, "country": "Malawi"}, {"unit": "cm", "email": "omhf@example.com", "status": "High", "state": "Minnesota", "city": "Monterrey", "amount": NaN, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "jkgcic@test.org", "status": "High", "state": "Virginia", "city": "Ponte Vedra Beach", "amount": 1644.98, "is_active": true, "country": "Malawi"}, {"unit": "kg", "email": "egaie@corp.io", "status": "High", "state": "Virginia", "city": "Ponte Vedra Beach", "amount": 4341.25, "is_active": true, "country": "Cocos (Keeling) Islands"}, {"unit": "cm", "email": "dehdj@mail.com", "status": "High", "state": "Colorado", "city": "Ormoc", "amount": NaN, "is_active": false, "country": "Malawi"}, {"unit": "cm", "email": "khlehk@test.org", "status": "High", "state": "Minnesota", "city": "Monterrey", "amount": 7817.76, "is_active": true, "country": "Cocos (Keeling) Islands"}, {"unit": "kg", "email": "ciehka@corp.io", "status": "High", "state": "Minnesota", "city": "Ponte Vedra Beach", "amount": 6887.56, "is_active": true, "country": "Bouvet Island"}, {"unit": "cm", "email": "kpjkfbj@corp.io", "status": "High", "state": "Connecticut", "city": "Ormoc", "amount": 2122.02, "is_active": true, "country": "Albania"}, {"unit": "kg", "email": "ialbdli@test.org", "status": "Medium", "state": "Connecticut", "city": "Prague", "amount": 2458.64, "is_active": true, "country": "Cocos (Keeling) Islands"}, {"unit": "kg", "email": "blfima@mail.com", "status": "Medium", "state": "Colorado", "city": "Ponte Vedra Beach", "amount": 1652.27, "is_active": true, "country": "Albania"}, {"unit": "cm", "email": "nhnbno@mail.com", "status": "High", "state": "Colorado", "city": "Ormoc", "amount": 4044.22, "is_active": false, "country": "Bouvet Island"}, {"unit": "kg", "email": "emhhpob@test.org", "status": "High", "state": "Wyoming", "city": "Prague", "amount": 6468.26, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "lofnk@mail.com", "status": "High", "state": "Wyoming", "city": "Ormoc", "amount": 2211.49, "is_active": false, "country": "New Zealand"}, {"unit": "cm", "email": "fconfcb@mail.com", "status": "High", "state": "Connecticut", "city": "Ponte Vedra Beach", "amount": NaN, "is_active": false, "country": "Malawi"}, {"unit": "kg", "email": "cmjf@mail.com", "status": "High", "state": "Minnesota", "city": "Ponte Vedra Beach", "amount": 8275.98, "is_active": true, "country": "Cocos (Keeling) Islands"}, {"unit": "cm", "email": "geecj@mail.com", "status": "Medium", "state": "Virginia", "city": "Ponte Vedra Beach", "amount": 3057.7, "is_active": true, "country": "New Zealand"}, {"unit": "kg", "email": "aplfje@example.com", "status": "High", "state": "Wyoming", "city": "Ormoc", "amount": 7256.77, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "fmbli@example.com", "status": "Medium", "state": "Colorado", "city": "Monterrey", "amount": 102.73, "is_active": true, "country": "Malawi"}, {"unit": "kg", "email": "nignc@mail.com", "status": "High", "state": "Virginia", "city": "Prague", "amount": 2726.09, "is_active": false, "country": "New Zealand"}, {"unit": "kg", "email": "mkkin@test.org", "status": "Medium", "state": "Minnesota", "city": "Ormoc", "amount": NaN, "is_active": false, "country": "Albania"}, {"unit": "cm", "email": "lbae@test.org", "status": "High", "state": "Wyoming", "city": "Prague", "amount": NaN, "is_active": true, "country": "Malawi"}, {"unit": "kg", "email": "jjce@corp.io", "status": "High", "state": "Connecticut", "city": "Ponte Vedra Beach", "amount": 8415.44, "is_active": false, "country": "Malawi"}, {"unit": "cm", "email": "bkfccnl@mail.com", "status": "High", "state": "Virginia", "city": "Monterrey", "amount": 1880.7, "is_active": false, "country": "Albania"}, {"unit": "cm", "email": "nnhemml@test.org", "status": "Medium", "state": "Virginia", "city": "Ormoc", "amount": 8759.69, "is_active": true, "country": "Albania"}, {"unit": "cm", "email": "dkddk@corp.io", "status": "High", "state": "Virginia", "city": "Prague", "amount": 2995.23, "is_active": false, "country": "Malawi"}, {"unit": "cm", "email": "jcfbp@corp.io", "status": "High", "state": "Connecticut", "city": "Prague", "amount": 8052.09, "is_active": false, "country": "New Zealand"}, {"unit": "cm", "email": "lgocg@mail.com", "status": "High", "state": "Wyoming", "city": "Prague", "amount": NaN, "is_active": false, "country": "Malawi"}, {"unit": "kg", "email": "bffpcef@corp.io", "status": "Medium", "state": "Minnesota", "city": "Monterrey", "amount": 4399.93, "is_active": false, "country": "New Zealand"}, {"unit": "kg", "email": "coio@test.org", "status": "High", "state": "Colorado", "city": "Ormoc", "amount": NaN, "is_active": true, "country": "Albania"}, {"unit": "cm", "email": "dhfke@mail.com", "status": "Medium", "state": "Connecticut", "city": "Monterrey", "amount": 8202.1, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "alfaphn@example.com", "status": "High", "state": "Connecticut", "city": "Monterrey", "amount": 7122.87, "is_active": true, "country": "Bouvet Island"}, {"unit": "cm", "email": "kiakef@corp.io", "status": "Medium", "state": "Virginia", "city": "Ormoc", "amount": 6257.99, "is_active": false, "country": "Albania"}, {"unit": "cm", "email": "iiddfp@test.org", "status": "Medium", "state": "Wyoming", "city": "Prague", "amount": 7307.29, "is_active": false, "country": "Bouvet Island"}, {"unit": "cm", "email": "ncedj@corp.io", "status": "Medium", "state": "Wyoming", "city": "Monterrey", "amount": NaN, "is_active": true, "country": "Bouvet Island"}, {"unit": "kg", "email": "dchppnm@mail.com", "status": "Medium", "state": "Wyoming", "city": "Prague", "amount": 6772.59, "is_active": false, "country": "Malawi"}, {"unit": "kg", "email": "pnmk@corp.io", "status": "High", "state": "Colorado", "city": "Monterrey", "amount": 7866.25, "is_active": false, "country": "Cocos (Keeling) Islands"}], "dirty_cols": ["unit", "email", "status", "state", "city", "amount", "is_active", "country", "extra"], "clean_cols": ["unit", "email", "status", "state", "city", "amount", "is_active", "country"], "plan": {"dataset_summary": "55 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilograms": "kg", "CM": "cm", "centimeters": "cm", "KG": "kg", "Ceentimeter": "cm", "Kg": "kg", "kilogrcms": "kg", "centimeter": "cm", "CENTIMETER": "cm", "KILOGRAMS": "kg", "centimeers": "cm", "centimater": "cm", "KILOGRAM": "kg", "Cm": "cm", "cntimeters": "cm", "Centimeter": "cm", "kilogram": "kg", "centiceter": "cm", "centieeter": "cm", "Kilogram": "kg"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"m": "Medium", "P2": "Medium", "MED": "Medium", "med": "Medium", "p1": "High", "H": "High", "P1": "High", "HIGH": "High", "high": "High", "medium": "Medium", "higgh": "High", "h": "High", "MEDIUM": "Medium", "p2": "Medium", "M": "Medium", "Medum": "Medium"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"VA": "Virginia", "mn": "Minnesota", "minnesota": "Minnesota", "co": "Colorado", "CO": "Colorado", "Wy": "Wyoming", "MN": "Minnesota", "MINNESOTA": "Minnesota", "virginia": "Virginia", "ct": "Connecticut", "Co": "Colorado", "WY": "Wyoming", "va": "Virginia", "connecticut": "Connecticut"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"monterrey": "Monterrey", "Pont eVedra Beach": "Ponte Vedra Beach", "PRAGUE": "Prague", "ORMOC": "Ormoc", "ormoc": "Ormoc", "ponte vedra beach": "Ponte Vedra Beach", "MONTERREY": "Monterrey", "PonteV edra Beach": "Ponte Vedra Beach", "Munterrey": "Monterrey", "ormooc": "Ormoc", "praguue": "Prague", "monnterrey": "Monterrey", "prague": "Prague"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"republic of malawi": "Malawi", "ALB": "Albania", "CC": "Cocos (Keeling) Islands", "malawi": "Malawi", "NZ": "New Zealand", "MW": "Malawi", "Republic of Malawi": "Malawi", "Cocos Keeling Islands": "Cocos (Keeling) Islands", "albvnia": "Albania", "MWI": "Malawi", "REPUBLIC OF MALAWI": "Malawi", "BVT": "Bouvet Island", "bvt": "Bouvet Island", "CCK": "Cocos (Keeling) Islands", "COCOS (KEELING) ISLANDS": "Cocos (Keeling) Islands", "alb": "Albania", "Republic of Albania": "Albania", "Bv": "Bouvet Island", "bv": "Bouvet Island", "Nwe Zealand": "New Zealand", "Malwi": "Malawi", "AL": "Albania", "Al": "Albania", "republi cof malawi": "Malawi", "Republic of aMlawi": "Malawi", "albania": "Albania", "BV": "Bouvet Island", "MALAWI": "Malawi", "cck": "Cocos (Keeling) Islands"}, "rationale": "Unified 29 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"status": "Paid", "job_title": "CTO", "amount": "$5,006.46", "email": "OLDK@EXAMPLE.COM", "state": "Indiana", "currency": "Fiji Dollar", "phone": "564.822.7391", "city": "mbinga"}, {"status": "   paid", "job_title": "CHIEF TECHNOLOGY OFFICER   ", "amount": "5.850,86", "email": " lfod@corp.io", "state": "Pennsylvania", "currency": " FJD", "phone": "585-250-1292", "city": "LIUTUAN"}, {"status": "paid", "job_title": "CTO", "amount": "$586.54", "email": "EIJIC@MAIL.COM", "state": "PENNSYLAVNIA", "currency": "Brunei Dollar", "phone": "589-303-5747", "city": "Minga"}, {"status": "PIAD", "job_title": "rep", "amount": "885,22", "email": "CNNC@EXAMPLE.COM", "state": "IN", "currency": "BND ", "phone": "(516)420-7059", "city": "Schoten"}, {"status": " paiid", "job_title": " sales representative", "amount": "null", "email": "   PCKCP@TEST.ORG", "state": "   PA", "currency": "FJD   ", "phone": "5186137021", "city": "SCHOTEN"}, {"status": "due", "job_title": " CTO", "amount": "null", "email": "   cefolho@test.org", "state": "north dakota", "currency": "fiji dollar", "phone": "5028584577", "city": "Mbinga "}, {"status": "due", "job_title": "CTO  ", "amount": "4.045,36", "email": "gdffak@test.org", "state": "Pennsylvania", "currency": "FIJI DOLALR", "phone": "588-125-9370", "city": "Schoten"}, {"status": "unpaid", "job_title": "  C.T.O.", "amount": "$8,099.51", "email": "ILILD@TEST.ORG  ", "state": "   ND", "currency": "bnd   ", "phone": "(523)691-3374", "city": "MBINGA"}, {"status": "  Pending", "job_title": "CTO", "amount": "$3,766.85", "email": " BJKIDJM@CORP.IO ", "state": "IN", "currency": "BND", "phone": "5935240650", "city": "Mbinga"}, {"status": "Due", "job_title": "c.t.o.", "amount": "445,04", "email": "madgop@example.com", "state": "Indiana", "currency": "FJD", "phone": "537-053-3928", "city": "schoten"}, {"status": "Pending", "job_title": "CTO", "amount": "5.993,63", "email": " DFOKB@MAIL.COM ", "state": "pa", "currency": "BND", "phone": "578-564-6112", "city": "Minga"}, {"status": "PAD", "job_title": "sales rep", "amount": "3.007,59", "email": "icno@example.com ", "state": "Indiana  ", "currency": "fjd", "phone": "525.654.8877", "city": "Schoten"}, {"status": "unpaid", "job_title": "Cto", "amount": "5.364,23", "email": "   IAGPBBD@EXAMPLE.COM", "state": "North Dakoa", "currency": "  BND", "phone": "576.641.4179", "city": "Liutuan"}, {"status": "PAID", "job_title": "Chief Technology Officer", "amount": "3.791,37", "email": "mdeg@test.org", "state": "IN", "currency": "BRUNEI DOLLAR", "phone": "(506)945-3832", "city": "Mbinga"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": "$6,279.05", "email": "PEAG@MAIL.COM", "state": "north dakota", "currency": "  BND", "phone": "552-360-5626", "city": "Liutuan"}, {"status": "Padi", "job_title": "   CTO", "amount": "$7,143.10", "email": "GPINBK@EXAMPLE.COM", "state": "   PA", "currency": "bruneid ollar  ", "phone": "520.722.4312", "city": "mbinga"}, {"status": "Paid", "job_title": "Sales Representative", "amount": "$1,438.00", "email": "  IGCP@CORP.IO", "state": "Indiana", "currency": "fjd", "phone": "5951802024", "city": "LIUTUAN"}, {"status": "   paid", "job_title": "C.T.O.", "amount": "676,90", "email": " ddoei@mail.com ", "state": "PA", "currency": "   BND", "phone": "574.283.3561", "city": "Liutuan"}, {"status": "pending", "job_title": " Chief Technology Officer", "amount": "4.737,62", "email": " kidhcd@example.com ", "state": "Indiana  ", "currency": "BND", "phone": "588-014-1244", "city": "Scohten"}, {"status": "paid", "job_title": "Sales Rpresentative", "amount": "8.790,38", "email": "ELLM@TEST.ORG", "state": "pennslyvania", "currency": "FIJI DOLLAR", "phone": "538.355.0248", "city": "mbinga"}, {"status": "paid", "job_title": "rep", "amount": "1.389,19", "email": "hplkpan@corp.io   ", "state": "North Dakota  ", "currency": "fjd", "phone": "544.395.1521", "city": "Schoten"}, {"status": "unpaid", "job_title": "c.t.o. ", "amount": "820,60", "email": "gggmap@corp.io", "state": "ND", "currency": "fjd", "phone": "(524)516-4980", "city": "Schoten"}, {"status": "Paid   ", "job_title": "cto", "amount": "null", "email": "   KMPGNOJ@MAIL.COM", "state": "IN", "currency": "BND  ", "phone": "557.206.9288", "city": "liutuan"}, {"status": "PAID", "job_title": "CTO", "amount": "516,64", "email": "  hmjdnem@mail.com", "state": "pa", "currency": "FIJI  DOLLAR", "phone": "530-201-1228", "city": "Schoten"}, {"status": "Pending", "job_title": "c.t.o.", "amount": "6.701,51", "email": "   LEGFLG@MAIL.COM", "state": "North Dakota", "currency": "BND", "phone": "538-762-4250", "city": "LIUTUAN"}, {"status": "paid", "job_title": "CTO", "amount": "$586.54", "email": "EIJIC@MAIL.COM", "state": "PENNSYLAVNIA", "currency": "Brunei Dollar", "phone": "589-303-5747", "city": "Minga"}], "clean": [{"status": "Paid", "job_title": "Chief Technology Officer", "amount": 5006.46, "email": "oldk@example.com", "state": "Indiana", "currency": "FJD", "phone": "(564) 822-7391", "city": "Mbinga"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 5850.86, "email": "lfod@corp.io", "state": "Pennsylvania", "currency": "FJD", "phone": "(585) 250-1292", "city": "Liutuan"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 586.54, "email": "eijic@mail.com", "state": "Pennsylvania", "currency": "BND", "phone": "(589) 303-5747", "city": "Mbinga"}, {"status": "Paid", "job_title": "Sales Representative", "amount": 885.22, "email": "cnnc@example.com", "state": "Indiana", "currency": "BND", "phone": "(516) 420-7059", "city": "Schoten"}, {"status": "Paid", "job_title": "Sales Representative", "amount": NaN, "email": "pckcp@test.org", "state": "Pennsylvania", "currency": "FJD", "phone": "(518) 613-7021", "city": "Schoten"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": NaN, "email": "cefolho@test.org", "state": "North Dakota", "currency": "FJD", "phone": "(502) 858-4577", "city": "Mbinga"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 4045.36, "email": "gdffak@test.org", "state": "Pennsylvania", "currency": "FJD", "phone": "(588) 125-9370", "city": "Schoten"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 8099.51, "email": "ilild@test.org", "state": "North Dakota", "currency": "BND", "phone": "(523) 691-3374", "city": "Mbinga"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 3766.85, "email": "bjkidjm@corp.io", "state": "Indiana", "currency": "BND", "phone": "(593) 524-0650", "city": "Mbinga"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 445.04, "email": "madgop@example.com", "state": "Indiana", "currency": "FJD", "phone": "(537) 053-3928", "city": "Schoten"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 5993.63, "email": "dfokb@mail.com", "state": "Pennsylvania", "currency": "BND", "phone": "(578) 564-6112", "city": "Mbinga"}, {"status": "Paid", "job_title": "Sales Representative", "amount": 3007.59, "email": "icno@example.com", "state": "Indiana", "currency": "FJD", "phone": "(525) 654-8877", "city": "Schoten"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 5364.23, "email": "iagpbbd@example.com", "state": "North Dakota", "currency": "BND", "phone": "(576) 641-4179", "city": "Liutuan"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 3791.37, "email": "mdeg@test.org", "state": "Indiana", "currency": "BND", "phone": "(506) 945-3832", "city": "Mbinga"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 6279.05, "email": "peag@mail.com", "state": "North Dakota", "currency": "BND", "phone": "(552) 360-5626", "city": "Liutuan"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 7143.1, "email": "gpinbk@example.com", "state": "Pennsylvania", "currency": "BND", "phone": "(520) 722-4312", "city": "Mbinga"}, {"status": "Paid", "job_title": "Sales Representative", "amount": 1438.0, "email": "igcp@corp.io", "state": "Indiana", "currency": "FJD", "phone": "(595) 180-2024", "city": "Liutuan"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 676.9, "email": "ddoei@mail.com", "state": "Pennsylvania", "currency": "BND", "phone": "(574) 283-3561", "city": "Liutuan"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 4737.62, "email": "kidhcd@example.com", "state": "Indiana", "currency": "BND", "phone": "(588) 014-1244", "city": "Schoten"}, {"status": "Paid", "job_title": "Sales Representative", "amount": 8790.38, "email": "ellm@test.org", "state": "Pennsylvania", "currency": "FJD", "phone": "(538) 355-0248", "city": "Mbinga"}, {"status": "Paid", "job_title": "Sales Representative", "amount": 1389.19, "email": "hplkpan@corp.io", "state": "North Dakota", "currency": "FJD", "phone": "(544) 395-1521", "city": "Schoten"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 820.6, "email": "gggmap@corp.io", "state": "North Dakota", "currency": "FJD", "phone": "(524) 516-4980", "city": "Schoten"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": NaN, "email": "kmpgnoj@mail.com", "state": "Indiana", "currency": "BND", "phone": "(557) 206-9288", "city": "Liutuan"}, {"status": "Paid", "job_title": "Chief Technology Officer", "amount": 516.64, "email": "hmjdnem@mail.com", "state": "Pennsylvania", "currency": "FJD", "phone": "(530) 201-1228", "city": "Schoten"}, {"status": "Pending", "job_title": "Chief Technology Officer", "amount": 6701.51, "email": "legflg@mail.com", "state": "North Dakota", "currency": "BND", "phone": "(538) 762-4250", "city": "Liutuan"}], "dirty_cols": ["status", "job_title", "amount", "email", "state", "currency", "phone", "city"], "clean_cols": ["status", "job_title", "amount", "email", "state", "currency", "phone", "city"], "plan": {"dataset_summary": "26 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paid": "Paid", "PIAD": "Paid", "paiid": "Paid", "due": "Pending", "unpaid": "Pending", "Due": "Pending", "PAD": "Paid", "PAID": "Paid", "Padi": "Paid", "pending": "Pending"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CTO": "Chief Technology Officer", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "rep": "Sales Representative", "sales representative": "Sales Representative", "C.T.O.": "Chief Technology Officer", "c.t.o.": "Chief Technology Officer", "sales rep": "Sales Representative", "Cto": "Chief Technology Officer", "Sales Rpresentative": "Sales Representative", "cto": "Chief Technology Officer"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PENNSYLAVNIA": "Pennsylvania", "IN": "Indiana", "PA": "Pennsylvania", "north dakota": "North Dakota", "ND": "North Dakota", "pa": "Pennsylvania", "North Dakoa": "North Dakota", "pennslyvania": "Pennsylvania"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fiji Dollar": "FJD", "Brunei Dollar": "BND", "fiji dollar": "FJD", "FIJI DOLALR": "FJD", "bnd": "BND", "fjd": "FJD", "BRUNEI DOLLAR": "BND", "bruneid ollar": "BND", "FIJI DOLLAR": "FJD"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mbinga": "Mbinga", "LIUTUAN": "Liutuan", "Minga": "Mbinga", "SCHOTEN": "Schoten", "MBINGA": "Mbinga", "schoten": "Schoten", "Scohten": "Schoten", "liutuan": "Liutuan"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"unit": "Kg", "signup_date": "8 May 2023", "is_active": "true"}, {"unit": " KG ", "signup_date": "45206", "is_active": "FALSE"}, {"unit": "Ml", "signup_date": "45095", "is_active": "FALSE"}, {"unit": "  miles", "signup_date": "12 Mar 2023", "is_active": "FALSE"}, {"unit": "MILE", "signup_date": "45231", "is_active": "N"}, {"unit": "KILOMETER", "signup_date": "1/11/2023", "is_active": "N"}, {"unit": "   Mi", "signup_date": "11 Aug 2023", "is_active": "false"}, {"unit": " km ", "signup_date": "18 Dec 2023", "is_active": "Yes"}, {"unit": "kilometters", "signup_date": "45241", "is_active": "Y"}, {"unit": " miles ", "signup_date": "2023-09-01", "is_active": "No"}, {"unit": "KM ", "signup_date": "2 Apr 2023", "is_active": "F"}, {"unit": "mL   ", "signup_date": "8 Feb 2023", "is_active": "No"}, {"unit": "MILE", "signup_date": "45231", "is_active": "N"}], "clean": [{"unit": "kg", "signup_date": "2023-05-08", "is_active": true}, {"unit": "kg", "signup_date": "2023-10-07", "is_active": false}, {"unit": "mL", "signup_date": "2023-06-18", "is_active": false}, {"unit": "mi", "signup_date": "2023-03-12", "is_active": false}, {"unit": "mi", "signup_date": "2023-11-01", "is_active": false}, {"unit": "km", "signup_date": "2023-01-11", "is_active": false}, {"unit": "mi", "signup_date": "2023-08-11", "is_active": false}, {"unit": "km", "signup_date": "2023-12-18", "is_active": true}, {"unit": "km", "signup_date": "2023-11-11", "is_active": true}, {"unit": "mi", "signup_date": "2023-09-01", "is_active": false}, {"unit": "km", "signup_date": "2023-04-02", "is_active": false}, {"unit": "mL", "signup_date": "2023-02-08", "is_active": false}], "dirty_cols": ["unit", "signup_date", "is_active"], "clean_cols": ["unit", "signup_date", "is_active"], "plan": {"dataset_summary": "13 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Kg": "kg", "KG": "kg", "Ml": "mL", "miles": "mi", "MILE": "mi", "KILOMETER": "km", "Mi": "mi", "kilometters": "km", "KM": "km"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"industry": "Telco", "department": "MARKETING", "job_title": "C.E.O.", "amount": "N/A", "phone": "(598)775-6002", "state": "GA", "email": "GGODLK@TEST.ORG", "rate": "17.7%", "extra": ""}, {"industry": "Telco", "department": "Product", "job_title": "ProductgManager", "amount": "2.342,42", "phone": "5495208574", "state": "oh", "email": " PPDNC@MAIL.COM ", "rate": "64.0%", "extra": ""}, {"industry": "RealEsttae", "department": " INFORMATION TECHNOLOGY", "job_title": "chief executive officer", "amount": "4.156,13", "phone": "558-753-7352", "state": "MN", "email": "cppipm@corp.io", "rate": "38.6%", "extra": ""}, {"industry": "comms", "department": "i.t.", "job_title": "   PROD MGR", "amount": "1.801,93", "phone": "592-269-9626", "state": "mn", "email": "LKHO@TEST.ORG", "rate": "67.1%", "extra": ""}, {"industry": "proptech", "department": "I.T.", "job_title": "Chief Executive Officer", "amount": "$1,276.62", "phone": "5236650229", "state": " oh ", "email": "hgjafjj@corp.io", "rate": "22.7%", "extra": ""}, {"industry": "RealEsttate", "department": " mkpg", "job_title": "Sr. Engineer ", "amount": "543,68", "phone": "523.671.9759", "state": "Oh", "email": "bnkdhc@mail.com", "rate": "53.5%", "extra": ""}, {"industry": "technology", "department": "Prouct", "job_title": "PROD MANAGER", "amount": "1.985,90", "phone": "5546044518", "state": "CA", "email": "nmlkclh@mail.com  ", "rate": "71.1%", "extra": ""}, {"industry": "Tech", "department": "Prod", "job_title": "senior  eng", "amount": "4.844,05", "phone": "(545)935-4584", "state": "Indiana", "email": "ldbjn@example.com", "rate": "6.0%", "extra": ""}, {"industry": "Health", "department": "marketing", "job_title": "C.E.O.  ", "amount": "--", "phone": "505-729-3913", "state": "  IN", "email": "   ngaiec@mail.com", "rate": "29.1%", "extra": ""}, {"industry": "Technology", "department": "pm", "job_title": "Pm", "amount": "$7,818.74", "phone": "542.926.1220", "state": "GA", "email": "  ebkkh@test.org", "rate": "67.6%", "extra": ""}, {"industry": "comms", "department": "Marketing", "job_title": "seinor  eng", "amount": "N/A", "phone": "5035855634", "state": " IN ", "email": " iiodipj@mail.com ", "rate": "52.1%", "extra": ""}, {"industry": " comms ", "department": "Mkt", "job_title": "Product Manager", "amount": "$4,728.12", "phone": "552.857.8771", "state": "CALIFORNIA", "email": "hdbg@mail.com", "rate": "14.2%", "extra": ""}, {"industry": "RealEstate", "department": "  GROWTH", "job_title": "prod manager", "amount": "4.655,81", "phone": "(507)784-2621", "state": "CALIFORNIA", "email": "JBMMNF@CORP.IO", "rate": "68.1%", "extra": ""}, {"industry": "HEALTH CARE   ", "department": "I..", "job_title": "prod mgr", "amount": "334,01", "phone": "541.783.9590", "state": "Georgia", "email": "   nbmc@test.org", "rate": "13.0%", "extra": ""}, {"industry": "Technology", "department": "  Marketing", "job_title": "Product Manager", "amount": "4.419,39", "phone": "(598)464-6317", "state": "Indiana", "email": "IFBP@MAIL.COM", "rate": "2.2%", "extra": ""}, {"industry": "telo", "department": "IT", "job_title": "senior  eng", "amount": "N/A", "phone": "5661732538", "state": " Minnesota", "email": "kmhmdim@example.com", "rate": "30.5%", "extra": ""}, {"industry": "REALESTATE", "department": "mkt", "job_title": "Product Managger", "amount": "4.406,14", "phone": "566.161.7057", "state": "California", "email": "afhaaln@mail.com  ", "rate": "5.6%", "extra": ""}, {"industry": "tech", "department": "product", "job_title": "Prod Manager", "amount": "$2,511.36", "phone": "557.007.5856", "state": "mn", "email": " BMOMF@MAIL.COM ", "rate": "54.3%", "extra": ""}, {"industry": "health", "department": "PM", "job_title": "prod manager", "amount": "8.204,43", "phone": "5206118571", "state": " California", "email": "mppl@example.com   ", "rate": "28.0%", "extra": ""}, {"industry": "Telecommunications", "department": "Mkt", "job_title": "sr engineer", "amount": "7.368,70", "phone": "527.883.8800", "state": "indiana", "email": "eeian@mail.com", "rate": "69.3%", "extra": ""}, {"industry": "Technology  ", "department": "growth", "job_title": "engineering manager", "amount": "$1,806.60", "phone": "(528)408-6535", "state": "OH", "email": "AKDME@CORP.IO", "rate": "10.2%", "extra": ""}, {"industry": "PropTech", "department": "growth", "job_title": " Engineering Manager", "amount": "-", "phone": "568-627-2456", "state": "ga ", "email": "ibmndb@example.com  ", "rate": "81.2%", "extra": ""}, {"industry": " Telecom ", "department": "PM", "job_title": "Eng Manager", "amount": "$8,499.86", "phone": "568-754-0858", "state": "IN", "email": "HHOD@MAIL.COM", "rate": "48.9%", "extra": ""}, {"industry": "  health care", "department": "   growth", "job_title": "C.EO.", "amount": "$3,268.03", "phone": "501-676-8706", "state": "Mn", "email": "nbbphjp@example.com", "rate": "42.1%", "extra": ""}, {"industry": "MEDICAL", "department": "it", "job_title": "Chief  Executive Officer", "amount": "8.230,92", "phone": "5320641026", "state": "   Georgia", "email": " dneg@example.com", "rate": "39.2%", "extra": ""}, {"industry": "   TECHNOLOGY", "department": " Product ", "job_title": "Eng Manager", "amount": "4.966,68", "phone": "(557)900-3563", "state": "  CA", "email": "BFFK@EXAMPLE.COM", "rate": "44.5%", "extra": ""}, {"industry": "RealEswate", "department": "mkt ", "job_title": "Prod Manager", "amount": "$3,990.88", "phone": "(504)286-9425", "state": " CA ", "email": "HFEAN@TEST.ORG", "rate": "90.1%", "extra": ""}, {"industry": "comms", "department": "Mkt", "job_title": "sr engineer", "amount": "4.068,50", "phone": "(533)295-0220", "state": "OH", "email": "   idgf@corp.io", "rate": "77.3%", "extra": ""}, {"industry": "medical", "department": "growth", "job_title": "ENG MGR", "amount": "$1,296.11", "phone": "541.166.2585", "state": "california", "email": "cepbbif@corp.io  ", "rate": "36.2%", "extra": ""}, {"industry": "MEDICAL", "department": "marketing", "job_title": "eng manager", "amount": "2.771,46", "phone": "(569)642-0112", "state": "MN", "email": "ejdaceh@mail.com", "rate": "26.9%", "extra": ""}, {"industry": "Health", "department": "Growth", "job_title": "snr engineer", "amount": "5.804,38", "phone": "513-773-0149", "state": "oiho", "email": "ibhmal@mail.com", "rate": "88.6%", "extra": ""}, {"industry": "Healthcare", "department": "information technology", "job_title": "prod manager", "amount": "$7,086.00", "phone": "526-718-7363", "state": "   CA", "email": "piaa@mail.com", "rate": "60.8%", "extra": ""}], "clean": [{"industry": "Telecommunications", "department": "Marketing", "job_title": "Chief Executive Officer", "amount": NaN, "phone": "(598) 775-6002", "state": "Georgia", "email": "ggodlk@test.org", "rate": 0.177}, {"industry": "Telecommunications", "department": "Product", "job_title": "Product Manager", "amount": 2342.42, "phone": "(549) 520-8574", "state": "Ohio", "email": "ppdnc@mail.com", "rate": 0.64}, {"industry": "Real Estate", "department": "Information Technology", "job_title": "Chief Executive Officer", "amount": 4156.13, "phone": "(558) 753-7352", "state": "Minnesota", "email": "cppipm@corp.io", "rate": 0.386}, {"industry": "Telecommunications", "department": "Information Technology", "job_title": "Product Manager", "amount": 1801.93, "phone": "(592) 269-9626", "state": "Minnesota", "email": "lkho@test.org", "rate": 0.6709999999999999}, {"industry": "Real Estate", "department": "Information Technology", "job_title": "Chief Executive Officer", "amount": 1276.62, "phone": "(523) 665-0229", "state": "Ohio", "email": "hgjafjj@corp.io", "rate": 0.22699999999999998}, {"industry": "Real Estate", "department": "Marketing", "job_title": "Senior Engineer", "amount": 543.68, "phone": "(523) 671-9759", "state": "Ohio", "email": "bnkdhc@mail.com", "rate": 0.535}, {"industry": "Technology", "department": "Product", "job_title": "Product Manager", "amount": 1985.9, "phone": "(554) 604-4518", "state": "California", "email": "nmlkclh@mail.com", "rate": 0.711}, {"industry": "Technology", "department": "Product", "job_title": "Senior Engineer", "amount": 4844.05, "phone": "(545) 935-4584", "state": "Indiana", "email": "ldbjn@example.com", "rate": 0.06}, {"industry": "Healthcare", "department": "Marketing", "job_title": "Chief Executive Officer", "amount": NaN, "phone": "(505) 729-3913", "state": "Indiana", "email": "ngaiec@mail.com", "rate": 0.29100000000000004}, {"industry": "Technology", "department": "Product", "job_title": "Product Manager", "amount": 7818.74, "phone": "(542) 926-1220", "state": "Georgia", "email": "ebkkh@test.org", "rate": 0.6759999999999999}, {"industry": "Telecommunications", "department": "Marketing", "job_title": "Senior Engineer", "amount": NaN, "phone": "(503) 585-5634", "state": "Indiana", "email": "iiodipj@mail.com", "rate": 0.521}, {"industry": "Telecommunications", "department": "Marketing", "job_title": "Product Manager", "amount": 4728.12, "phone": "(552) 857-8771", "state": "California", "email": "hdbg@mail.com", "rate": 0.142}, {"industry": "Real Estate", "department": "Marketing", "job_title": "Product Manager", "amount": 4655.81, "phone": "(507) 784-2621", "state": "California", "email": "jbmmnf@corp.io", "rate": 0.6809999999999999}, {"industry": "Healthcare", "department": "Information Technology", "job_title": "Product Manager", "amount": 334.01, "phone": "(541) 783-9590", "state": "Georgia", "email": "nbmc@test.org", "rate": 0.13}, {"industry": "Technology", "department": "Marketing", "job_title": "Product Manager", "amount": 4419.39, "phone": "(598) 464-6317", "state": "Indiana", "email": "ifbp@mail.com", "rate": 0.022000000000000002}, {"industry": "Telecommunications", "department": "Information Technology", "job_title": "Senior Engineer", "amount": NaN, "phone": "(566) 173-2538", "state": "Minnesota", "email": "kmhmdim@example.com", "rate": 0.305}, {"industry": "Real Estate", "department": "Marketing", "job_title": "Product Manager", "amount": 4406.14, "phone": "(566) 161-7057", "state": "California", "email": "afhaaln@mail.com", "rate": 0.055999999999999994}, {"industry": "Technology", "department": "Product", "job_title": "Product Manager", "amount": 2511.36, "phone": "(557) 007-5856", "state": "Minnesota", "email": "bmomf@mail.com", "rate": 0.5429999999999999}, {"industry": "Healthcare", "department": "Product", "job_title": "Product Manager", "amount": 8204.43, "phone": "(520) 611-8571", "state": "California", "email": "mppl@example.com", "rate": 0.28}, {"industry": "Telecommunications", "department": "Marketing", "job_title": "Senior Engineer", "amount": 7368.7, "phone": "(527) 883-8800", "state": "Indiana", "email": "eeian@mail.com", "rate": 0.693}, {"industry": "Technology", "department": "Marketing", "job_title": "Engineering Manager", "amount": 1806.6, "phone": "(528) 408-6535", "state": "Ohio", "email": "akdme@corp.io", "rate": 0.102}, {"industry": "Real Estate", "department": "Marketing", "job_title": "Engineering Manager", "amount": NaN, "phone": "(568) 627-2456", "state": "Georgia", "email": "ibmndb@example.com", "rate": 0.812}, {"industry": "Telecommunications", "department": "Product", "job_title": "Engineering Manager", "amount": 8499.86, "phone": "(568) 754-0858", "state": "Indiana", "email": "hhod@mail.com", "rate": 0.489}, {"industry": "Healthcare", "department": "Marketing", "job_title": "Chief Executive Officer", "amount": 3268.03, "phone": "(501) 676-8706", "state": "Minnesota", "email": "nbbphjp@example.com", "rate": 0.42100000000000004}, {"industry": "Healthcare", "department": "Information Technology", "job_title": "Chief Executive Officer", "amount": 8230.92, "phone": "(532) 064-1026", "state": "Georgia", "email": "dneg@example.com", "rate": 0.392}, {"industry": "Technology", "department": "Product", "job_title": "Engineering Manager", "amount": 4966.68, "phone": "(557) 900-3563", "state": "California", "email": "bffk@example.com", "rate": 0.445}, {"industry": "Real Estate", "department": "Marketing", "job_title": "Product Manager", "amount": 3990.88, "phone": "(504) 286-9425", "state": "California", "email": "hfean@test.org", "rate": 0.9009999999999999}, {"industry": "Telecommunications", "department": "Marketing", "job_title": "Senior Engineer", "amount": 4068.5, "phone": "(533) 295-0220", "state": "Ohio", "email": "idgf@corp.io", "rate": 0.773}, {"industry": "Healthcare", "department": "Marketing", "job_title": "Engineering Manager", "amount": 1296.11, "phone": "(541) 166-2585", "state": "California", "email": "cepbbif@corp.io", "rate": 0.36200000000000004}, {"industry": "Healthcare", "department": "Marketing", "job_title": "Engineering Manager", "amount": 2771.46, "phone": "(569) 642-0112", "state": "Minnesota", "email": "ejdaceh@mail.com", "rate": 0.26899999999999996}, {"industry": "Healthcare", "department": "Marketing", "job_title": "Senior Engineer", "amount": 5804.38, "phone": "(513) 773-0149", "state": "Ohio", "email": "ibhmal@mail.com", "rate": 0.8859999999999999}, {"industry": "Healthcare", "department": "Information Technology", "job_title": "Product Manager", "amount": 7086.0, "phone": "(526) 718-7363", "state": "California", "email": "piaa@mail.com", "rate": 0.608}], "dirty_cols": ["industry", "department", "job_title", "amount", "phone", "state", "email", "rate", "extra"], "clean_cols": ["industry", "department", "job_title", "amount", "phone", "state", "email", "rate"], "plan": {"dataset_summary": "32 rows × 9 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Telco": "Telecommunications", "RealEsttae": "Real Estate", "comms": "Telecommunications", "proptech": "Real Estate", "RealEsttate": "Real Estate", "technology": "Technology", "Tech": "Technology", "Health": "Healthcare", "RealEstate": "Real Estate", "HEALTH CARE": "Healthcare", "telo": "Telecommunications", "REALESTATE": "Real Estate", "tech": "Technology", "health": "Healthcare", "PropTech": "Real Estate", "Telecom": "Telecommunications", "health care": "Healthcare", "MEDICAL": "Healthcare", "TECHNOLOGY": "Technology", "RealEswate": "Real Estate", "medical": "Healthcare"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MARKETING": "Marketing", "INFORMATION TECHNOLOGY": "Information Technology", "i.t.": "Information Technology", "I.T.": "Information Technology", "mkpg": "Marketing", "Prouct": "Product", "Prod": "Product", "marketing": "Marketing", "pm": "Product", "Mkt": "Marketing", "GROWTH": "Marketing", "I..": "Information Technology", "IT": "Information Technology", "mkt": "Marketing", "product": "Product", "PM": "Product", "growth": "Marketing", "it": "Information Technology", "Growth": "Marketing", "information technology": "Information Technology"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"C.E.O.": "Chief Executive Officer", "ProductgManager": "Product Manager", "chief executive officer": "Chief Executive Officer", "PROD MGR": "Product Manager", "Sr. Engineer": "Senior Engineer", "PROD MANAGER": "Product Manager", "senior eng": "Senior Engineer", "Pm": "Product Manager", "seinor eng": "Senior Engineer", "prod manager": "Product Manager", "prod mgr": "Product Manager", "Product Managger": "Product Manager", "Prod Manager": "Product Manager", "sr engineer": "Senior Engineer", "engineering manager": "Engineering Manager", "Eng Manager": "Engineering Manager", "C.EO.": "Chief Executive Officer", "ENG MGR": "Engineering Manager", "eng manager": "Engineering Manager", "snr engineer": "Senior Engineer"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GA": "Georgia", "oh": "Ohio", "MN": "Minnesota", "mn": "Minnesota", "Oh": "Ohio", "CA": "California", "IN": "Indiana", "CALIFORNIA": "California", "indiana": "Indiana", "OH": "Ohio", "ga": "Georgia", "Mn": "Minnesota", "california": "California", "oiho": "Ohio"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"department": "Sales", "phone": "599.944.7821", "email": "jacodmn@mail.com", "state": "De", "is_active": "TRUE", "status": "HIGH"}, {"department": " Legal ", "phone": "5845658853", "email": "aljj@test.org", "state": "  HI", "is_active": "FALSE", "status": "High"}, {"department": "legal", "phone": "569-927-2845", "email": "PBNEG@TEST.ORG", "state": "delaware", "is_active": "F", "status": "P1"}, {"department": "BIZ DEV", "phone": "593.078.1620", "email": "iagdnjl@example.com", "state": "DE", "is_active": "TRUE", "status": "P2"}, {"department": "  Biz Dev", "phone": "(579)675-8558", "email": "FNILBGK@TEST.ORG", "state": "De", "is_active": "1", "status": "H"}, {"department": "Legal", "phone": "521.523.1959", "email": "   pabb@corp.io", "state": "HI", "is_active": "Y", "status": "medium"}, {"department": "Biz Dev", "phone": "572-704-8854", "email": "BGAF@EXAMPLE.COM", "state": "hawaii   ", "is_active": "0", "status": "medimu"}, {"department": "  Biz Dev", "phone": "513-786-8759", "email": "   jnmbpn@corp.io", "state": "HAWAII", "is_active": "Y", "status": "HIIGH"}, {"department": "Biz  Dev", "phone": "561.432.5007", "email": "AICKD@TEST.ORG", "state": "hawaii", "is_active": "N", "status": " med"}, {"department": "SLAES", "phone": "(566)972-4407", "email": "MAAKLKF@CORP.IO  ", "state": "HAWAII", "is_active": "false", "status": "medium"}, {"department": "Biz Dev", "phone": "5738103773", "email": "enmdf@corp.io  ", "state": "delaware", "is_active": "1", "status": "high"}, {"department": " Legal ", "phone": "(594)527-6881", "email": "PFODHJ@EXAMPLE.COM", "state": "HAWAII", "is_active": "1", "status": "P2"}, {"department": "Sales  ", "phone": "(535)456-4116", "email": "MCNCFOO@CORP.IO", "state": "DE", "is_active": "No", "status": " h"}, {"department": "legal & compliance", "phone": "5191816612", "email": " ecnad@corp.io ", "state": "Delaware", "is_active": "T", "status": "Medium"}, {"department": " sales ", "phone": "593-239-4767", "email": "CJMOLF@EXAMPLE.COM", "state": "Delaware", "is_active": "Yes", "status": "medium"}, {"department": "sales ", "phone": "(537)002-7902", "email": "lfod@test.org", "state": "Delaware  ", "is_active": "Yes", "status": "P1"}, {"department": "Sales", "phone": "5360576770", "email": "OGOEOLJ@CORP.IO", "state": "hi", "is_active": "FALSE", "status": "   High"}, {"department": "Legal", "phone": "5151716536", "email": "   clekgkc@test.org", "state": "DE", "is_active": "F", "status": " P1"}, {"department": "Legal", "phone": "5894684307", "email": "MGOP@EXAMPLE.COM", "state": " HI", "is_active": "No", "status": "H"}, {"department": "Legal", "phone": "5643066010", "email": "MGOO@MAIL.COM", "state": "Delaware", "is_active": "0", "status": "M"}, {"department": "   Biz Dev", "phone": "(531)564-8635", "email": "jgeijo@example.com", "state": "HAWAII", "is_active": "Yes", "status": "high  "}, {"department": "LEGAL & COPMLIANCE", "phone": "519-479-5318", "email": "mloh@mail.com", "state": "DE", "is_active": "FALSE", "status": "MED  "}, {"department": "Legal", "phone": "534-180-2532", "email": "FKKEDA@EXAMPLE.COM ", "state": "DE", "is_active": "F", "status": "high"}, {"department": "legal ", "phone": "(561)080-9708", "email": "cnmlob@mail.com", "state": "Hawaii", "is_active": "T", "status": "HIGH"}, {"department": " legal ", "phone": "(560)185-7211", "email": "gpcg@test.org", "state": "HI", "is_active": "0", "status": "H"}, {"department": "Biz Dev", "phone": "5872980791", "email": " akfgc@mail.com ", "state": "DE", "is_active": "No", "status": "HIGH"}, {"department": "sales", "phone": "(596)252-5270", "email": "  pghlcl@corp.io", "state": "Hawaii", "is_active": "1", "status": "high"}, {"department": "Biz Dev", "phone": "554.248.8511", "email": "  KHFELAI@MAIL.COM", "state": "Htwaii", "is_active": "N", "status": "p2"}, {"department": "Legl", "phone": "5911336950", "email": "ecmk@test.org", "state": "HI", "is_active": "N", "status": "   P1"}, {"department": "Sales  ", "phone": "(546)580-0529", "email": "oejp@example.com  ", "state": "hawaii", "is_active": "F", "status": "Hiigh"}, {"department": "  Legal", "phone": "5375943223", "email": " dpimanb@test.org", "state": "Delaware", "is_active": "T", "status": "high"}, {"department": "Biz Dve", "phone": "(564)985-8720", "email": "clnfpb@test.org  ", "state": "Delaware", "is_active": "Yes", "status": "  med"}, {"department": "", "phone": "", "email": "", "state": "", "is_active": "", "status": ""}], "clean": [{"department": "Sales", "phone": "(599) 944-7821", "email": "jacodmn@mail.com", "state": "Delaware", "is_active": true, "status": "High"}, {"department": "Legal", "phone": "(584) 565-8853", "email": "aljj@test.org", "state": "Hawaii", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(569) 927-2845", "email": "pbneg@test.org", "state": "Delaware", "is_active": false, "status": "High"}, {"department": "Sales", "phone": "(593) 078-1620", "email": "iagdnjl@example.com", "state": "Delaware", "is_active": true, "status": "Medium"}, {"department": "Sales", "phone": "(579) 675-8558", "email": "fnilbgk@test.org", "state": "Delaware", "is_active": true, "status": "High"}, {"department": "Legal", "phone": "(521) 523-1959", "email": "pabb@corp.io", "state": "Hawaii", "is_active": true, "status": "Medium"}, {"department": "Sales", "phone": "(572) 704-8854", "email": "bgaf@example.com", "state": "Hawaii", "is_active": false, "status": "Medium"}, {"department": "Sales", "phone": "(513) 786-8759", "email": "jnmbpn@corp.io", "state": "Hawaii", "is_active": true, "status": "High"}, {"department": "Sales", "phone": "(561) 432-5007", "email": "aickd@test.org", "state": "Hawaii", "is_active": false, "status": "Medium"}, {"department": "Sales", "phone": "(566) 972-4407", "email": "maaklkf@corp.io", "state": "Hawaii", "is_active": false, "status": "Medium"}, {"department": "Sales", "phone": "(573) 810-3773", "email": "enmdf@corp.io", "state": "Delaware", "is_active": true, "status": "High"}, {"department": "Legal", "phone": "(594) 527-6881", "email": "pfodhj@example.com", "state": "Hawaii", "is_active": true, "status": "Medium"}, {"department": "Sales", "phone": "(535) 456-4116", "email": "mcncfoo@corp.io", "state": "Delaware", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(519) 181-6612", "email": "ecnad@corp.io", "state": "Delaware", "is_active": true, "status": "Medium"}, {"department": "Sales", "phone": "(593) 239-4767", "email": "cjmolf@example.com", "state": "Delaware", "is_active": true, "status": "Medium"}, {"department": "Sales", "phone": "(537) 002-7902", "email": "lfod@test.org", "state": "Delaware", "is_active": true, "status": "High"}, {"department": "Sales", "phone": "(536) 057-6770", "email": "ogoeolj@corp.io", "state": "Hawaii", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(515) 171-6536", "email": "clekgkc@test.org", "state": "Delaware", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(589) 468-4307", "email": "mgop@example.com", "state": "Hawaii", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(564) 306-6010", "email": "mgoo@mail.com", "state": "Delaware", "is_active": false, "status": "Medium"}, {"department": "Sales", "phone": "(531) 564-8635", "email": "jgeijo@example.com", "state": "Hawaii", "is_active": true, "status": "High"}, {"department": "Legal", "phone": "(519) 479-5318", "email": "mloh@mail.com", "state": "Delaware", "is_active": false, "status": "Medium"}, {"department": "Legal", "phone": "(534) 180-2532", "email": "fkkeda@example.com", "state": "Delaware", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(561) 080-9708", "email": "cnmlob@mail.com", "state": "Hawaii", "is_active": true, "status": "High"}, {"department": "Legal", "phone": "(560) 185-7211", "email": "gpcg@test.org", "state": "Hawaii", "is_active": false, "status": "High"}, {"department": "Sales", "phone": "(587) 298-0791", "email": "akfgc@mail.com", "state": "Delaware", "is_active": false, "status": "High"}, {"department": "Sales", "phone": "(596) 252-5270", "email": "pghlcl@corp.io", "state": "Hawaii", "is_active": true, "status": "High"}, {"department": "Sales", "phone": "(554) 248-8511", "email": "khfelai@mail.com", "state": "Hawaii", "is_active": false, "status": "Medium"}, {"department": "Legal", "phone": "(591) 133-6950", "email": "ecmk@test.org", "state": "Hawaii", "is_active": false, "status": "High"}, {"department": "Sales", "phone": "(546) 580-0529", "email": "oejp@example.com", "state": "Hawaii", "is_active": false, "status": "High"}, {"department": "Legal", "phone": "(537) 594-3223", "email": "dpimanb@test.org", "state": "Delaware", "is_active": true, "status": "High"}, {"department": "Sales", "phone": "(564) 985-8720", "email": "clnfpb@test.org", "state": "Delaware", "is_active": true, "status": "Medium"}], "dirty_cols": ["department", "phone", "email", "state", "is_active", "status"], "clean_cols": ["department", "phone", "email", "state", "is_active", "status"], "plan": {"dataset_summary": "33 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"legal": "Legal", "BIZ DEV": "Sales", "Biz Dev": "Sales", "SLAES": "Sales", "legal & compliance": "Legal", "sales": "Sales", "LEGAL & COPMLIANCE": "Legal", "Legl": "Legal", "Biz Dve": "Sales"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"De": "Delaware", "HI": "Hawaii", "delaware": "Delaware", "DE": "Delaware", "hawaii": "Hawaii", "HAWAII": "Hawaii", "hi": "Hawaii", "Htwaii": "Hawaii"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HIGH": "High", "P1": "High", "P2": "Medium", "H": "High", "medium": "Medium", "medimu": "Medium", "HIIGH": "High", "med": "Medium", "high": "High", "h": "High", "M": "Medium", "MED": "Medium", "p2": "Medium", "Hiigh": "High"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "Colombian Peso", "amount": "7.601,61", "industry": " SOFTWARE ", "phone": "(541)900-9913", "department": "Legal", "city": "Sundbyberg", "extra": ""}, {"currency": "CDF", "amount": "--", "industry": "Tech", "phone": "(547)282-8643", "department": "legal & compliance", "city": " Boromo ", "extra": ""}, {"currency": "COP", "amount": "317,09", "industry": "Oil & Gas", "phone": "579-818-8173", "department": "HR", "city": "sundbyberg", "extra": ""}, {"currency": "colommbian peso", "amount": "5.948,88", "industry": "Cotms", "phone": "511-799-9029", "department": "Legal", "city": " Yuulong ", "extra": ""}, {"currency": "   cdf", "amount": "7.303,27", "industry": "RealEstate", "phone": "500.909.4802", "department": " PEOPLE ", "city": "  Boromo", "extra": ""}, {"currency": "CONGOLESE FRANC", "amount": "3.439,33", "industry": " Comms ", "phone": "5849635142", "department": "Leegal", "city": "Trenque  Lauquen", "extra": ""}, {"currency": "colombian peso", "amount": "5.067,81", "industry": "utilities", "phone": "(534)213-2931", "department": " HR", "city": "Trenque Lauquen", "extra": ""}, {"currency": "CONGOLESE FRANC", "amount": "8.657,97", "industry": "property", "phone": "5744184949", "department": "hr", "city": "Boromo", "extra": ""}, {"currency": "cop", "amount": "$7,219.15", "industry": " PropTech ", "phone": "(557)075-3026", "department": "People  Ops", "city": "Sundbyberg", "extra": ""}, {"currency": "Congolese Franc", "amount": "$4,468.28", "industry": "teelecom", "phone": "5915806631", "department": "People Ops", "city": "   Sundbyberg", "extra": ""}, {"currency": "colombian peso", "amount": "1.325,25", "industry": "Soctware", "phone": "(535)907-4911", "department": "   Legal", "city": "trenque lauquen", "extra": ""}, {"currency": "Cdf", "amount": "8.246,73", "industry": "PROPTECH", "phone": "(508)098-1298", "department": "Legal", "city": "trenque lauquen", "extra": ""}, {"currency": " Cdf", "amount": "6.243,67", "industry": "TELECOM", "phone": "590.567.0048", "department": "PEOPLE OPS", "city": "Boromo", "extra": ""}, {"currency": "Colomiban Peso", "amount": "8.371,44", "industry": "real estate", "phone": "(529)878-9245", "department": "HR", "city": "TRENQUE LAUQUEN", "extra": ""}, {"currency": "cop", "amount": "$536.78", "industry": "Energy", "phone": "(556)795-7094", "department": "legal & compilance", "city": "trenque lauquen", "extra": ""}, {"currency": "CDF", "amount": "$8,128.56", "industry": "tech", "phone": "522.881.6434", "department": "legal & compliance", "city": "Yulong", "extra": ""}, {"currency": "  Congolese Franc", "amount": "4.868,21", "industry": "energy", "phone": "(586)645-0582", "department": "  HR", "city": "SUNDBYBERG", "extra": ""}, {"currency": " cdf", "amount": "850,61", "industry": " TECH ", "phone": "5869206326", "department": "People", "city": " SUNDBYBERRG ", "extra": ""}, {"currency": "Congolese Franc", "amount": "$7,368.44", "industry": "technology", "phone": "5371756960", "department": "legal & compliance", "city": "BOROMO", "extra": ""}, {"currency": "cop", "amount": "5.893,11", "industry": "RealEstate   ", "phone": "544.491.3892", "department": "legal", "city": "  Yulong", "extra": ""}, {"currency": "colombian peso", "amount": "$4,427.25", "industry": "Technology", "phone": "572.803.4441", "department": " Human Resources", "city": "Sundbyberg", "extra": ""}, {"currency": "   cdf", "amount": "7.303,27", "industry": "RealEstate", "phone": "500.909.4802", "department": " PEOPLE ", "city": "  Boromo", "extra": ""}, {"currency": "", "amount": "", "industry": "", "phone": "", "department": "", "city": "", "extra": ""}], "clean": [{"currency": "COP", "amount": 7601.61, "industry": "Technology", "phone": "(541) 900-9913", "department": "Legal", "city": "Sundbyberg"}, {"currency": "CDF", "amount": NaN, "industry": "Technology", "phone": "(547) 282-8643", "department": "Legal", "city": "Boromo"}, {"currency": "COP", "amount": 317.09, "industry": "Energy", "phone": "(579) 818-8173", "department": "Human Resources", "city": "Sundbyberg"}, {"currency": "COP", "amount": 5948.88, "industry": "Telecommunications", "phone": "(511) 799-9029", "department": "Legal", "city": "Yulong"}, {"currency": "CDF", "amount": 7303.27, "industry": "Real Estate", "phone": "(500) 909-4802", "department": "Human Resources", "city": "Boromo"}, {"currency": "CDF", "amount": 3439.33, "industry": "Telecommunications", "phone": "(584) 963-5142", "department": "Legal", "city": "Trenque Lauquen"}, {"currency": "COP", "amount": 5067.81, "industry": "Energy", "phone": "(534) 213-2931", "department": "Human Resources", "city": "Trenque Lauquen"}, {"currency": "CDF", "amount": 8657.97, "industry": "Real Estate", "phone": "(574) 418-4949", "department": "Human Resources", "city": "Boromo"}, {"currency": "COP", "amount": 7219.15, "industry": "Real Estate", "phone": "(557) 075-3026", "department": "Human Resources", "city": "Sundbyberg"}, {"currency": "CDF", "amount": 4468.28, "industry": "Telecommunications", "phone": "(591) 580-6631", "department": "Human Resources", "city": "Sundbyberg"}, {"currency": "COP", "amount": 1325.25, "industry": "Technology", "phone": "(535) 907-4911", "department": "Legal", "city": "Trenque Lauquen"}, {"currency": "CDF", "amount": 8246.73, "industry": "Real Estate", "phone": "(508) 098-1298", "department": "Legal", "city": "Trenque Lauquen"}, {"currency": "CDF", "amount": 6243.67, "industry": "Telecommunications", "phone": "(590) 567-0048", "department": "Human Resources", "city": "Boromo"}, {"currency": "COP", "amount": 8371.44, "industry": "Real Estate", "phone": "(529) 878-9245", "department": "Human Resources", "city": "Trenque Lauquen"}, {"currency": "COP", "amount": 536.78, "industry": "Energy", "phone": "(556) 795-7094", "department": "Legal", "city": "Trenque Lauquen"}, {"currency": "CDF", "amount": 8128.56, "industry": "Technology", "phone": "(522) 881-6434", "department": "Legal", "city": "Yulong"}, {"currency": "CDF", "amount": 4868.21, "industry": "Energy", "phone": "(586) 645-0582", "department": "Human Resources", "city": "Sundbyberg"}, {"currency": "CDF", "amount": 850.61, "industry": "Technology", "phone": "(586) 920-6326", "department": "Human Resources", "city": "Sundbyberg"}, {"currency": "CDF", "amount": 7368.44, "industry": "Technology", "phone": "(537) 175-6960", "department": "Legal", "city": "Boromo"}, {"currency": "COP", "amount": 5893.11, "industry": "Real Estate", "phone": "(544) 491-3892", "department": "Legal", "city": "Yulong"}, {"currency": "COP", "amount": 4427.25, "industry": "Technology", "phone": "(572) 803-4441", "department": "Human Resources", "city": "Sundbyberg"}], "dirty_cols": ["currency", "amount", "industry", "phone", "department", "city", "extra"], "clean_cols": ["currency", "amount", "industry", "phone", "department", "city"], "plan": {"dataset_summary": "23 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Colombian Peso": "COP", "colommbian peso": "COP", "cdf": "CDF", "CONGOLESE FRANC": "CDF", "colombian peso": "COP", "cop": "COP", "Congolese Franc": "CDF", "Cdf": "CDF", "Colomiban Peso": "COP"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SOFTWARE": "Technology", "Tech": "Technology", "Oil & Gas": "Energy", "Cotms": "Telecommunications", "RealEstate": "Real Estate", "Comms": "Telecommunications", "utilities": "Energy", "property": "Real Estate", "PropTech": "Real Estate", "teelecom": "Telecommunications", "Soctware": "Technology", "PROPTECH": "Real Estate", "TELECOM": "Telecommunications", "real estate": "Real Estate", "tech": "Technology", "energy": "Energy", "TECH": "Technology", "technology": "Technology"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"legal & compliance": "Legal", "HR": "Human Resources", "PEOPLE": "Human Resources", "Leegal": "Legal", "hr": "Human Resources", "People Ops": "Human Resources", "PEOPLE OPS": "Human Resources", "legal & compilance": "Legal", "People": "Human Resources", "legal": "Legal"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"sundbyberg": "Sundbyberg", "Yuulong": "Yulong", "trenque lauquen": "Trenque Lauquen", "TRENQUE LAUQUEN": "Trenque Lauquen", "SUNDBYBERG": "Sundbyberg", "SUNDBYBERRG": "Sundbyberg", "BOROMO": "Boromo"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "AUD", "state": " Pennsylvania ", "phone": "529.846.0456", "country": " Mwi", "unit": "inch", "amount": "6.413,29", "city": "an muileann gcearr"}, {"currency": "BYN", "state": "IN", "phone": "(539)338-5722", "country": "   Nigeria", "unit": "  in", "amount": "$1,076.43", "city": "an muileann gcearr"}, {"currency": "Australian Dollar", "state": "   New Mexico", "phone": "5238775371", "country": "nga", "unit": " In ", "amount": "$2,514.89", "city": "An Muileann gCearr"}, {"currency": "AUD", "state": "in", "phone": "531.921.5885", "country": "FEDERAL REPUBLIC OF  NIGERIA", "unit": "   kg", "amount": "1.000,47", "city": "Fortress Hill"}, {"currency": "clp", "state": " indiana ", "phone": "5390365053", "country": "MWI", "unit": "in", "amount": "$7,642.98", "city": "Fortress Hill "}, {"currency": "AUD", "state": "Nebraska", "phone": "5539650110", "country": "ng", "unit": "Kg", "amount": "8.420,32", "city": "An Muileann Gcearr"}, {"currency": "aud", "state": " IN ", "phone": "584.018.2808", "country": "Republic of Malawi", "unit": "Mi", "amount": "7.036,91", "city": "Fortress Hill"}, {"currency": "Bahamian Dollar ", "state": "Pennsylvania", "phone": "595.422.8287", "country": "NG", "unit": " MI ", "amount": "6.554,93", "city": "an muileann gcearr"}, {"currency": "AUD", "state": "Nm", "phone": "550.568.8226", "country": "MWI", "unit": "mi", "amount": "$5,863.41", "city": "Fortrses Hill"}, {"currency": "BSD", "state": "IN", "phone": "569.005.5205", "country": "federal republic of nigeria", "unit": "   INCH", "amount": "2.700,61", "city": "Fortress Hill"}, {"currency": "AUD", "state": "NEBRASKA", "phone": "5831867691", "country": "federal republic of nigeria ", "unit": "miles", "amount": "?", "city": "Fortress Hill"}, {"currency": "AUD", "state": "nm", "phone": "579-867-4706", "country": "nga", "unit": "mi", "amount": "$7,513.41", "city": "Fortress Hill   "}, {"currency": "Cihlean Peso", "state": "IN   ", "phone": "(564)898-4419", "country": " nigeria", "unit": "inches", "amount": "--", "city": "An Muileann gCearr"}, {"currency": "chilean peso", "state": "ne", "phone": "(553)489-8969", "country": " republic of malawi", "unit": "milse ", "amount": "null", "city": "Foortress Hill"}, {"currency": "BELARUSIAN RUBLE", "state": "Indiana", "phone": "(571)514-5465", "country": "mw", "unit": "mi", "amount": "8.944,81", "city": " fortress hill"}, {"currency": "Chilean Peso", "state": "ne", "phone": "5421202289", "country": "NGA  ", "unit": "Mi", "amount": "$618.20", "city": "An Muileann gCearr"}, {"currency": "  australian dollar", "state": "PA", "phone": "5325273579", "country": "  nigeria", "unit": "mi", "amount": "$3,188.62", "city": "fortress hill"}, {"currency": "byn", "state": " PA ", "phone": "(512)629-0093", "country": "Federal Republic of Nigeria", "unit": "Mi", "amount": "$3,603.21", "city": "Fortress Hill"}, {"currency": "BYN", "state": "Pa", "phone": "(555)181-3223", "country": "NGA", "unit": "KG", "amount": "0", "city": "Fortress Hill"}, {"currency": "BSD   ", "state": "AL", "phone": "576.576.8970", "country": "Federal Republic of Nigeria", "unit": " inch ", "amount": "null", "city": "An Muileann gCearr"}, {"currency": "BYN", "state": "nebraska", "phone": "578.904.3747", "country": "republiczof malawi", "unit": " mile", "amount": "8.838,57", "city": "An Muileann gCearr"}, {"currency": "BYN", "state": "Pennsylvania   ", "phone": "522-029-7948", "country": "Federal Republic of Nigeria", "unit": "kiilogram", "amount": "4.653,17", "city": "Fortress Hill"}, {"currency": "Australian Dollar", "state": "New  Mexico", "phone": "534-017-9207", "country": "Federal Republic of Nigeria", "unit": "inch", "amount": "$5,050.55", "city": "Fortress Hill"}, {"currency": "BSD", "state": "NEW MEICO   ", "phone": "(535)858-6352", "country": "NG", "unit": "mle  ", "amount": "8.497,44", "city": "An Muileann gCearr"}, {"currency": "Clp", "state": "New  Mexico", "phone": "510-971-7277", "country": "Malawi", "unit": " mi ", "amount": "7.573,57", "city": "Fortress Hill"}, {"currency": "Chilean Peso", "state": "IN", "phone": "(576)749-8259", "country": "federal republic of nigeria", "unit": "mile", "amount": "$6,216.40", "city": "Fortress Hill"}, {"currency": "Bahamian Dollar ", "state": "Pennsylvania", "phone": "595.422.8287", "country": "NG", "unit": " MI ", "amount": "6.554,93", "city": "an muileann gcearr"}], "clean": [{"currency": "AUD", "state": "Pennsylvania", "phone": "(529) 846-0456", "country": "Malawi", "unit": "in", "amount": 6413.29, "city": "An Muileann gCearr"}, {"currency": "BYN", "state": "Indiana", "phone": "(539) 338-5722", "country": "Nigeria", "unit": "in", "amount": 1076.43, "city": "An Muileann gCearr"}, {"currency": "AUD", "state": "New Mexico", "phone": "(523) 877-5371", "country": "Nigeria", "unit": "in", "amount": 2514.89, "city": "An Muileann gCearr"}, {"currency": "AUD", "state": "Indiana", "phone": "(531) 921-5885", "country": "Nigeria", "unit": "kg", "amount": 1000.47, "city": "Fortress Hill"}, {"currency": "CLP", "state": "Indiana", "phone": "(539) 036-5053", "country": "Malawi", "unit": "in", "amount": 7642.98, "city": "Fortress Hill"}, {"currency": "AUD", "state": "Nebraska", "phone": "(553) 965-0110", "country": "Nigeria", "unit": "kg", "amount": 8420.32, "city": "An Muileann gCearr"}, {"currency": "AUD", "state": "Indiana", "phone": "(584) 018-2808", "country": "Malawi", "unit": "mi", "amount": 7036.91, "city": "Fortress Hill"}, {"currency": "BSD", "state": "Pennsylvania", "phone": "(595) 422-8287", "country": "Nigeria", "unit": "mi", "amount": 6554.93, "city": "An Muileann gCearr"}, {"currency": "AUD", "state": "New Mexico", "phone": "(550) 568-8226", "country": "Malawi", "unit": "mi", "amount": 5863.41, "city": "Fortress Hill"}, {"currency": "BSD", "state": "Indiana", "phone": "(569) 005-5205", "country": "Nigeria", "unit": "in", "amount": 2700.61, "city": "Fortress Hill"}, {"currency": "AUD", "state": "Nebraska", "phone": "(583) 186-7691", "country": "Nigeria", "unit": "mi", "amount": NaN, "city": "Fortress Hill"}, {"currency": "AUD", "state": "New Mexico", "phone": "(579) 867-4706", "country": "Nigeria", "unit": "mi", "amount": 7513.41, "city": "Fortress Hill"}, {"currency": "CLP", "state": "Indiana", "phone": "(564) 898-4419", "country": "Nigeria", "unit": "in", "amount": NaN, "city": "An Muileann gCearr"}, {"currency": "CLP", "state": "Nebraska", "phone": "(553) 489-8969", "country": "Malawi", "unit": "mi", "amount": NaN, "city": "Fortress Hill"}, {"currency": "BYN", "state": "Indiana", "phone": "(571) 514-5465", "country": "Malawi", "unit": "mi", "amount": 8944.81, "city": "Fortress Hill"}, {"currency": "CLP", "state": "Nebraska", "phone": "(542) 120-2289", "country": "Nigeria", "unit": "mi", "amount": 618.2, "city": "An Muileann gCearr"}, {"currency": "AUD", "state": "Pennsylvania", "phone": "(532) 527-3579", "country": "Nigeria", "unit": "mi", "amount": 3188.62, "city": "Fortress Hill"}, {"currency": "BYN", "state": "Pennsylvania", "phone": "(512) 629-0093", "country": "Nigeria", "unit": "mi", "amount": 3603.21, "city": "Fortress Hill"}, {"currency": "BYN", "state": "Pennsylvania", "phone": "(555) 181-3223", "country": "Nigeria", "unit": "kg", "amount": 0.0, "city": "Fortress Hill"}, {"currency": "BSD", "state": "Alabama", "phone": "(576) 576-8970", "country": "Nigeria", "unit": "in", "amount": NaN, "city": "An Muileann gCearr"}, {"currency": "BYN", "state": "Nebraska", "phone": "(578) 904-3747", "country": "Malawi", "unit": "mi", "amount": 8838.57, "city": "An Muileann gCearr"}, {"currency": "BYN", "state": "Pennsylvania", "phone": "(522) 029-7948", "country": "Nigeria", "unit": "kg", "amount": 4653.17, "city": "Fortress Hill"}, {"currency": "AUD", "state": "New Mexico", "phone": "(534) 017-9207", "country": "Nigeria", "unit": "in", "amount": 5050.55, "city": "Fortress Hill"}, {"currency": "BSD", "state": "New Mexico", "phone": "(535) 858-6352", "country": "Nigeria", "unit": "mi", "amount": 8497.44, "city": "An Muileann gCearr"}, {"currency": "CLP", "state": "New Mexico", "phone": "(510) 971-7277", "country": "Malawi", "unit": "mi", "amount": 7573.57, "city": "Fortress Hill"}, {"currency": "CLP", "state": "Indiana", "phone": "(576) 749-8259", "country": "Nigeria", "unit": "mi", "amount": 6216.4, "city": "Fortress Hill"}], "dirty_cols": ["currency", "state", "phone", "country", "unit", "amount", "city"], "clean_cols": ["currency", "state", "phone", "country", "unit", "amount", "city"], "plan": {"dataset_summary": "27 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Australian Dollar": "AUD", "clp": "CLP", "aud": "AUD", "Bahamian Dollar": "BSD", "Cihlean Peso": "CLP", "chilean peso": "CLP", "BELARUSIAN RUBLE": "BYN", "Chilean Peso": "CLP", "australian dollar": "AUD", "byn": "BYN", "Clp": "CLP"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IN": "Indiana", "in": "Indiana", "indiana": "Indiana", "Nm": "New Mexico", "NEBRASKA": "Nebraska", "nm": "New Mexico", "ne": "Nebraska", "PA": "Pennsylvania", "Pa": "Pennsylvania", "AL": "Alabama", "nebraska": "Nebraska", "NEW MEICO": "New Mexico"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mwi": "Malawi", "nga": "Nigeria", "FEDERAL REPUBLIC OF NIGERIA": "Nigeria", "MWI": "Malawi", "ng": "Nigeria", "Republic of Malawi": "Malawi", "NG": "Nigeria", "federal republic of nigeria": "Nigeria", "nigeria": "Nigeria", "republic of malawi": "Malawi", "mw": "Malawi", "NGA": "Nigeria", "Federal Republic of Nigeria": "Nigeria", "republiczof malawi": "Malawi"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"inch": "in", "In": "in", "Kg": "kg", "Mi": "mi", "MI": "mi", "INCH": "in", "miles": "mi", "inches": "in", "milse": "mi", "KG": "kg", "mile": "mi", "kiilogram": "kg", "mle": "mi"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"an muileann gcearr": "An Muileann gCearr", "An Muileann Gcearr": "An Muileann gCearr", "Fortrses Hill": "Fortress Hill", "Foortress Hill": "Fortress Hill", "fortress hill": "Fortress Hill"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 0 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"state": "  united states minor outlying islands", "company": "Acme Inc ", "city": "Mao", "unit": "mile   ", "col_x": ""}, {"state": " MS", "company": "Acme Inc", "city": "Mao", "unit": "Lbs", "col_x": ""}, {"state": "Oklahoma", "company": " Umbrella ", "city": "BACCHUS MARSH", "unit": " cm ", "col_x": ""}, {"state": "UM", "company": "Globex", "city": "Bacchus Marsh", "unit": "lb", "col_x": ""}, {"state": "United States Minor  Outlying Islands", "company": " Vehement", "city": "bacchus marsh", "unit": "mile  ", "col_x": ""}, {"state": "UM", "company": "Umbrella", "city": "Mao", "unit": "MILE", "col_x": ""}, {"state": "OK", "company": "Hooli   ", "city": "Mao", "unit": " lbs ", "col_x": ""}, {"state": "wv", "company": "Initech   ", "city": "Bacchus Marsh", "unit": "mi", "col_x": ""}, {"state": " UM ", "company": "Initech", "city": "Mao", "unit": "mi   ", "col_x": ""}, {"state": "oklkhoma", "company": "   Soylent Corp", "city": "BACCHUSM ARSH", "unit": "miles", "col_x": ""}, {"state": "ms", "company": "   Umbrella", "city": "bacchus marsh", "unit": "CM", "col_x": ""}, {"state": "MISSOURI", "company": " Vehement ", "city": "mao", "unit": "  centimeter", "col_x": ""}, {"state": "OKLAHOMA", "company": "Wonka  Co", "city": "Bacchus Marsh", "unit": " kg ", "col_x": ""}, {"state": "MO", "company": "Stark Industries  ", "city": "bcchus marsh", "unit": "cm", "col_x": ""}, {"state": "MS", "company": "Initech", "city": " Bacchus Marsh", "unit": "Mi", "col_x": ""}, {"state": "MS", "company": " Globex ", "city": "Mao", "unit": "kilograms", "col_x": ""}, {"state": "ok", "company": "Initech", "city": "MAO ", "unit": "Cm", "col_x": ""}], "clean": [{"state": "United States Minor Outlying Islands", "company": "Acme Inc", "city": "Mao", "unit": "mi"}, {"state": "Mississippi", "company": "Acme Inc", "city": "Mao", "unit": "lb"}, {"state": "Oklahoma", "company": "Umbrella", "city": "Bacchus Marsh", "unit": "cm"}, {"state": "United States Minor Outlying Islands", "company": "Globex", "city": "Bacchus Marsh", "unit": "lb"}, {"state": "United States Minor Outlying Islands", "company": "Vehement", "city": "Bacchus Marsh", "unit": "mi"}, {"state": "United States Minor Outlying Islands", "company": "Umbrella", "city": "Mao", "unit": "mi"}, {"state": "Oklahoma", "company": "Hooli", "city": "Mao", "unit": "lb"}, {"state": "West Virginia", "company": "Initech", "city": "Bacchus Marsh", "unit": "mi"}, {"state": "United States Minor Outlying Islands", "company": "Initech", "city": "Mao", "unit": "mi"}, {"state": "Oklahoma", "company": "Soylent Corp", "city": "Bacchus Marsh", "unit": "mi"}, {"state": "Mississippi", "company": "Umbrella", "city": "Bacchus Marsh", "unit": "cm"}, {"state": "Missouri", "company": "Vehement", "city": "Mao", "unit": "cm"}, {"state": "Oklahoma", "company": "Wonka Co", "city": "Bacchus Marsh", "unit": "kg"}, {"state": "Missouri", "company": "Stark Industries", "city": "Bacchus Marsh", "unit": "cm"}, {"state": "Mississippi", "company": "Initech", "city": "Bacchus Marsh", "unit": "mi"}, {"state": "Mississippi", "company": "Globex", "city": "Mao", "unit": "kg"}, {"state": "Oklahoma", "company": "Initech", "city": "Mao", "unit": "cm"}], "dirty_cols": ["state", "company", "city", "unit", "col_x"], "clean_cols": ["state", "company", "city", "unit"], "plan": {"dataset_summary": "17 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"united states minor outlying islands": "United States Minor Outlying Islands", "MS": "Mississippi", "UM": "United States Minor Outlying Islands", "OK": "Oklahoma", "wv": "West Virginia", "oklkhoma": "Oklahoma", "ms": "Mississippi", "MISSOURI": "Missouri", "OKLAHOMA": "Oklahoma", "MO": "Missouri", "ok": "Oklahoma"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BACCHUS MARSH": "Bacchus Marsh", "bacchus marsh": "Bacchus Marsh", "BACCHUSM ARSH": "Bacchus Marsh", "mao": "Mao", "bcchus marsh": "Bacchus Marsh", "MAO": "Mao"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mile": "mi", "Lbs": "lb", "MILE": "mi", "lbs": "lb", "miles": "mi", "CM": "cm", "centimeter": "cm", "Mi": "mi", "kilograms": "kg", "Cm": "cm"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "Guajr Khan", "phone": "571-962-4070", "name": "Karl Adams", "job_title": "Senior Engineer", "extra": ""}, {"city": "GUJAR KHAN", "phone": "(523)117-7971", "name": "  David Reyes", "job_title": "Vice President", "extra": ""}, {"city": "  BLOEMHOF", "phone": "(580)461-0110", "name": "  Carol Ali", "job_title": "SNR ENGINEER", "extra": ""}, {"city": "Bloemhof", "phone": "527-272-1623", "name": "   Grace Petrov", "job_title": "Vice President", "extra": ""}, {"city": "bloemhof", "phone": "5116642885", "name": "Sara Diaz", "job_title": "Sales Representative", "extra": ""}, {"city": "BLOEMHOF", "phone": "(500)004-1723", "name": "Judy  Johnson", "job_title": "sr engineer", "extra": ""}, {"city": "Gujar Khan", "phone": "584-390-1792", "name": "Sara Brandt  ", "job_title": "Snr Engineer", "extra": ""}, {"city": "gujar  khan", "phone": "544.848.3508", "name": "Priya Reyes", "job_title": "VICE PRESIDENT", "extra": ""}, {"city": "bloemhof", "phone": "579-064-4765", "name": "Alice  Diaz", "job_title": "sales rep", "extra": ""}, {"city": "bloemhof", "phone": "572-129-2037", "name": "Grace Smith  ", "job_title": "SALES REP", "extra": ""}, {"city": " Gujar Khan", "phone": "(550)596-2244", "name": "Lena Fischer ", "job_title": "Vce Pres", "extra": ""}, {"city": "bloemhof", "phone": "5400019558", "name": "Judy Johnson   ", "job_title": "Snr Engineer", "extra": ""}, {"city": "Gujar Khan", "phone": "5830734143", "name": "Ivan Khan", "job_title": "Sr.  Engineer", "extra": ""}, {"city": "bloemhof   ", "phone": "524.517.3060", "name": "Carol  Reyes", "job_title": "Sales Representative", "extra": ""}, {"city": "Bloemhof  ", "phone": "5035145468", "name": "Mona Reyes", "job_title": "Senior Engineer", "extra": ""}, {"city": "Gujar Khan", "phone": "(566)937-4996", "name": "Bob Lee", "job_title": "VP", "extra": ""}, {"city": "GUJAR KHAN", "phone": "5162488336", "name": "Sara Smith", "job_title": "Sales Rep", "extra": ""}, {"city": "Gujar Khan", "phone": "541-387-7816", "name": "Karl Brandt ", "job_title": "SENIOR ENG", "extra": ""}, {"city": "Gujar Khan", "phone": "(546)748-7440", "name": "Heidi  Novak", "job_title": " SR. ENGINEER", "extra": ""}, {"city": "bloemhof", "phone": "(543)895-5941", "name": "Lena Khan", "job_title": "rep", "extra": ""}, {"city": "Gujar Khan", "phone": "(527)304-2384", "name": "Bob Reyes ", "job_title": "sr engineer", "extra": ""}, {"city": "Bloemhof", "phone": "536-068-5406", "name": "  Sara Novak", "job_title": "sales representative", "extra": ""}, {"city": "Gujar Khan ", "phone": "524-220-5394", "name": "Karl Adams", "job_title": "Rep ", "extra": ""}, {"city": "GUJARK HAN", "phone": "540.836.5714", "name": " Ivan Ali", "job_title": "Vice President", "extra": ""}, {"city": "gujar  khan", "phone": "544.848.3508", "name": "Priya Reyes", "job_title": "VICE PRESIDENT", "extra": ""}, {"city": "GUJARK HAN", "phone": "540.836.5714", "name": " Ivan Ali", "job_title": "Vice President", "extra": ""}, {"city": "", "phone": "", "name": "", "job_title": "", "extra": ""}], "clean": [{"city": "Gujar Khan", "phone": "(571) 962-4070", "name": "Karl Adams", "job_title": "Senior Engineer"}, {"city": "Gujar Khan", "phone": "(523) 117-7971", "name": "David Reyes", "job_title": "Vice President"}, {"city": "Bloemhof", "phone": "(580) 461-0110", "name": "Carol Ali", "job_title": "Senior Engineer"}, {"city": "Bloemhof", "phone": "(527) 272-1623", "name": "Grace Petrov", "job_title": "Vice President"}, {"city": "Bloemhof", "phone": "(511) 664-2885", "name": "Sara Diaz", "job_title": "Sales Representative"}, {"city": "Bloemhof", "phone": "(500) 004-1723", "name": "Judy Johnson", "job_title": "Senior Engineer"}, {"city": "Gujar Khan", "phone": "(584) 390-1792", "name": "Sara Brandt", "job_title": "Senior Engineer"}, {"city": "Gujar Khan", "phone": "(544) 848-3508", "name": "Priya Reyes", "job_title": "Vice President"}, {"city": "Bloemhof", "phone": "(579) 064-4765", "name": "Alice Diaz", "job_title": "Sales Representative"}, {"city": "Bloemhof", "phone": "(572) 129-2037", "name": "Grace Smith", "job_title": "Sales Representative"}, {"city": "Gujar Khan", "phone": "(550) 596-2244", "name": "Lena Fischer", "job_title": "Vice President"}, {"city": "Bloemhof", "phone": "(540) 001-9558", "name": "Judy Johnson", "job_title": "Senior Engineer"}, {"city": "Gujar Khan", "phone": "(583) 073-4143", "name": "Ivan Khan", "job_title": "Senior Engineer"}, {"city": "Bloemhof", "phone": "(524) 517-3060", "name": "Carol Reyes", "job_title": "Sales Representative"}, {"city": "Bloemhof", "phone": "(503) 514-5468", "name": "Mona Reyes", "job_title": "Senior Engineer"}, {"city": "Gujar Khan", "phone": "(566) 937-4996", "name": "Bob Lee", "job_title": "Vice President"}, {"city": "Gujar Khan", "phone": "(516) 248-8336", "name": "Sara Smith", "job_title": "Sales Representative"}, {"city": "Gujar Khan", "phone": "(541) 387-7816", "name": "Karl Brandt", "job_title": "Senior Engineer"}, {"city": "Gujar Khan", "phone": "(546) 748-7440", "name": "Heidi Novak", "job_title": "Senior Engineer"}, {"city": "Bloemhof", "phone": "(543) 895-5941", "name": "Lena Khan", "job_title": "Sales Representative"}, {"city": "Gujar Khan", "phone": "(527) 304-2384", "name": "Bob Reyes", "job_title": "Senior Engineer"}, {"city": "Bloemhof", "phone": "(536) 068-5406", "name": "Sara Novak", "job_title": "Sales Representative"}, {"city": "Gujar Khan", "phone": "(524) 220-5394", "name": "Karl Adams", "job_title": "Sales Representative"}, {"city": "Gujar Khan", "phone": "(540) 836-5714", "name": "Ivan Ali", "job_title": "Vice President"}], "dirty_cols": ["city", "phone", "name", "job_title", "extra"], "clean_cols": ["city", "phone", "name", "job_title"], "plan": {"dataset_summary": "27 rows × 5 columns. 4 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Guajr Khan": "Gujar Khan", "GUJAR KHAN": "Gujar Khan", "BLOEMHOF": "Bloemhof", "bloemhof": "Bloemhof", "gujar khan": "Gujar Khan", "GUJARK HAN": "Gujar Khan"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SNR ENGINEER": "Senior Engineer", "sr engineer": "Senior Engineer", "Snr Engineer": "Senior Engineer", "VICE PRESIDENT": "Vice President", "sales rep": "Sales Representative", "SALES REP": "Sales Representative", "Vce Pres": "Vice President", "Sr. Engineer": "Senior Engineer", "VP": "Vice President", "Sales Rep": "Sales Representative", "SENIOR ENG": "Senior Engineer", "SR. ENGINEER": "Senior Engineer", "rep": "Sales Representative", "sales representative": "Sales Representative", "Rep": "Sales Representative"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "Federative Republic of Brazil", "company": " Initech", "currency": "cny", "phone": "570.685.0144", "city": "Koblenz  ", "notes2": ""}, {"country": "Brazil", "company": " Cyberdyne ", "currency": "Fjd", "phone": "554.876.1862", "city": "aix-en-provence", "notes2": ""}, {"country": "kiribait", "company": " Globex ", "currency": "FJD", "phone": "5859958737", "city": "koblenz", "notes2": ""}, {"country": "SAINT PIERRE AND MIQUELON", "company": "Umbrella", "currency": "cny", "phone": "589.121.8618", "city": "Koblenz", "notes2": ""}, {"country": "kiribati", "company": "   Soylent Corp", "currency": "Fiji Dollar", "phone": "538.712.4819", "city": "Aix-En-Provence", "notes2": ""}, {"country": "Tunisia", "company": "Umbrella", "currency": "Fiji Dollar", "phone": "5063342188", "city": "  aixenprovence", "notes2": ""}, {"country": "  pry", "company": "Globex", "currency": "fjd", "phone": "504-089-9784", "city": "Koblenz", "notes2": ""}, {"country": "Paraguay", "company": "Stark Industries", "currency": "FJD", "phone": "5826597386", "city": "Aix-En-Provence", "notes2": ""}, {"country": "KI", "company": "Globex", "currency": "CNY", "phone": "5996970141", "city": "   AIXENPROVVENCE", "notes2": ""}, {"country": "SPM", "company": "Acme  Inc", "currency": "Yuan Renminbi", "phone": "591.143.4110", "city": "Koblenz", "notes2": ""}, {"country": "TUNISIA", "company": "Acme Inc", "currency": "Fiji Dollar", "phone": "5228950225", "city": "Koblenz", "notes2": ""}, {"country": "Brazil", "company": "Umbrella", "currency": "Fiji  Dollar", "phone": "503.407.6435", "city": "  Aix-En-Provence", "notes2": ""}, {"country": "py", "company": " Vehement ", "currency": "yuan renminbi", "phone": "5604856555", "city": "Aix-en-Provence", "notes2": ""}, {"country": " br ", "company": "Acme Inc", "currency": "fjd", "phone": "5616903628", "city": "Koblenz", "notes2": ""}, {"country": "PY", "company": "Acme Inc", "currency": "CNY", "phone": "528.314.9478", "city": "aix-en-provence", "notes2": ""}, {"country": "Republic of Tunisia", "company": "Acme Inc", "currency": "Yuan Renminbi", "phone": "504.965.9546", "city": "Koblenz", "notes2": ""}, {"country": "kir", "company": "Umbrella", "currency": "fiji dollar", "phone": "(561)288-3372", "city": "Koblenz", "notes2": ""}, {"country": "Republic of Kiribati", "company": "Vehement", "currency": "Fiji  Dollar", "phone": "5346603907", "city": "aix-en-provence", "notes2": ""}, {"country": "", "company": "", "currency": "", "phone": "", "city": "", "notes2": ""}], "clean": [{"country": "Brazil", "company": "Initech", "currency": "CNY", "phone": "(570) 685-0144", "city": "Koblenz"}, {"country": "Brazil", "company": "Cyberdyne", "currency": "FJD", "phone": "(554) 876-1862", "city": "Aix-en-Provence"}, {"country": "Kiribati", "company": "Globex", "currency": "FJD", "phone": "(585) 995-8737", "city": "Koblenz"}, {"country": "Saint Pierre and Miquelon", "company": "Umbrella", "currency": "CNY", "phone": "(589) 121-8618", "city": "Koblenz"}, {"country": "Kiribati", "company": "Soylent Corp", "currency": "FJD", "phone": "(538) 712-4819", "city": "Aix-en-Provence"}, {"country": "Tunisia", "company": "Umbrella", "currency": "FJD", "phone": "(506) 334-2188", "city": "Aix-en-Provence"}, {"country": "Paraguay", "company": "Globex", "currency": "FJD", "phone": "(504) 089-9784", "city": "Koblenz"}, {"country": "Paraguay", "company": "Stark Industries", "currency": "FJD", "phone": "(582) 659-7386", "city": "Aix-en-Provence"}, {"country": "Kiribati", "company": "Globex", "currency": "CNY", "phone": "(599) 697-0141", "city": "Aix-en-Provence"}, {"country": "Saint Pierre and Miquelon", "company": "Acme Inc", "currency": "CNY", "phone": "(591) 143-4110", "city": "Koblenz"}, {"country": "Tunisia", "company": "Acme Inc", "currency": "FJD", "phone": "(522) 895-0225", "city": "Koblenz"}, {"country": "Brazil", "company": "Umbrella", "currency": "FJD", "phone": "(503) 407-6435", "city": "Aix-en-Provence"}, {"country": "Paraguay", "company": "Vehement", "currency": "CNY", "phone": "(560) 485-6555", "city": "Aix-en-Provence"}, {"country": "Brazil", "company": "Acme Inc", "currency": "FJD", "phone": "(561) 690-3628", "city": "Koblenz"}, {"country": "Paraguay", "company": "Acme Inc", "currency": "CNY", "phone": "(528) 314-9478", "city": "Aix-en-Provence"}, {"country": "Tunisia", "company": "Acme Inc", "currency": "CNY", "phone": "(504) 965-9546", "city": "Koblenz"}, {"country": "Kiribati", "company": "Umbrella", "currency": "FJD", "phone": "(561) 288-3372", "city": "Koblenz"}, {"country": "Kiribati", "company": "Vehement", "currency": "FJD", "phone": "(534) 660-3907", "city": "Aix-en-Provence"}], "dirty_cols": ["country", "company", "currency", "phone", "city", "notes2"], "clean_cols": ["country", "company", "currency", "phone", "city"], "plan": {"dataset_summary": "19 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Federative Republic of Brazil": "Brazil", "kiribait": "Kiribati", "SAINT PIERRE AND MIQUELON": "Saint Pierre and Miquelon", "kiribati": "Kiribati", "pry": "Paraguay", "KI": "Kiribati", "SPM": "Saint Pierre and Miquelon", "TUNISIA": "Tunisia", "py": "Paraguay", "br": "Brazil", "PY": "Paraguay", "Republic of Tunisia": "Tunisia", "kir": "Kiribati", "Republic of Kiribati": "Kiribati"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cny": "CNY", "Fjd": "FJD", "Fiji Dollar": "FJD", "fjd": "FJD", "Yuan Renminbi": "CNY", "yuan renminbi": "CNY", "fiji dollar": "FJD"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"aix-en-provence": "Aix-en-Provence", "koblenz": "Koblenz", "Aix-En-Provence": "Aix-en-Provence", "aixenprovence": "Aix-en-Provence", "AIXENPROVVENCE": "Aix-en-Provence"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "$4,031.57", "status": "free trial", "phone": "5413836016", "rate": "14.3%"}, {"amount": "8.624,56", "status": " Trial", "phone": "(547)593-7035", "rate": "70.0%"}, {"amount": "7.328,39", "status": "   active", "phone": "5926299658", "rate": "53.1%"}, {"amount": "$287.23", "status": "free trial", "phone": "598-984-4758", "rate": "60.6%"}, {"amount": "$1,362.91", "status": " Paused ", "phone": "(533)897-3263", "rate": "94.1%"}, {"amount": "1.882,32", "status": "Canxelled ", "phone": "516.549.6627", "rate": "73.5%"}, {"amount": "4.081,60", "status": "on hold", "phone": "5588676269", "rate": "51.1%"}, {"amount": "3.481,79", "status": " TRIAL ", "phone": "538-280-7323", "rate": "20.7%"}, {"amount": "None", "status": "cancelled   ", "phone": "589-420-9798", "rate": "88.2%"}, {"amount": "5.770,99", "status": "CANCWLLED", "phone": "559-818-8087", "rate": "45.2%"}, {"amount": "$5,640.85", "status": "active", "phone": "523-422-5555", "rate": "48.0%"}, {"amount": "8.263,41", "status": "Chuurned", "phone": "5578248224", "rate": "97.2%"}, {"amount": "6.262,16", "status": "trial ", "phone": "5062364481", "rate": "14.6%"}, {"amount": "1.463,89", "status": "FREE TRIAL", "phone": "(565)850-5077", "rate": "13.1%"}, {"amount": "2.585,50", "status": "PAUSDE", "phone": "(509)767-3410", "rate": "74.8%"}, {"amount": "$8,404.26", "status": "Active", "phone": "(510)866-6766", "rate": "7.1%"}, {"amount": "$4,200.39", "status": "active", "phone": "524.514.7893", "rate": "71.8%"}, {"amount": "$3,360.68", "status": "Active", "phone": "(526)198-7603", "rate": "31.2%"}, {"amount": "$4,891.86", "status": "ACTIVE", "phone": "(552)321-6894", "rate": "67.7%"}, {"amount": "$1,844.51", "status": " active ", "phone": "506.649.3709", "rate": "8.3%"}, {"amount": "$6,148.28", "status": "CHURNED", "phone": "579-407-2370", "rate": "26.7%"}, {"amount": "$6,148.28", "status": "CHURNED", "phone": "579-407-2370", "rate": "26.7%"}], "clean": [{"amount": 4031.57, "status": "Trial", "phone": "(541) 383-6016", "rate": 0.14300000000000002}, {"amount": 8624.56, "status": "Trial", "phone": "(547) 593-7035", "rate": 0.7}, {"amount": 7328.39, "status": "Active", "phone": "(592) 629-9658", "rate": 0.531}, {"amount": 287.23, "status": "Trial", "phone": "(598) 984-4758", "rate": 0.606}, {"amount": 1362.91, "status": "Paused", "phone": "(533) 897-3263", "rate": 0.941}, {"amount": 1882.32, "status": "Churned", "phone": "(516) 549-6627", "rate": 0.735}, {"amount": 4081.6, "status": "Paused", "phone": "(558) 867-6269", "rate": 0.511}, {"amount": 3481.79, "status": "Trial", "phone": "(538) 280-7323", "rate": 0.207}, {"amount": NaN, "status": "Churned", "phone": "(589) 420-9798", "rate": 0.882}, {"amount": 5770.99, "status": "Churned", "phone": "(559) 818-8087", "rate": 0.452}, {"amount": 5640.85, "status": "Active", "phone": "(523) 422-5555", "rate": 0.48}, {"amount": 8263.41, "status": "Churned", "phone": "(557) 824-8224", "rate": 0.972}, {"amount": 6262.16, "status": "Trial", "phone": "(506) 236-4481", "rate": 0.146}, {"amount": 1463.89, "status": "Trial", "phone": "(565) 850-5077", "rate": 0.131}, {"amount": 2585.5, "status": "Paused", "phone": "(509) 767-3410", "rate": 0.748}, {"amount": 8404.26, "status": "Active", "phone": "(510) 866-6766", "rate": 0.071}, {"amount": 4200.39, "status": "Active", "phone": "(524) 514-7893", "rate": 0.718}, {"amount": 3360.68, "status": "Active", "phone": "(526) 198-7603", "rate": 0.312}, {"amount": 4891.86, "status": "Active", "phone": "(552) 321-6894", "rate": 0.677}, {"amount": 1844.51, "status": "Active", "phone": "(506) 649-3709", "rate": 0.083}, {"amount": 6148.28, "status": "Churned", "phone": "(579) 407-2370", "rate": 0.267}], "dirty_cols": ["amount", "status", "phone", "rate"], "clean_cols": ["amount", "status", "phone", "rate"], "plan": {"dataset_summary": "22 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"free trial": "Trial", "active": "Active", "Canxelled": "Churned", "on hold": "Paused", "TRIAL": "Trial", "cancelled": "Churned", "CANCWLLED": "Churned", "Chuurned": "Churned", "trial": "Trial", "FREE TRIAL": "Trial", "PAUSDE": "Paused", "ACTIVE": "Active", "CHURNED": "Churned"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"city": "bentley", "state": "Northern Mariana Islands", "name": "Eve Petrov"}, {"city": "São Paulo", "state": "AZ", "name": " Priya Johnson"}, {"city": " bentey ", "state": " iowwa ", "name": "Lena Diaz"}, {"city": "naranjal", "state": "MP ", "name": "Frank Lee"}, {"city": "shahecheng", "state": "iowa", "name": "Omar Ali"}, {"city": "Sao Paulo", "state": "  MI", "name": "David Smith"}, {"city": "SP", "state": "Northern Mariana Islands", "name": " Mona Ali"}, {"city": "   Naarnjal", "state": "   AZ", "name": "Frank Park"}, {"city": "naranjal", "state": "MP", "name": "  Bob Reyes"}, {"city": "seattlle", "state": " Iowa ", "name": "Omar Cruz"}, {"city": "Naranjal ", "state": "IA", "name": "Eve Cruz"}, {"city": "   Bentley", "state": "northern mariana islands", "name": "  Priya Park"}, {"city": "sao paulo", "state": "Northern Mariana Islands", "name": "Frank Diaz"}, {"city": "SP", "state": "Iowa", "name": "Mona  Park"}, {"city": "Shahecheng", "state": "  Mi", "name": "David Brandt"}], "clean": [{"city": "Bentley", "state": "Northern Mariana Islands", "name": "Eve Petrov"}, {"city": "Sao Paulo", "state": "Arizona", "name": "Priya Johnson"}, {"city": "Bentley", "state": "Iowa", "name": "Lena Diaz"}, {"city": "Naranjal", "state": "Northern Mariana Islands", "name": "Frank Lee"}, {"city": "Shahecheng", "state": "Iowa", "name": "Omar Ali"}, {"city": "Sao Paulo", "state": "Michigan", "name": "David Smith"}, {"city": "Sao Paulo", "state": "Northern Mariana Islands", "name": "Mona Ali"}, {"city": "Naranjal", "state": "Arizona", "name": "Frank Park"}, {"city": "Naranjal", "state": "Northern Mariana Islands", "name": "Bob Reyes"}, {"city": "Seattle", "state": "Iowa", "name": "Omar Cruz"}, {"city": "Naranjal", "state": "Iowa", "name": "Eve Cruz"}, {"city": "Bentley", "state": "Northern Mariana Islands", "name": "Priya Park"}, {"city": "Sao Paulo", "state": "Northern Mariana Islands", "name": "Frank Diaz"}, {"city": "Sao Paulo", "state": "Iowa", "name": "Mona Park"}, {"city": "Shahecheng", "state": "Michigan", "name": "David Brandt"}], "dirty_cols": ["city", "state", "name"], "clean_cols": ["city", "state", "name"], "plan": {"dataset_summary": "15 rows × 3 columns. 3 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bentley": "Bentley", "São Paulo": "Sao Paulo", "bentey": "Bentley", "naranjal": "Naranjal", "shahecheng": "Shahecheng", "SP": "Sao Paulo", "Naarnjal": "Naranjal", "seattlle": "Seattle", "sao paulo": "Sao Paulo"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AZ": "Arizona", "iowwa": "Iowa", "MP": "Northern Mariana Islands", "iowa": "Iowa", "MI": "Michigan", "IA": "Iowa", "northern mariana islands": "Northern Mariana Islands", "Mi": "Michigan"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"status": "late", "rate": "41.4%", "company": "Stark Industries", "city": "Adelaide", "currency": "  Gibraltar Pound"}, {"status": "overdue", "rate": "19.9%", "company": " Acme Inc", "city": "adelaide", "currency": "wir franc"}, {"status": "refunded   ", "rate": "55.4%", "company": " Umbrella ", "city": "Adelaide", "currency": "Gibraltar  Pownd"}, {"status": "Pending", "rate": "19.9%", "company": " Hooli ", "city": "Dessau", "currency": "UAE Dirham"}, {"status": "Paid", "rate": "13.6%", "company": "Initech", "city": "Calasiao", "currency": "Forint"}, {"status": "refunded", "rate": "29.2%", "company": "Cyberdyne", "city": "DESSAU   ", "currency": "WIR Franc"}, {"status": " paid ", "rate": "66.9%", "company": "Cyberdyne", "city": "Dessau   ", "currency": "HUF"}, {"status": "PAID", "rate": "50.9%", "company": "Initech", "city": "DESSAU", "currency": " AED"}, {"status": "Pendifg", "rate": "66.9%", "company": "Soylent Corp ", "city": "Calasiao", "currency": "uae dirham"}, {"status": "Pid", "rate": "87.3%", "company": " Cyberdyne", "city": "Calasiao", "currency": "HUF"}, {"status": "unpaid", "rate": "74.1%", "company": " Hooli ", "city": "CALASIAO  ", "currency": "Forint   "}, {"status": "refund", "rate": "91.6%", "company": "Stark Industries", "city": "Dessau", "currency": "   Uae Diram"}, {"status": "late  ", "rate": "77.9%", "company": "Wonka Co", "city": "Calasiao", "currency": "huf"}, {"status": "late", "rate": "88.1%", "company": "Wonka  Co", "city": "QUARO", "currency": "  WIR Franc"}, {"status": "overdue ", "rate": "5.9%", "company": "Cyberdyne  ", "city": "adelaide", "currency": "Uae Dirahm"}, {"status": "Overdue", "rate": "64.1%", "company": " Cyberdyne", "city": "Adeladie  ", "currency": "uae dirham"}, {"status": "  refund", "rate": "8.5%", "company": "Soylent Corp", "city": "Calasiao", "currency": " HUF "}, {"status": " paid ", "rate": "8.8%", "company": "Cyberdyne", "city": "QUARTO", "currency": "AED"}, {"status": "Refund", "rate": "16.5%", "company": " Cyberdyne ", "city": "ADELAIDE", "currency": "huf"}, {"status": " overdue", "rate": "90.8%", "company": "Globex", "city": "dessau", "currency": "chw"}, {"status": "Overdue", "rate": "19.3%", "company": "Vehement ", "city": " Adlaide", "currency": "GIP"}, {"status": "refunded", "rate": "6.0%", "company": "Cyberdyne", "city": "ADELAIDE", "currency": "CHW"}, {"status": "PAID", "rate": "53.2%", "company": "Stark Industries", "city": "adelaide", "currency": "Gibraltar Jound"}, {"status": "Refund", "rate": "99.1%", "company": "Vehement", "city": "Quarto", "currency": "HUF "}, {"status": "Overdue", "rate": "43.0%", "company": " Globex", "city": "CALASIAO ", "currency": "CHW"}, {"status": "overdue", "rate": "30.5%", "company": "Umbrella", "city": "Calasiao", "currency": "Gibraltr Pound"}, {"status": "refunded", "rate": "15.0%", "company": "Hooli  ", "city": "QUARTO", "currency": "FORINT"}, {"status": "pending", "rate": "64.8%", "company": "Soylent Corp", "city": "  Adelaide", "currency": "HUF"}, {"status": "pending", "rate": "24.2%", "company": "Hooli", "city": "Adelaide", "currency": "Forint"}, {"status": "overdue", "rate": "25.2%", "company": "Stark Industries", "city": " ADELAIDE ", "currency": "Gibraxtar Pound"}, {"status": " Due", "rate": "50.9%", "company": " Stark Industries", "city": "CALASIOA", "currency": "GIP"}, {"status": " DUE ", "rate": "54.5%", "company": "Acme Inc", "city": "Quarto", "currency": "   CHW"}, {"status": "PENDING", "rate": "62.3%", "company": " Umbrella ", "city": "Adelaide", "currency": "Forint"}, {"status": "Overdue", "rate": "0.5%", "company": "Soylent Corp", "city": "quarto", "currency": "   chw"}, {"status": " unpaid ", "rate": "28.2%", "company": "Cyberdyne", "city": "   calasiao", "currency": "huf"}, {"status": "refunded", "rate": "92.9%", "company": "Soylent Corp", "city": "Adelaide", "currency": "Gip   "}, {"status": "refund", "rate": "74.0%", "company": "Wonka Co  ", "city": "Dessau  ", "currency": "wir  franc"}, {"status": "pendong", "rate": "51.9%", "company": "Stark Industries", "city": "Calasiao", "currency": "huf"}, {"status": "paid", "rate": "55.3%", "company": "Hooli", "city": "adelaide", "currency": "HUF"}, {"status": "Paid", "rate": "11.7%", "company": "  Umbrella", "city": "Dessau", "currency": "  WIR Franc"}, {"status": "overdue", "rate": "61.3%", "company": "Soylent Corp", "city": "ADEAIDE", "currency": "AED   "}, {"status": "overdue", "rate": "19.9%", "company": " Acme Inc", "city": "adelaide", "currency": "wir franc"}, {"status": "paid", "rate": "55.3%", "company": "Hooli", "city": "adelaide", "currency": "HUF"}], "clean": [{"status": "Overdue", "rate": 0.414, "company": "Stark Industries", "city": "Adelaide", "currency": "GIP"}, {"status": "Overdue", "rate": 0.19899999999999998, "company": "Acme Inc", "city": "Adelaide", "currency": "CHW"}, {"status": "Refunded", "rate": 0.5539999999999999, "company": "Umbrella", "city": "Adelaide", "currency": "GIP"}, {"status": "Pending", "rate": 0.19899999999999998, "company": "Hooli", "city": "Dessau", "currency": "AED"}, {"status": "Paid", "rate": 0.136, "company": "Initech", "city": "Calasiao", "currency": "HUF"}, {"status": "Refunded", "rate": 0.292, "company": "Cyberdyne", "city": "Dessau", "currency": "CHW"}, {"status": "Paid", "rate": 0.669, "company": "Cyberdyne", "city": "Dessau", "currency": "HUF"}, {"status": "Paid", "rate": 0.509, "company": "Initech", "city": "Dessau", "currency": "AED"}, {"status": "Pending", "rate": 0.669, "company": "Soylent Corp", "city": "Calasiao", "currency": "AED"}, {"status": "Paid", "rate": 0.873, "company": "Cyberdyne", "city": "Calasiao", "currency": "HUF"}, {"status": "Pending", "rate": 0.741, "company": "Hooli", "city": "Calasiao", "currency": "HUF"}, {"status": "Refunded", "rate": 0.9159999999999999, "company": "Stark Industries", "city": "Dessau", "currency": "AED"}, {"status": "Overdue", "rate": 0.779, "company": "Wonka Co", "city": "Calasiao", "currency": "HUF"}, {"status": "Overdue", "rate": 0.8809999999999999, "company": "Wonka Co", "city": "Quarto", "currency": "CHW"}, {"status": "Overdue", "rate": 0.059000000000000004, "company": "Cyberdyne", "city": "Adelaide", "currency": "AED"}, {"status": "Overdue", "rate": 0.6409999999999999, "company": "Cyberdyne", "city": "Adelaide", "currency": "AED"}, {"status": "Refunded", "rate": 0.085, "company": "Soylent Corp", "city": "Calasiao", "currency": "HUF"}, {"status": "Paid", "rate": 0.08800000000000001, "company": "Cyberdyne", "city": "Quarto", "currency": "AED"}, {"status": "Refunded", "rate": 0.165, "company": "Cyberdyne", "city": "Adelaide", "currency": "HUF"}, {"status": "Overdue", "rate": 0.9079999999999999, "company": "Globex", "city": "Dessau", "currency": "CHW"}, {"status": "Overdue", "rate": 0.193, "company": "Vehement", "city": "Adelaide", "currency": "GIP"}, {"status": "Refunded", "rate": 0.06, "company": "Cyberdyne", "city": "Adelaide", "currency": "CHW"}, {"status": "Paid", "rate": 0.532, "company": "Stark Industries", "city": "Adelaide", "currency": "GIP"}, {"status": "Refunded", "rate": 0.991, "company": "Vehement", "city": "Quarto", "currency": "HUF"}, {"status": "Overdue", "rate": 0.43, "company": "Globex", "city": "Calasiao", "currency": "CHW"}, {"status": "Overdue", "rate": 0.305, "company": "Umbrella", "city": "Calasiao", "currency": "GIP"}, {"status": "Refunded", "rate": 0.15, "company": "Hooli", "city": "Quarto", "currency": "HUF"}, {"status": "Pending", "rate": 0.648, "company": "Soylent Corp", "city": "Adelaide", "currency": "HUF"}, {"status": "Pending", "rate": 0.242, "company": "Hooli", "city": "Adelaide", "currency": "HUF"}, {"status": "Overdue", "rate": 0.252, "company": "Stark Industries", "city": "Adelaide", "currency": "GIP"}, {"status": "Pending", "rate": 0.509, "company": "Stark Industries", "city": "Calasiao", "currency": "GIP"}, {"status": "Pending", "rate": 0.545, "company": "Acme Inc", "city": "Quarto", "currency": "CHW"}, {"status": "Pending", "rate": 0.623, "company": "Umbrella", "city": "Adelaide", "currency": "HUF"}, {"status": "Overdue", "rate": 0.005, "company": "Soylent Corp", "city": "Quarto", "currency": "CHW"}, {"status": "Pending", "rate": 0.282, "company": "Cyberdyne", "city": "Calasiao", "currency": "HUF"}, {"status": "Refunded", "rate": 0.929, "company": "Soylent Corp", "city": "Adelaide", "currency": "GIP"}, {"status": "Refunded", "rate": 0.74, "company": "Wonka Co", "city": "Dessau", "currency": "CHW"}, {"status": "Pending", "rate": 0.519, "company": "Stark Industries", "city": "Calasiao", "currency": "HUF"}, {"status": "Paid", "rate": 0.5529999999999999, "company": "Hooli", "city": "Adelaide", "currency": "HUF"}, {"status": "Paid", "rate": 0.11699999999999999, "company": "Umbrella", "city": "Dessau", "currency": "CHW"}, {"status": "Overdue", "rate": 0.613, "company": "Soylent Corp", "city": "Adelaide", "currency": "AED"}], "dirty_cols": ["status", "rate", "company", "city", "currency"], "clean_cols": ["status", "rate", "company", "city", "currency"], "plan": {"dataset_summary": "43 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"late": "Overdue", "overdue": "Overdue", "refunded": "Refunded", "paid": "Paid", "PAID": "Paid", "Pendifg": "Pending", "Pid": "Paid", "unpaid": "Pending", "refund": "Refunded", "Refund": "Refunded", "pending": "Pending", "Due": "Pending", "DUE": "Pending", "PENDING": "Pending", "pendong": "Pending"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"adelaide": "Adelaide", "DESSAU": "Dessau", "CALASIAO": "Calasiao", "QUARO": "Quarto", "Adeladie": "Adelaide", "QUARTO": "Quarto", "ADELAIDE": "Adelaide", "dessau": "Dessau", "Adlaide": "Adelaide", "CALASIOA": "Calasiao", "quarto": "Quarto", "calasiao": "Calasiao", "ADEAIDE": "Adelaide"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Gibraltar Pound": "GIP", "wir franc": "CHW", "Gibraltar Pownd": "GIP", "UAE Dirham": "AED", "Forint": "HUF", "WIR Franc": "CHW", "uae dirham": "AED", "Uae Diram": "AED", "huf": "HUF", "Uae Dirahm": "AED", "chw": "CHW", "Gibraltar Jound": "GIP", "Gibraltr Pound": "GIP", "FORINT": "HUF", "Gibraxtar Pound": "GIP", "Gip": "GIP"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "CHE   ", "phone": "507.622.8116", "department": "support", "rate": "26.6%"}, {"currency": "DOP", "phone": "5448503812", "department": "mktg", "rate": "16.2%"}, {"currency": "gmd", "phone": "(518)332-7567", "department": "   Sales", "rate": "23.5%"}, {"currency": "Fiji Dollar", "phone": "(527)588-1037", "department": "operations", "rate": "16.5%"}, {"currency": "Fiji Dollar", "phone": "587-195-1490", "department": "CS", "rate": "19.0%"}, {"currency": "  che", "phone": "5480542792", "department": "   Marketing", "rate": "99.1%"}, {"currency": "dominican peso ", "phone": "(538)397-3374", "department": "OPS", "rate": "2.2%"}, {"currency": "fiji dollar", "phone": "(562)247-9756", "department": "sales", "rate": "65.6%"}, {"currency": "dominican  peso", "phone": "557.193.9366", "department": "Marketing", "rate": "93.7%"}, {"currency": "DOP", "phone": "524-739-9572", "department": "Mktg", "rate": "46.5%"}, {"currency": " WIR EURO", "phone": "(508)105-7850", "department": "Sales", "rate": "42.3%"}, {"currency": "fiji dollar", "phone": "(574)853-1375", "department": "Mktg", "rate": "38.5%"}, {"currency": " fjd ", "phone": "5280368290", "department": "OPS ", "rate": "46.2%"}, {"currency": "CHE", "phone": "(569)798-7349", "department": "cust support", "rate": "69.5%"}, {"currency": "WIR Euro", "phone": "574.691.5051", "department": " GROWTH ", "rate": "74.9%"}, {"currency": "FJD", "phone": "504-717-6640", "department": "Mktg", "rate": "17.9%"}, {"currency": "Dop", "phone": "(537)880-4200", "department": " mktg ", "rate": "93.0%"}, {"currency": "dalasi", "phone": "528-422-8835", "department": "Cust  Support", "rate": "2.4%"}, {"currency": "Gmd", "phone": "5001594697", "department": " Marketing ", "rate": "67.6%"}, {"currency": "Fij iDollar", "phone": "(516)876-7261", "department": "SALES   ", "rate": "86.3%"}, {"currency": "DOP", "phone": "524-739-9572", "department": "Mktg", "rate": "46.5%"}], "clean": [{"currency": "CHE", "phone": "(507) 622-8116", "department": "Customer Support", "rate": 0.266}, {"currency": "DOP", "phone": "(544) 850-3812", "department": "Marketing", "rate": 0.162}, {"currency": "GMD", "phone": "(518) 332-7567", "department": "Sales", "rate": 0.235}, {"currency": "FJD", "phone": "(527) 588-1037", "department": "Operations", "rate": 0.165}, {"currency": "FJD", "phone": "(587) 195-1490", "department": "Customer Support", "rate": 0.19}, {"currency": "CHE", "phone": "(548) 054-2792", "department": "Marketing", "rate": 0.991}, {"currency": "DOP", "phone": "(538) 397-3374", "department": "Operations", "rate": 0.022000000000000002}, {"currency": "FJD", "phone": "(562) 247-9756", "department": "Sales", "rate": 0.6559999999999999}, {"currency": "DOP", "phone": "(557) 193-9366", "department": "Marketing", "rate": 0.937}, {"currency": "DOP", "phone": "(524) 739-9572", "department": "Marketing", "rate": 0.465}, {"currency": "CHE", "phone": "(508) 105-7850", "department": "Sales", "rate": 0.423}, {"currency": "FJD", "phone": "(574) 853-1375", "department": "Marketing", "rate": 0.385}, {"currency": "FJD", "phone": "(528) 036-8290", "department": "Operations", "rate": 0.462}, {"currency": "CHE", "phone": "(569) 798-7349", "department": "Customer Support", "rate": 0.695}, {"currency": "CHE", "phone": "(574) 691-5051", "department": "Marketing", "rate": 0.7490000000000001}, {"currency": "FJD", "phone": "(504) 717-6640", "department": "Marketing", "rate": 0.179}, {"currency": "DOP", "phone": "(537) 880-4200", "department": "Marketing", "rate": 0.93}, {"currency": "GMD", "phone": "(528) 422-8835", "department": "Customer Support", "rate": 0.024}, {"currency": "GMD", "phone": "(500) 159-4697", "department": "Marketing", "rate": 0.6759999999999999}, {"currency": "FJD", "phone": "(516) 876-7261", "department": "Sales", "rate": 0.863}], "dirty_cols": ["currency", "phone", "department", "rate"], "clean_cols": ["currency", "phone", "department", "rate"], "plan": {"dataset_summary": "21 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"gmd": "GMD", "Fiji Dollar": "FJD", "che": "CHE", "dominican peso": "DOP", "fiji dollar": "FJD", "WIR EURO": "CHE", "fjd": "FJD", "WIR Euro": "CHE", "Dop": "DOP", "dalasi": "GMD", "Gmd": "GMD", "Fij iDollar": "FJD"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"support": "Customer Support", "mktg": "Marketing", "operations": "Operations", "CS": "Customer Support", "OPS": "Operations", "sales": "Sales", "Mktg": "Marketing", "cust support": "Customer Support", "GROWTH": "Marketing", "Cust Support": "Customer Support", "SALES": "Sales"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"name": "Lena Petrov ", "phone": "533-928-3458", "is_active": "true", "company": "Globex ", "city": "SF", "col_x": ""}, {"name": "Frank  Ali", "phone": "(556)474-7219", "is_active": "TRUE", "company": "Wonka Co", "city": "  Valparai", "col_x": ""}, {"name": " Karl Brandt", "phone": "5444247985", "is_active": "F", "company": "Hooli ", "city": "Aldaia", "col_x": ""}, {"name": "Frank Petrov ", "phone": "570-111-7318", "is_active": "0", "company": "Soylent Corp", "city": "SF", "col_x": ""}, {"name": "Sara  Ali", "phone": "553.568.3893", "is_active": "F", "company": " Stark Industries", "city": "Aldaia", "col_x": ""}, {"name": "  Grace Petrov", "phone": "594.778.7503", "is_active": "T", "company": "Initech", "city": "Aldaia", "col_x": ""}, {"name": "Karl  Wong", "phone": "562.961.4792", "is_active": "F", "company": "Wonka  Co", "city": "SUNGAI UDANG", "col_x": ""}, {"name": "Heidi Wong", "phone": "(598)907-9274", "is_active": "FALSE", "company": "  Initech", "city": "ILE-IFE", "col_x": ""}, {"name": "Priya Johnson", "phone": "5686244943", "is_active": "TRUE", "company": "Hooli", "city": "Valparai", "col_x": ""}, {"name": "Ivan Adams   ", "phone": "513-406-2479", "is_active": "1", "company": "Stark Industries", "city": "Aldaia", "col_x": ""}], "clean": [{"name": "Lena Petrov", "phone": "(533) 928-3458", "is_active": true, "company": "Globex", "city": "San Francisco"}, {"name": "Frank Ali", "phone": "(556) 474-7219", "is_active": true, "company": "Wonka Co", "city": "Valparai"}, {"name": "Karl Brandt", "phone": "(544) 424-7985", "is_active": false, "company": "Hooli", "city": "Aldaia"}, {"name": "Frank Petrov", "phone": "(570) 111-7318", "is_active": false, "company": "Soylent Corp", "city": "San Francisco"}, {"name": "Sara Ali", "phone": "(553) 568-3893", "is_active": false, "company": "Stark Industries", "city": "Aldaia"}, {"name": "Grace Petrov", "phone": "(594) 778-7503", "is_active": true, "company": "Initech", "city": "Aldaia"}, {"name": "Karl Wong", "phone": "(562) 961-4792", "is_active": false, "company": "Wonka Co", "city": "Sungai Udang"}, {"name": "Heidi Wong", "phone": "(598) 907-9274", "is_active": false, "company": "Initech", "city": "Ile-Ife"}, {"name": "Priya Johnson", "phone": "(568) 624-4943", "is_active": true, "company": "Hooli", "city": "Valparai"}, {"name": "Ivan Adams", "phone": "(513) 406-2479", "is_active": true, "company": "Stark Industries", "city": "Aldaia"}], "dirty_cols": ["name", "phone", "is_active", "company", "city", "col_x"], "clean_cols": ["name", "phone", "is_active", "company", "city"], "plan": {"dataset_summary": "10 rows × 6 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SF": "San Francisco", "SUNGAI UDANG": "Sungai Udang", "ILE-IFE": "Ile-Ife"}, "rationale": "Unified 3 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "Omar Moore ", "department": "operations", "status": "Medium", "job_title": " CEO"}, {"name": "Priya Diaz   ", "department": " CS ", "status": "LOW", "job_title": "Prod Manager"}, {"name": "Mona  Reyes", "department": "  Legal & Compliance", "status": "Low  ", "job_title": "ceo"}, {"name": "Alice Smith", "department": "CS   ", "status": "  med", "job_title": " Prod Mgr"}, {"name": "Omar Moore   ", "department": "OPERATIONS ", "status": "L", "job_title": "CEO"}, {"name": "Mona  Johnson", "department": "Cs", "status": "L", "job_title": "chief executive officer"}, {"name": "Ivan Reyes", "department": "   Cust Support", "status": "low", "job_title": "C.E.O. "}, {"name": "Lena Moore ", "department": "cust suppport", "status": "P2", "job_title": "pm"}, {"name": "Sara  Johnson", "department": "CS", "status": "med", "job_title": "PM"}, {"name": "Heidi  Lee", "department": "Legal & Compliance", "status": "low", "job_title": "Chief Executive Officer   "}, {"name": "Mona  Reyes", "department": "  Legal & Compliance", "status": "Low  ", "job_title": "ceo"}], "clean": [{"name": "Omar Moore", "department": "Operations", "status": "Medium", "job_title": "Chief Executive Officer"}, {"name": "Priya Diaz", "department": "Customer Support", "status": "Low", "job_title": "Product Manager"}, {"name": "Mona Reyes", "department": "Legal", "status": "Low", "job_title": "Chief Executive Officer"}, {"name": "Alice Smith", "department": "Customer Support", "status": "Medium", "job_title": "Product Manager"}, {"name": "Omar Moore", "department": "Operations", "status": "Low", "job_title": "Chief Executive Officer"}, {"name": "Mona Johnson", "department": "Customer Support", "status": "Low", "job_title": "Chief Executive Officer"}, {"name": "Ivan Reyes", "department": "Customer Support", "status": "Low", "job_title": "Chief Executive Officer"}, {"name": "Lena Moore", "department": "Customer Support", "status": "Medium", "job_title": "Product Manager"}, {"name": "Sara Johnson", "department": "Customer Support", "status": "Medium", "job_title": "Product Manager"}, {"name": "Heidi Lee", "department": "Legal", "status": "Low", "job_title": "Chief Executive Officer"}], "dirty_cols": ["name", "department", "status", "job_title"], "clean_cols": ["name", "department", "status", "job_title"], "plan": {"dataset_summary": "11 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"operations": "Operations", "CS": "Customer Support", "Legal & Compliance": "Legal", "OPERATIONS": "Operations", "Cs": "Customer Support", "Cust Support": "Customer Support", "cust suppport": "Customer Support"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LOW": "Low", "med": "Medium", "L": "Low", "low": "Low", "P2": "Medium"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CEO": "Chief Executive Officer", "Prod Manager": "Product Manager", "ceo": "Chief Executive Officer", "Prod Mgr": "Product Manager", "chief executive officer": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer", "pm": "Product Manager", "PM": "Product Manager"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "De", "signup_date": "11 Mar 2023", "status": "Oepn", "currency": "BWP", "is_active": "No", "phone": "558.310.9800", "name": "Ivan Novak", "unit": " centimeter ", "col_x": ""}, {"state": "vermont", "signup_date": "45271", "status": "Lost", "currency": "Fiji Dollar", "is_active": "Yes", "phone": "516.626.6179", "name": " Sara Lee", "unit": "mi", "col_x": ""}, {"state": "Co", "signup_date": "2023-08-17", "status": "New", "currency": "Aud", "is_active": "Yes", "phone": "568-764-7552", "name": "David Novak", "unit": "CENTIMETERS", "col_x": ""}, {"state": "vt", "signup_date": "2023-02-10", "status": "lost", "currency": "WIR Euro", "is_active": "Y", "phone": "592-989-8814", "name": "Priya Moore ", "unit": " cm ", "col_x": ""}, {"state": "CO", "signup_date": "16 Dec 2023", "status": "open", "currency": "wi euro", "is_active": "N", "phone": "(558)257-7085", "name": "David  Brandt", "unit": " mi ", "col_x": ""}, {"state": "Illinois", "signup_date": "2023-06-05", "status": "Closed Lost", "currency": "WIR Euro", "is_active": "false", "phone": "(563)679-0728", "name": "Ivan Wong", "unit": "Mile", "col_x": ""}, {"state": "Vermont", "signup_date": "7/16/2023", "status": "Lot", "currency": "che", "is_active": "TRUE", "phone": "5107301874", "name": "Judy Khan", "unit": "KM", "col_x": ""}, {"state": "CO", "signup_date": "2/18/2023", "status": "lost", "currency": "AUD", "is_active": "Y", "phone": "(595)458-2180", "name": "   Judy Brandt", "unit": "kilomeetr", "col_x": ""}, {"state": "Vermont", "signup_date": "5/18/2023", "status": " Closed Lost", "currency": "che", "is_active": "F", "phone": "5301927658", "name": "Priya Smith", "unit": "cm", "col_x": ""}, {"state": " vt ", "signup_date": "2023-12-21", "status": "lost", "currency": " aud ", "is_active": "N", "phone": "5621004748", "name": "Carol Moore", "unit": " km ", "col_x": ""}, {"state": "vt", "signup_date": "2023-08-02", "status": "new", "currency": "Fiji Dollar", "is_active": "F", "phone": "503-159-5469", "name": "Alice Johnson", "unit": "mi", "col_x": ""}, {"state": "Illinois", "signup_date": "2023-11-14", "status": "Lost", "currency": "AUSTRALIAN DOLLAR", "is_active": "Yes", "phone": "588.220.8602", "name": "Grace  Ali", "unit": "centimeters", "col_x": ""}, {"state": "il", "signup_date": "25 Sep 2023", "status": " Open ", "currency": "fiji dollar", "is_active": "Y", "phone": "512-734-2618", "name": "Frank Diaz  ", "unit": "Mi", "col_x": ""}, {"state": "ILLINOIS", "signup_date": "10/6/2023", "status": "open", "currency": "FJD", "is_active": "N", "phone": "(550)904-0116", "name": "Sara Novak   ", "unit": "MI", "col_x": ""}, {"state": "IL", "signup_date": "10 Nov 2023", "status": "  lost", "currency": "Australian Dollr", "is_active": "F", "phone": "5220594990", "name": " Priya Brandt", "unit": "Cm", "col_x": ""}, {"state": "ILLINOIS  ", "signup_date": "2023-10-16", "status": "  Open", "currency": "CHE", "is_active": "true", "phone": "5437511504", "name": "Alice  Khan", "unit": "Kilometer", "col_x": ""}, {"state": "CO", "signup_date": "45211", "status": "CLOSED LOST   ", "currency": "australian  dollar", "is_active": "FALSE", "phone": "531.060.2109", "name": "Priya Adams", "unit": "mie", "col_x": ""}, {"state": "IL", "signup_date": "2023-08-05", "status": "  Open", "currency": "Australian Dollar", "is_active": "Y", "phone": "558-864-3847", "name": "Grace Moore", "unit": "mi", "col_x": ""}, {"state": "Illinois", "signup_date": "45026", "status": "open", "currency": "  WIR Euro", "is_active": "false", "phone": "501.312.2987", "name": "Judy Cruz", "unit": "Centmieter", "col_x": ""}, {"state": "DE", "signup_date": "23 Nov 2023", "status": "Lost", "currency": "Fiji Dollar", "is_active": "Y", "phone": "537-805-9718", "name": "Carol Lee", "unit": "cm", "col_x": ""}, {"state": "Vermont", "signup_date": "2023-06-02", "status": "  Closed Lost", "currency": "   Pula", "is_active": "T", "phone": "589.309.2417", "name": "Judy Khan", "unit": "mi", "col_x": ""}, {"state": "DE  ", "signup_date": "2023-02-14", "status": "   open", "currency": "FJD", "is_active": "TRUE", "phone": "595.607.9011", "name": "Frank  Novak", "unit": "  mi", "col_x": ""}, {"state": "delaware", "signup_date": "24 Mar 2023", "status": " NEW ", "currency": "Pula", "is_active": "Y", "phone": "5153554217", "name": "Sara Johnson", "unit": "kilometers", "col_x": ""}, {"state": "CO", "signup_date": "9/14/2023", "status": "closed lost", "currency": "AUD", "is_active": "true", "phone": "(570)828-2467", "name": "Grace  Wong", "unit": "km", "col_x": ""}, {"state": "IL", "signup_date": "2023-03-10", "status": "New", "currency": "Australian Dollar", "is_active": "false", "phone": "529-058-3072", "name": "Frank Fischer   ", "unit": "Mi", "col_x": ""}, {"state": "De", "signup_date": "45113", "status": "Open", "currency": "Australian Dollar", "is_active": "1", "phone": "(572)286-1868", "name": "Frank Diaz   ", "unit": "kilometer", "col_x": ""}, {"state": "delaware", "signup_date": "2023-04-22", "status": "   Lost", "currency": "Fiji  Dollar", "is_active": "Yes", "phone": "(575)994-9353", "name": "   David Ali", "unit": "kilometers", "col_x": ""}, {"state": "  il", "signup_date": "22 Nov 2023", "status": " open ", "currency": "WIR uEro", "is_active": "true", "phone": "578.533.7870", "name": "Eve Johnson   ", "unit": "Mi", "col_x": ""}, {"state": "Delaware", "signup_date": "3/24/2023", "status": "lost", "currency": "AUSTRALIAN DOLLAR", "is_active": "0", "phone": "(525)274-3461", "name": "Omar Moore   ", "unit": " kilometers", "col_x": ""}, {"state": "IL", "signup_date": "12 Jan 2023", "status": " Closed ost", "currency": "fjd", "is_active": "No", "phone": "(555)448-5508", "name": "Carol  Brandt", "unit": "kilometer", "col_x": ""}, {"state": "Illinois ", "signup_date": "2023-02-28", "status": "   Lot", "currency": "Australian Dollar", "is_active": "No", "phone": "564-178-2230", "name": "Sara Wong", "unit": "miles", "col_x": ""}, {"state": "VT", "signup_date": "1/23/2023", "status": "new", "currency": "WIR  EURO", "is_active": "N", "phone": "(542)940-4976", "name": "Carol Lee", "unit": "kilometers", "col_x": ""}], "clean": [{"state": "Delaware", "signup_date": "2023-03-11", "status": "Open", "currency": "BWP", "is_active": false, "phone": "(558) 310-9800", "name": "Ivan Novak", "unit": "cm"}, {"state": "Vermont", "signup_date": "2023-12-11", "status": "Lost", "currency": "FJD", "is_active": true, "phone": "(516) 626-6179", "name": "Sara Lee", "unit": "mi"}, {"state": "Colorado", "signup_date": "2023-08-17", "status": "Open", "currency": "AUD", "is_active": true, "phone": "(568) 764-7552", "name": "David Novak", "unit": "cm"}, {"state": "Vermont", "signup_date": "2023-02-10", "status": "Lost", "currency": "CHE", "is_active": true, "phone": "(592) 989-8814", "name": "Priya Moore", "unit": "cm"}, {"state": "Colorado", "signup_date": "2023-12-16", "status": "Open", "currency": "CHE", "is_active": false, "phone": "(558) 257-7085", "name": "David Brandt", "unit": "mi"}, {"state": "Illinois", "signup_date": "2023-06-05", "status": "Lost", "currency": "CHE", "is_active": false, "phone": "(563) 679-0728", "name": "Ivan Wong", "unit": "mi"}, {"state": "Vermont", "signup_date": "2023-07-16", "status": "Lost", "currency": "CHE", "is_active": true, "phone": "(510) 730-1874", "name": "Judy Khan", "unit": "km"}, {"state": "Colorado", "signup_date": "2023-02-18", "status": "Lost", "currency": "AUD", "is_active": true, "phone": "(595) 458-2180", "name": "Judy Brandt", "unit": "km"}, {"state": "Vermont", "signup_date": "2023-05-18", "status": "Lost", "currency": "CHE", "is_active": false, "phone": "(530) 192-7658", "name": "Priya Smith", "unit": "cm"}, {"state": "Vermont", "signup_date": "2023-12-21", "status": "Lost", "currency": "AUD", "is_active": false, "phone": "(562) 100-4748", "name": "Carol Moore", "unit": "km"}, {"state": "Vermont", "signup_date": "2023-08-02", "status": "Open", "currency": "FJD", "is_active": false, "phone": "(503) 159-5469", "name": "Alice Johnson", "unit": "mi"}, {"state": "Illinois", "signup_date": "2023-11-14", "status": "Lost", "currency": "AUD", "is_active": true, "phone": "(588) 220-8602", "name": "Grace Ali", "unit": "cm"}, {"state": "Illinois", "signup_date": "2023-09-25", "status": "Open", "currency": "FJD", "is_active": true, "phone": "(512) 734-2618", "name": "Frank Diaz", "unit": "mi"}, {"state": "Illinois", "signup_date": "2023-10-06", "status": "Open", "currency": "FJD", "is_active": false, "phone": "(550) 904-0116", "name": "Sara Novak", "unit": "mi"}, {"state": "Illinois", "signup_date": "2023-11-10", "status": "Lost", "currency": "AUD", "is_active": false, "phone": "(522) 059-4990", "name": "Priya Brandt", "unit": "cm"}, {"state": "Illinois", "signup_date": "2023-10-16", "status": "Open", "currency": "CHE", "is_active": true, "phone": "(543) 751-1504", "name": "Alice Khan", "unit": "km"}, {"state": "Colorado", "signup_date": "2023-10-12", "status": "Lost", "currency": "AUD", "is_active": false, "phone": "(531) 060-2109", "name": "Priya Adams", "unit": "mi"}, {"state": "Illinois", "signup_date": "2023-08-05", "status": "Open", "currency": "AUD", "is_active": true, "phone": "(558) 864-3847", "name": "Grace Moore", "unit": "mi"}, {"state": "Illinois", "signup_date": "2023-04-10", "status": "Open", "currency": "CHE", "is_active": false, "phone": "(501) 312-2987", "name": "Judy Cruz", "unit": "cm"}, {"state": "Delaware", "signup_date": "2023-11-23", "status": "Lost", "currency": "FJD", "is_active": true, "phone": "(537) 805-9718", "name": "Carol Lee", "unit": "cm"}, {"state": "Vermont", "signup_date": "2023-06-02", "status": "Lost", "currency": "BWP", "is_active": true, "phone": "(589) 309-2417", "name": "Judy Khan", "unit": "mi"}, {"state": "Delaware", "signup_date": "2023-02-14", "status": "Open", "currency": "FJD", "is_active": true, "phone": "(595) 607-9011", "name": "Frank Novak", "unit": "mi"}, {"state": "Delaware", "signup_date": "2023-03-24", "status": "Open", "currency": "BWP", "is_active": true, "phone": "(515) 355-4217", "name": "Sara Johnson", "unit": "km"}, {"state": "Colorado", "signup_date": "2023-09-14", "status": "Lost", "currency": "AUD", "is_active": true, "phone": "(570) 828-2467", "name": "Grace Wong", "unit": "km"}, {"state": "Illinois", "signup_date": "2023-03-10", "status": "Open", "currency": "AUD", "is_active": false, "phone": "(529) 058-3072", "name": "Frank Fischer", "unit": "mi"}, {"state": "Delaware", "signup_date": "2023-07-06", "status": "Open", "currency": "AUD", "is_active": true, "phone": "(572) 286-1868", "name": "Frank Diaz", "unit": "km"}, {"state": "Delaware", "signup_date": "2023-04-22", "status": "Lost", "currency": "FJD", "is_active": true, "phone": "(575) 994-9353", "name": "David Ali", "unit": "km"}, {"state": "Illinois", "signup_date": "2023-11-22", "status": "Open", "currency": "CHE", "is_active": true, "phone": "(578) 533-7870", "name": "Eve Johnson", "unit": "mi"}, {"state": "Delaware", "signup_date": "2023-03-24", "status": "Lost", "currency": "AUD", "is_active": false, "phone": "(525) 274-3461", "name": "Omar Moore", "unit": "km"}, {"state": "Illinois", "signup_date": "2023-01-12", "status": "Lost", "currency": "FJD", "is_active": false, "phone": "(555) 448-5508", "name": "Carol Brandt", "unit": "km"}, {"state": "Illinois", "signup_date": "2023-02-28", "status": "Lost", "currency": "AUD", "is_active": false, "phone": "(564) 178-2230", "name": "Sara Wong", "unit": "mi"}, {"state": "Vermont", "signup_date": "2023-01-23", "status": "Open", "currency": "CHE", "is_active": false, "phone": "(542) 940-4976", "name": "Carol Lee", "unit": "km"}], "dirty_cols": ["state", "signup_date", "status", "currency", "is_active", "phone", "name", "unit", "col_x"], "clean_cols": ["state", "signup_date", "status", "currency", "is_active", "phone", "name", "unit"], "plan": {"dataset_summary": "32 rows × 9 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"De": "Delaware", "vermont": "Vermont", "Co": "Colorado", "vt": "Vermont", "CO": "Colorado", "il": "Illinois", "ILLINOIS": "Illinois", "IL": "Illinois", "DE": "Delaware", "delaware": "Delaware", "VT": "Vermont"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Oepn": "Open", "New": "Open", "lost": "Lost", "open": "Open", "Closed Lost": "Lost", "Lot": "Lost", "new": "Open", "CLOSED LOST": "Lost", "NEW": "Open", "closed lost": "Lost", "Closed ost": "Lost"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fiji Dollar": "FJD", "Aud": "AUD", "WIR Euro": "CHE", "wi euro": "CHE", "che": "CHE", "aud": "AUD", "AUSTRALIAN DOLLAR": "AUD", "fiji dollar": "FJD", "Australian Dollr": "AUD", "australian dollar": "AUD", "Australian Dollar": "AUD", "Pula": "BWP", "WIR uEro": "CHE", "fjd": "FJD", "WIR EURO": "CHE"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"centimeter": "cm", "CENTIMETERS": "cm", "Mile": "mi", "KM": "km", "kilomeetr": "km", "centimeters": "cm", "Mi": "mi", "MI": "mi", "Cm": "cm", "Kilometer": "km", "mie": "mi", "Centmieter": "cm", "kilometers": "km", "kilometer": "km", "miles": "mi"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": "Ops", "unit": "inches", "job_title": "eng mgr", "city": "Ile-Ife", "phone": "5344225075", "state": "Nj", "notes2": ""}, {"department": "I.T.", "unit": "l", "job_title": "  Engineering Mgr", "city": "Kidderminster", "phone": "(571)309-3108", "state": "WA   ", "notes2": ""}, {"department": "   I.T.", "unit": "inch", "job_title": "Ceo", "city": "kidderminster", "phone": "5690703624", "state": " New Jersey", "notes2": ""}, {"department": " operations ", "unit": "INCHES", "job_title": "Eng Mgr", "city": "Kidderminster", "phone": "559-690-7311", "state": "NJ", "notes2": ""}, {"department": "it", "unit": "kilometers", "job_title": "C..EO.", "city": "ile-ife", "phone": "(595)918-8413", "state": "WA", "notes2": ""}, {"department": "it", "unit": "L", "job_title": "ENG MANAGER   ", "city": "Kidderminster", "phone": "(576)646-2752", "state": "  alabama", "notes2": ""}, {"department": "Ops", "unit": "kilometers", "job_title": "PM", "city": "Kidderminster", "phone": "(583)636-0003", "state": "NJ", "notes2": ""}, {"department": "Operations", "unit": "litre", "job_title": "PROD MANAGER", "city": "Kidderminster", "phone": "546.665.7200", "state": "Al", "notes2": ""}, {"department": "It", "unit": "litre", "job_title": "PM ", "city": "Ile-Ise", "phone": "558-931-1958", "state": "WASHINGTON", "notes2": ""}, {"department": "I.T.", "unit": "\"", "job_title": "C.E.O.", "city": "ILE-IFE", "phone": "5238968847", "state": "alabama", "notes2": ""}, {"department": "OPERATIONS", "unit": "inch", "job_title": "product manager", "city": "ileief", "phone": "527-572-9593", "state": "Al", "notes2": ""}, {"department": "operations", "unit": "LITER", "job_title": "Eng  Manager", "city": "KIDDERMINSTER", "phone": "(592)300-7750", "state": "NJ", "notes2": ""}, {"department": "It", "unit": "litre", "job_title": "PM ", "city": "Ile-Ise", "phone": "558-931-1958", "state": "WASHINGTON", "notes2": ""}, {"department": " operations ", "unit": "INCHES", "job_title": "Eng Mgr", "city": "Kidderminster", "phone": "559-690-7311", "state": "NJ", "notes2": ""}, {"department": "", "unit": "", "job_title": "", "city": "", "phone": "", "state": "", "notes2": ""}], "clean": [{"department": "Operations", "unit": "in", "job_title": "Engineering Manager", "city": "Ile-Ife", "phone": "(534) 422-5075", "state": "New Jersey"}, {"department": "Information Technology", "unit": "L", "job_title": "Engineering Manager", "city": "Kidderminster", "phone": "(571) 309-3108", "state": "Washington"}, {"department": "Information Technology", "unit": "in", "job_title": "Chief Executive Officer", "city": "Kidderminster", "phone": "(569) 070-3624", "state": "New Jersey"}, {"department": "Operations", "unit": "in", "job_title": "Engineering Manager", "city": "Kidderminster", "phone": "(559) 690-7311", "state": "New Jersey"}, {"department": "Information Technology", "unit": "km", "job_title": "Chief Executive Officer", "city": "Ile-Ife", "phone": "(595) 918-8413", "state": "Washington"}, {"department": "Information Technology", "unit": "L", "job_title": "Engineering Manager", "city": "Kidderminster", "phone": "(576) 646-2752", "state": "Alabama"}, {"department": "Operations", "unit": "km", "job_title": "Product Manager", "city": "Kidderminster", "phone": "(583) 636-0003", "state": "New Jersey"}, {"department": "Operations", "unit": "L", "job_title": "Product Manager", "city": "Kidderminster", "phone": "(546) 665-7200", "state": "Alabama"}, {"department": "Information Technology", "unit": "L", "job_title": "Product Manager", "city": "Ile-Ife", "phone": "(558) 931-1958", "state": "Washington"}, {"department": "Information Technology", "unit": "in", "job_title": "Chief Executive Officer", "city": "Ile-Ife", "phone": "(523) 896-8847", "state": "Alabama"}, {"department": "Operations", "unit": "in", "job_title": "Product Manager", "city": "Ile-Ife", "phone": "(527) 572-9593", "state": "Alabama"}, {"department": "Operations", "unit": "L", "job_title": "Engineering Manager", "city": "Kidderminster", "phone": "(592) 300-7750", "state": "New Jersey"}], "dirty_cols": ["department", "unit", "job_title", "city", "phone", "state", "notes2"], "clean_cols": ["department", "unit", "job_title", "city", "phone", "state"], "plan": {"dataset_summary": "15 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ops": "Operations", "I.T.": "Information Technology", "operations": "Operations", "it": "Information Technology", "It": "Information Technology", "OPERATIONS": "Operations"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"inches": "in", "l": "L", "inch": "in", "INCHES": "in", "kilometers": "km", "litre": "L", "\"": "in", "LITER": "L"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"eng mgr": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "Ceo": "Chief Executive Officer", "Eng Mgr": "Engineering Manager", "C..EO.": "Chief Executive Officer", "ENG MANAGER": "Engineering Manager", "PM": "Product Manager", "PROD MANAGER": "Product Manager", "C.E.O.": "Chief Executive Officer", "product manager": "Product Manager", "Eng Manager": "Engineering Manager"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"kidderminster": "Kidderminster", "ile-ife": "Ile-Ife", "Ile-Ise": "Ile-Ife", "ILE-IFE": "Ile-Ife", "ileief": "Ile-Ife", "KIDDERMINSTER": "Kidderminster"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Nj": "New Jersey", "WA": "Washington", "NJ": "New Jersey", "alabama": "Alabama", "Al": "Alabama", "WASHINGTON": "Washington"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "Yes", "rate": "47.2%", "department": "prrod", "currency": "GIP", "industry": "education", "col_x": ""}, {"is_active": "Yes", "rate": "82.2%", "department": "PM", "currency": "Gibraltar Pound", "industry": "Fintech", "col_x": ""}, {"is_active": "false", "rate": "84.3%", "department": "Biz Dev", "currency": "gibraltar pound", "industry": " Financial Services", "col_x": ""}, {"is_active": "Yes", "rate": "59.0%", "department": "PRODUCT", "currency": " EGP ", "industry": "Utilitils", "col_x": ""}, {"is_active": "1", "rate": "73.9%", "department": "Prod", "currency": "egyptian pound", "industry": "finance", "col_x": ""}, {"is_active": "Yes", "rate": "8.4%", "department": "prod", "currency": "EGP", "industry": "Energy", "col_x": ""}, {"is_active": "FALSE", "rate": "43.3%", "department": "Product", "currency": "egp", "industry": "edtech", "col_x": ""}, {"is_active": "FALSE", "rate": "60.8%", "department": "Sales", "currency": "gip", "industry": "finserv", "col_x": ""}, {"is_active": "Y", "rate": "21.9%", "department": "BIZ DEV ", "currency": "GIP", "industry": "   EDTECH", "col_x": ""}, {"is_active": "F", "rate": "93.4%", "department": "  SALES", "currency": "Gibraltar Pound  ", "industry": "UTILITIES", "col_x": ""}, {"is_active": "Yes", "rate": "35.8%", "department": "Biz Dev", "currency": "GIP", "industry": "finserv", "col_x": ""}, {"is_active": "Y", "rate": "55.6%", "department": "PM", "currency": "GIP", "industry": "Utilities", "col_x": ""}, {"is_active": "Y", "rate": "74.4%", "department": "Prod", "currency": "Gibraaltar Pound", "industry": " Technology", "col_x": ""}, {"is_active": "true", "rate": "1.0%", "department": "SALES", "currency": "  GIBRALTAR POUND", "industry": " Sotware", "col_x": ""}, {"is_active": "0", "rate": "58.3%", "department": "biz dev", "currency": "Egyptian Puond", "industry": " FinServ ", "col_x": ""}, {"is_active": "true", "rate": "93.9%", "department": "Biz Dev", "currency": "egyptian pound ", "industry": "   EdTech", "col_x": ""}, {"is_active": "Y", "rate": "21.9%", "department": "BIZ DEV ", "currency": "GIP", "industry": "   EDTECH", "col_x": ""}, {"is_active": "FALSE", "rate": "60.8%", "department": "Sales", "currency": "gip", "industry": "finserv", "col_x": ""}, {"is_active": "", "rate": "", "department": "", "currency": "", "industry": "", "col_x": ""}], "clean": [{"is_active": true, "rate": 0.47200000000000003, "department": "Product", "currency": "GIP", "industry": "Education"}, {"is_active": true, "rate": 0.8220000000000001, "department": "Product", "currency": "GIP", "industry": "Financial Services"}, {"is_active": false, "rate": 0.843, "department": "Sales", "currency": "GIP", "industry": "Financial Services"}, {"is_active": true, "rate": 0.59, "department": "Product", "currency": "EGP", "industry": "Energy"}, {"is_active": true, "rate": 0.7390000000000001, "department": "Product", "currency": "EGP", "industry": "Financial Services"}, {"is_active": true, "rate": 0.084, "department": "Product", "currency": "EGP", "industry": "Energy"}, {"is_active": false, "rate": 0.433, "department": "Product", "currency": "EGP", "industry": "Education"}, {"is_active": false, "rate": 0.608, "department": "Sales", "currency": "GIP", "industry": "Financial Services"}, {"is_active": true, "rate": 0.21899999999999997, "department": "Sales", "currency": "GIP", "industry": "Education"}, {"is_active": false, "rate": 0.934, "department": "Sales", "currency": "GIP", "industry": "Energy"}, {"is_active": true, "rate": 0.358, "department": "Sales", "currency": "GIP", "industry": "Financial Services"}, {"is_active": true, "rate": 0.556, "department": "Product", "currency": "GIP", "industry": "Energy"}, {"is_active": true, "rate": 0.7440000000000001, "department": "Product", "currency": "GIP", "industry": "Technology"}, {"is_active": true, "rate": 0.01, "department": "Sales", "currency": "GIP", "industry": "Technology"}, {"is_active": false, "rate": 0.583, "department": "Sales", "currency": "EGP", "industry": "Financial Services"}, {"is_active": true, "rate": 0.9390000000000001, "department": "Sales", "currency": "EGP", "industry": "Education"}], "dirty_cols": ["is_active", "rate", "department", "currency", "industry", "col_x"], "clean_cols": ["is_active", "rate", "department", "currency", "industry"], "plan": {"dataset_summary": "19 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"prrod": "Product", "PM": "Product", "Biz Dev": "Sales", "PRODUCT": "Product", "Prod": "Product", "prod": "Product", "BIZ DEV": "Sales", "SALES": "Sales", "biz dev": "Sales"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Gibraltar Pound": "GIP", "gibraltar pound": "GIP", "egyptian pound": "EGP", "egp": "EGP", "gip": "GIP", "Gibraaltar Pound": "GIP", "GIBRALTAR POUND": "GIP", "Egyptian Puond": "EGP"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"education": "Education", "Fintech": "Financial Services", "Utilitils": "Energy", "finance": "Financial Services", "edtech": "Education", "finserv": "Financial Services", "EDTECH": "Education", "UTILITIES": "Energy", "Utilities": "Energy", "Sotware": "Technology", "FinServ": "Financial Services", "EdTech": "Education"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "Alice Ali", "is_active": "F", "industry": "Tourism", "status": " paid ", "company": " Umbrella ", "department": "Ops", "extra": ""}, {"name": " David Khan", "is_active": "1", "industry": "healthcare", "status": "overdue", "company": " Vehement", "department": "ops", "extra": ""}, {"name": "Omar Moore", "is_active": "false", "industry": "Health Care", "status": "refund", "company": "Stark Industries", "department": "ops   ", "extra": ""}, {"name": "Sara Lee", "is_active": "Y", "industry": "TOURISM", "status": "unpaid", "company": "Vehement   ", "department": "PM", "extra": ""}, {"name": "Frank Adams", "is_active": "Y", "industry": "hospitality", "status": "Overdeu", "company": "Globex", "department": "PM", "extra": ""}, {"name": "  Eve Brandt", "is_active": "T", "industry": "TELECOMMUNICATIONS", "status": "pending", "company": "Hooli ", "department": "Operations ", "extra": ""}, {"name": "Sara Wong", "is_active": "TRUE", "industry": "HEALTH CARE", "status": " PAID ", "company": "Initech", "department": "pm", "extra": ""}, {"name": "Mona Wong", "is_active": "Y", "industry": "TELECOMMUNICATIONS", "status": "paid   ", "company": "Hooli", "department": "PM", "extra": ""}, {"name": "Frank Khan   ", "is_active": "true", "industry": "toruism", "status": " pending ", "company": "  Globex", "department": "OPERATIONS", "extra": ""}, {"name": "   Judy Novak", "is_active": "Y", "industry": "travel", "status": "late", "company": "Soylent Corp", "department": "Customer Support ", "extra": ""}, {"name": "Heidi Wong", "is_active": "FALSE", "industry": "tourism", "status": "Overdue", "company": " Hooli ", "department": "cs", "extra": ""}, {"name": "Eve  Moore", "is_active": "N", "industry": "telecom", "status": "Paid", "company": "  Soylent Corp", "department": "CS", "extra": ""}, {"name": "David Johnson", "is_active": "Y", "industry": "Medical   ", "status": "refund", "company": "Stark Industries ", "department": "Customer Support", "extra": ""}, {"name": "Bob Wong", "is_active": "1", "industry": "Teecom", "status": "refunded", "company": "Globex", "department": "Operations", "extra": ""}, {"name": "David Lee   ", "is_active": "N", "industry": "Telecommunications", "status": "unpaid", "company": " Vehement ", "department": "product", "extra": ""}, {"name": "Heidi  Park", "is_active": "F", "industry": "MEDICAL", "status": "REFUNDED", "company": "Cyberdyne", "department": "OPS ", "extra": ""}, {"name": "   Ivan Cruz", "is_active": "false", "industry": "Tourism", "status": "REFUNDED", "company": "Initech", "department": "Prod", "extra": ""}, {"name": "Eve Johnson  ", "is_active": "true", "industry": "Health Care", "status": "paid", "company": " Globex ", "department": "Customer Support", "extra": ""}, {"name": "Ivan Park ", "is_active": "F", "industry": "Comms", "status": "pending", "company": "Wonka Co", "department": "PM", "extra": ""}, {"name": "   Karl Diaz", "is_active": "F", "industry": "health care", "status": " PAID ", "company": "Globex", "department": "Suppotr  ", "extra": ""}, {"name": "Bob Khan", "is_active": "Yes", "industry": "telecommunications", "status": "Refunded", "company": "Cyberdyne", "department": "cus support", "extra": ""}, {"name": "Karl Ali", "is_active": "Yes", "industry": " Health Care", "status": "Paid", "company": " Vehement", "department": "Product", "extra": ""}, {"name": " Alice Diaz", "is_active": "true", "industry": "  comms", "status": "Overdue", "company": "  Wonka Co", "department": " prod ", "extra": ""}, {"name": "Judy Park", "is_active": "T", "industry": " TEELCO", "status": "unpaid  ", "company": "Stark Industries   ", "department": "  Ops", "extra": ""}, {"name": "Karl Petrov", "is_active": "N", "industry": "health care", "status": "overdue ", "company": "Vehement  ", "department": "Prod", "extra": ""}, {"name": "Frank Lee", "is_active": "N", "industry": "Comms", "status": "Refunded", "company": "Soylent Corp", "department": "Prod", "extra": ""}, {"name": "Lena Ali", "is_active": "true", "industry": "TRAVEL", "status": " PERDING ", "company": "Vehement", "department": "supvort", "extra": ""}, {"name": "Judy Park", "is_active": "FALSE", "industry": "Tourism", "status": "reefunded", "company": "Initech", "department": "ops", "extra": ""}, {"name": "   Sara Smith", "is_active": "FALSE", "industry": "Telco", "status": " PAID ", "company": " Umbrella ", "department": "Cust Support", "extra": ""}, {"name": "Omar Lee", "is_active": "F", "industry": "TRAVEL  ", "status": "PAID  ", "company": " Initech ", "department": " ops ", "extra": ""}, {"name": "Grace Park", "is_active": "T", "industry": "Health Care", "status": "refund", "company": "Acme Inc", "department": "Cust Support  ", "extra": ""}, {"name": "   David Khan", "is_active": "0", "industry": "health care", "status": "REFUNDDE", "company": "Initech  ", "department": "   Prod", "extra": ""}, {"name": "Carol Moore", "is_active": "0", "industry": " Medical", "status": "Overdue", "company": "Wonka Co  ", "department": "operations", "extra": ""}, {"name": "Ivan  Brandt", "is_active": "1", "industry": "Travel", "status": "OVERDUE", "company": "  Initech", "department": "prod", "extra": ""}, {"name": "Grace Park", "is_active": "Yes", "industry": "Telco", "status": "overdue", "company": "Acme Inc", "department": "Product", "extra": ""}, {"name": " Omar Adams", "is_active": "Yes", "industry": "tourism", "status": "Pending", "company": "  Wonka Co", "department": "pm", "extra": ""}, {"name": "   Priya Diaz", "is_active": "1", "industry": "Medical  ", "status": "REFUND", "company": " Vehement ", "department": "Cust Support", "extra": ""}, {"name": "Karl Smith ", "is_active": "N", "industry": "Hottels &  Travel", "status": "OVERDUE", "company": "Vehement", "department": "Product", "extra": ""}, {"name": "Eve  Reyes", "is_active": "Yes", "industry": "Health Care", "status": "Refunded", "company": "  Cyberdyne", "department": "operations", "extra": ""}, {"name": "David Reyes", "is_active": "Yes", "industry": "travel", "status": "Overdue", "company": "Soylent Corp", "department": "Cust Support", "extra": ""}, {"name": "Alice Petrov", "is_active": "FALSE", "industry": "Hospitality", "status": "refunded", "company": "Wonka Co", "department": "Operatins", "extra": ""}, {"name": "Priya Wong  ", "is_active": "0", "industry": "HEATH", "status": "refunded", "company": "Umbrella", "department": "Operations", "extra": ""}, {"name": "David Novak   ", "is_active": "false", "industry": "   Trbvel", "status": "Refunded", "company": "Initech ", "department": "Customer Support", "extra": ""}, {"name": "Eve Cruz", "is_active": "TRUE", "industry": "Comms", "status": " refunded", "company": "Soylent Corp ", "department": "  SUPPORT", "extra": ""}, {"name": "Bob Diaz", "is_active": "1", "industry": "Hospitality", "status": " peid ", "company": "   Umbrella", "department": " SUPPORT ", "extra": ""}, {"name": "Ivan Cruz", "is_active": "Y", "industry": "Tourism", "status": "pending", "company": "Cyberdyne", "department": "Product", "extra": ""}, {"name": "Grace Ali", "is_active": "TRUE", "industry": "Medical", "status": "unpaid", "company": "Hooli", "department": "prod", "extra": ""}, {"name": "Sara Wong", "is_active": "TRUE", "industry": "HEALTH CARE", "status": " PAID ", "company": "Initech", "department": "pm", "extra": ""}, {"name": "", "is_active": "", "industry": "", "status": "", "company": "", "department": "", "extra": ""}], "clean": [{"name": "Alice Ali", "is_active": false, "industry": "Hospitality", "status": "Paid", "company": "Umbrella", "department": "Operations"}, {"name": "David Khan", "is_active": true, "industry": "Healthcare", "status": "Overdue", "company": "Vehement", "department": "Operations"}, {"name": "Omar Moore", "is_active": false, "industry": "Healthcare", "status": "Refunded", "company": "Stark Industries", "department": "Operations"}, {"name": "Sara Lee", "is_active": true, "industry": "Hospitality", "status": "Pending", "company": "Vehement", "department": "Product"}, {"name": "Frank Adams", "is_active": true, "industry": "Hospitality", "status": "Overdue", "company": "Globex", "department": "Product"}, {"name": "Eve Brandt", "is_active": true, "industry": "Telecommunications", "status": "Pending", "company": "Hooli", "department": "Operations"}, {"name": "Sara Wong", "is_active": true, "industry": "Healthcare", "status": "Paid", "company": "Initech", "department": "Product"}, {"name": "Mona Wong", "is_active": true, "industry": "Telecommunications", "status": "Paid", "company": "Hooli", "department": "Product"}, {"name": "Frank Khan", "is_active": true, "industry": "Hospitality", "status": "Pending", "company": "Globex", "department": "Operations"}, {"name": "Judy Novak", "is_active": true, "industry": "Hospitality", "status": "Overdue", "company": "Soylent Corp", "department": "Customer Support"}, {"name": "Heidi Wong", "is_active": false, "industry": "Hospitality", "status": "Overdue", "company": "Hooli", "department": "Customer Support"}, {"name": "Eve Moore", "is_active": false, "industry": "Telecommunications", "status": "Paid", "company": "Soylent Corp", "department": "Customer Support"}, {"name": "David Johnson", "is_active": true, "industry": "Healthcare", "status": "Refunded", "company": "Stark Industries", "department": "Customer Support"}, {"name": "Bob Wong", "is_active": true, "industry": "Telecommunications", "status": "Refunded", "company": "Globex", "department": "Operations"}, {"name": "David Lee", "is_active": false, "industry": "Telecommunications", "status": "Pending", "company": "Vehement", "department": "Product"}, {"name": "Heidi Park", "is_active": false, "industry": "Healthcare", "status": "Refunded", "company": "Cyberdyne", "department": "Operations"}, {"name": "Ivan Cruz", "is_active": false, "industry": "Hospitality", "status": "Refunded", "company": "Initech", "department": "Product"}, {"name": "Eve Johnson", "is_active": true, "industry": "Healthcare", "status": "Paid", "company": "Globex", "department": "Customer Support"}, {"name": "Ivan Park", "is_active": false, "industry": "Telecommunications", "status": "Pending", "company": "Wonka Co", "department": "Product"}, {"name": "Karl Diaz", "is_active": false, "industry": "Healthcare", "status": "Paid", "company": "Globex", "department": "Customer Support"}, {"name": "Bob Khan", "is_active": true, "industry": "Telecommunications", "status": "Refunded", "company": "Cyberdyne", "department": "Customer Support"}, {"name": "Karl Ali", "is_active": true, "industry": "Healthcare", "status": "Paid", "company": "Vehement", "department": "Product"}, {"name": "Alice Diaz", "is_active": true, "industry": "Telecommunications", "status": "Overdue", "company": "Wonka Co", "department": "Product"}, {"name": "Judy Park", "is_active": true, "industry": "Telecommunications", "status": "Pending", "company": "Stark Industries", "department": "Operations"}, {"name": "Karl Petrov", "is_active": false, "industry": "Healthcare", "status": "Overdue", "company": "Vehement", "department": "Product"}, {"name": "Frank Lee", "is_active": false, "industry": "Telecommunications", "status": "Refunded", "company": "Soylent Corp", "department": "Product"}, {"name": "Lena Ali", "is_active": true, "industry": "Hospitality", "status": "Pending", "company": "Vehement", "department": "Customer Support"}, {"name": "Judy Park", "is_active": false, "industry": "Hospitality", "status": "Refunded", "company": "Initech", "department": "Operations"}, {"name": "Sara Smith", "is_active": false, "industry": "Telecommunications", "status": "Paid", "company": "Umbrella", "department": "Customer Support"}, {"name": "Omar Lee", "is_active": false, "industry": "Hospitality", "status": "Paid", "company": "Initech", "department": "Operations"}, {"name": "Grace Park", "is_active": true, "industry": "Healthcare", "status": "Refunded", "company": "Acme Inc", "department": "Customer Support"}, {"name": "David Khan", "is_active": false, "industry": "Healthcare", "status": "Refunded", "company": "Initech", "department": "Product"}, {"name": "Carol Moore", "is_active": false, "industry": "Healthcare", "status": "Overdue", "company": "Wonka Co", "department": "Operations"}, {"name": "Ivan Brandt", "is_active": true, "industry": "Hospitality", "status": "Overdue", "company": "Initech", "department": "Product"}, {"name": "Grace Park", "is_active": true, "industry": "Telecommunications", "status": "Overdue", "company": "Acme Inc", "department": "Product"}, {"name": "Omar Adams", "is_active": true, "industry": "Hospitality", "status": "Pending", "company": "Wonka Co", "department": "Product"}, {"name": "Priya Diaz", "is_active": true, "industry": "Healthcare", "status": "Refunded", "company": "Vehement", "department": "Customer Support"}, {"name": "Karl Smith", "is_active": false, "industry": "Hospitality", "status": "Overdue", "company": "Vehement", "department": "Product"}, {"name": "Eve Reyes", "is_active": true, "industry": "Healthcare", "status": "Refunded", "company": "Cyberdyne", "department": "Operations"}, {"name": "David Reyes", "is_active": true, "industry": "Hospitality", "status": "Overdue", "company": "Soylent Corp", "department": "Customer Support"}, {"name": "Alice Petrov", "is_active": false, "industry": "Hospitality", "status": "Refunded", "company": "Wonka Co", "department": "Operations"}, {"name": "Priya Wong", "is_active": false, "industry": "Healthcare", "status": "Refunded", "company": "Umbrella", "department": "Operations"}, {"name": "David Novak", "is_active": false, "industry": "Hospitality", "status": "Refunded", "company": "Initech", "department": "Customer Support"}, {"name": "Eve Cruz", "is_active": true, "industry": "Telecommunications", "status": "Refunded", "company": "Soylent Corp", "department": "Customer Support"}, {"name": "Bob Diaz", "is_active": true, "industry": "Hospitality", "status": "Paid", "company": "Umbrella", "department": "Customer Support"}, {"name": "Ivan Cruz", "is_active": true, "industry": "Hospitality", "status": "Pending", "company": "Cyberdyne", "department": "Product"}, {"name": "Grace Ali", "is_active": true, "industry": "Healthcare", "status": "Pending", "company": "Hooli", "department": "Product"}], "dirty_cols": ["name", "is_active", "industry", "status", "company", "department", "extra"], "clean_cols": ["name", "is_active", "industry", "status", "company", "department"], "plan": {"dataset_summary": "49 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Tourism": "Hospitality", "healthcare": "Healthcare", "Health Care": "Healthcare", "TOURISM": "Hospitality", "hospitality": "Hospitality", "TELECOMMUNICATIONS": "Telecommunications", "HEALTH CARE": "Healthcare", "toruism": "Hospitality", "travel": "Hospitality", "tourism": "Hospitality", "telecom": "Telecommunications", "Medical": "Healthcare", "Teecom": "Telecommunications", "MEDICAL": "Healthcare", "Comms": "Telecommunications", "health care": "Healthcare", "telecommunications": "Telecommunications", "comms": "Telecommunications", "TEELCO": "Telecommunications", "TRAVEL": "Hospitality", "Telco": "Telecommunications", "Travel": "Hospitality", "Hottels & Travel": "Hospitality", "HEATH": "Healthcare", "Trbvel": "Hospitality"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paid": "Paid", "overdue": "Overdue", "refund": "Refunded", "unpaid": "Pending", "Overdeu": "Overdue", "pending": "Pending", "PAID": "Paid", "late": "Overdue", "refunded": "Refunded", "REFUNDED": "Refunded", "PERDING": "Pending", "reefunded": "Refunded", "REFUNDDE": "Refunded", "OVERDUE": "Overdue", "REFUND": "Refunded", "peid": "Paid"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ops": "Operations", "ops": "Operations", "PM": "Product", "pm": "Product", "OPERATIONS": "Operations", "cs": "Customer Support", "CS": "Customer Support", "product": "Product", "OPS": "Operations", "Prod": "Product", "Suppotr": "Customer Support", "cus support": "Customer Support", "prod": "Product", "supvort": "Customer Support", "Cust Support": "Customer Support", "operations": "Operations", "Operatins": "Operations", "SUPPORT": "Customer Support"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "Afif", "phone": "(541)870-4382", "rate": "83.9%", "email": "  EOJLIO@CORP.IO", "currency": "  Colombian Peso", "unit": "Centimeter", "notes2": ""}, {"city": "  Aif", "phone": "(540)410-6731", "rate": "99.4%", "email": "  ejdddj@corp.io", "currency": "CAD", "unit": "kilometers", "notes2": ""}, {"city": " Dugda", "phone": "517.022.5029", "rate": "67.4%", "email": "BEKNI@MAIL.COM", "currency": "COP", "unit": "Cm", "notes2": ""}, {"city": "Tokyo", "phone": "596.393.4574", "rate": "95.7%", "email": "nlfggk@mail.com", "currency": "COP", "unit": "km", "notes2": ""}, {"city": "dugda", "phone": "(589)795-5199", "rate": "47.1%", "email": "POMMAA@MAIL.COM  ", "currency": "Cop", "unit": "kilometers", "notes2": ""}, {"city": "Dguda", "phone": "506-000-7324", "rate": "44.6%", "email": "mcllea@test.org", "currency": "GIP", "unit": "kilometers", "notes2": ""}, {"city": " affi", "phone": "534.133.3942", "rate": "42.7%", "email": "GEJHMOB@MAIL.COM", "currency": "cop   ", "unit": "  Cm", "notes2": ""}, {"city": "AFIF", "phone": "523.141.2953", "rate": "47.9%", "email": "bpdjbd@example.com", "currency": "  COP", "unit": "Kilometer", "notes2": ""}, {"city": "Dugda", "phone": "543-913-0200", "rate": "62.3%", "email": "fgick@corp.io", "currency": "CAD", "unit": "kilometers", "notes2": ""}, {"city": "Tokyo", "phone": "533.703.6475", "rate": "59.2%", "email": "dfajife@corp.io", "currency": "swis franc", "unit": "km", "notes2": ""}, {"city": "TŌKYŌ  ", "phone": "5299146643", "rate": "44.7%", "email": "  fjecb@mail.com", "currency": "GIP", "unit": "cm", "notes2": ""}, {"city": "afif", "phone": "572-916-1727", "rate": "7.5%", "email": "FJKAC@MAIL.COM", "currency": "  Gibraltar Pound", "unit": "Km", "notes2": ""}, {"city": "tokyo", "phone": "511-072-6207", "rate": "78.9%", "email": "OIFI@MAIL.COM", "currency": "COP", "unit": "cm", "notes2": ""}, {"city": "dugda", "phone": "(592)512-1816", "rate": "37.6%", "email": "ajamlhc@mail.com   ", "currency": "Gibraltar Pound", "unit": " km", "notes2": ""}, {"city": "   afif", "phone": "5733556586", "rate": "67.6%", "email": "   ojkmdf@test.org", "currency": "swiss franc", "unit": "kilometers", "notes2": ""}, {"city": "Afif ", "phone": "530-148-1599", "rate": "81.1%", "email": "gafan@example.com  ", "currency": "   GIP", "unit": "centimeter", "notes2": ""}], "clean": [{"city": "Afif", "phone": "(541) 870-4382", "rate": 0.8390000000000001, "email": "eojlio@corp.io", "currency": "COP", "unit": "cm"}, {"city": "Afif", "phone": "(540) 410-6731", "rate": 0.9940000000000001, "email": "ejdddj@corp.io", "currency": "CAD", "unit": "km"}, {"city": "Dugda", "phone": "(517) 022-5029", "rate": 0.674, "email": "bekni@mail.com", "currency": "COP", "unit": "cm"}, {"city": "Tokyo", "phone": "(596) 393-4574", "rate": 0.9570000000000001, "email": "nlfggk@mail.com", "currency": "COP", "unit": "km"}, {"city": "Dugda", "phone": "(589) 795-5199", "rate": 0.47100000000000003, "email": "pommaa@mail.com", "currency": "COP", "unit": "km"}, {"city": "Dugda", "phone": "(506) 000-7324", "rate": 0.446, "email": "mcllea@test.org", "currency": "GIP", "unit": "km"}, {"city": "Afif", "phone": "(534) 133-3942", "rate": 0.42700000000000005, "email": "gejhmob@mail.com", "currency": "COP", "unit": "cm"}, {"city": "Afif", "phone": "(523) 141-2953", "rate": 0.479, "email": "bpdjbd@example.com", "currency": "COP", "unit": "km"}, {"city": "Dugda", "phone": "(543) 913-0200", "rate": 0.623, "email": "fgick@corp.io", "currency": "CAD", "unit": "km"}, {"city": "Tokyo", "phone": "(533) 703-6475", "rate": 0.5920000000000001, "email": "dfajife@corp.io", "currency": "CHF", "unit": "km"}, {"city": "Tokyo", "phone": "(529) 914-6643", "rate": 0.447, "email": "fjecb@mail.com", "currency": "GIP", "unit": "cm"}, {"city": "Afif", "phone": "(572) 916-1727", "rate": 0.075, "email": "fjkac@mail.com", "currency": "GIP", "unit": "km"}, {"city": "Tokyo", "phone": "(511) 072-6207", "rate": 0.789, "email": "oifi@mail.com", "currency": "COP", "unit": "cm"}, {"city": "Dugda", "phone": "(592) 512-1816", "rate": 0.376, "email": "ajamlhc@mail.com", "currency": "GIP", "unit": "km"}, {"city": "Afif", "phone": "(573) 355-6586", "rate": 0.6759999999999999, "email": "ojkmdf@test.org", "currency": "CHF", "unit": "km"}, {"city": "Afif", "phone": "(530) 148-1599", "rate": 0.8109999999999999, "email": "gafan@example.com", "currency": "GIP", "unit": "cm"}], "dirty_cols": ["city", "phone", "rate", "email", "currency", "unit", "notes2"], "clean_cols": ["city", "phone", "rate", "email", "currency", "unit"], "plan": {"dataset_summary": "16 rows × 7 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Aif": "Afif", "dugda": "Dugda", "Dguda": "Dugda", "affi": "Afif", "AFIF": "Afif", "TŌKYŌ": "Tokyo", "afif": "Afif", "tokyo": "Tokyo"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Colombian Peso": "COP", "Cop": "COP", "cop": "COP", "swis franc": "CHF", "Gibraltar Pound": "GIP", "swiss franc": "CHF"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Centimeter": "cm", "kilometers": "km", "Cm": "cm", "Kilometer": "km", "Km": "km", "centimeter": "cm"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "Chf", "phone": "528.809.9822", "department": "Sales"}, {"currency": "ERN   ", "phone": "(554)547-2224", "department": "operations"}, {"currency": "bob", "phone": "521-242-8525", "department": "operations"}, {"currency": "   Bob", "phone": "(527)512-2179", "department": "Operations  "}, {"currency": "chf", "phone": "(530)691-9756", "department": "ops"}, {"currency": "ERN", "phone": "525-798-0156", "department": "ops"}, {"currency": "ERN", "phone": "5781019831", "department": " dev"}, {"currency": "bob", "phone": "(538)480-7484", "department": "Ops"}, {"currency": "Nakfa", "phone": "596.795.2731", "department": " Ops"}, {"currency": "nakfa", "phone": "5061649952", "department": "Sales"}, {"currency": "CHF", "phone": "505.605.8184", "department": "biz dev "}, {"currency": "   ERN", "phone": "5648764796", "department": "Finance"}, {"currency": " chf ", "phone": "5604766539", "department": "  Operations"}, {"currency": "bob", "phone": "(512)103-2323", "department": "Finance"}, {"currency": " kwanza", "phone": "(531)178-0147", "department": "SALES"}, {"currency": "bob", "phone": "(512)103-2323", "department": "Finance"}], "clean": [{"currency": "CHF", "phone": "(528) 809-9822", "department": "Sales"}, {"currency": "ERN", "phone": "(554) 547-2224", "department": "Operations"}, {"currency": "BOB", "phone": "(521) 242-8525", "department": "Operations"}, {"currency": "BOB", "phone": "(527) 512-2179", "department": "Operations"}, {"currency": "CHF", "phone": "(530) 691-9756", "department": "Operations"}, {"currency": "ERN", "phone": "(525) 798-0156", "department": "Operations"}, {"currency": "ERN", "phone": "(578) 101-9831", "department": "Engineering"}, {"currency": "BOB", "phone": "(538) 480-7484", "department": "Operations"}, {"currency": "ERN", "phone": "(596) 795-2731", "department": "Operations"}, {"currency": "ERN", "phone": "(506) 164-9952", "department": "Sales"}, {"currency": "CHF", "phone": "(505) 605-8184", "department": "Sales"}, {"currency": "ERN", "phone": "(564) 876-4796", "department": "Finance"}, {"currency": "CHF", "phone": "(560) 476-6539", "department": "Operations"}, {"currency": "BOB", "phone": "(512) 103-2323", "department": "Finance"}, {"currency": "AOA", "phone": "(531) 178-0147", "department": "Sales"}], "dirty_cols": ["currency", "phone", "department"], "clean_cols": ["currency", "phone", "department"], "plan": {"dataset_summary": "16 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Chf": "CHF", "bob": "BOB", "Bob": "BOB", "chf": "CHF", "Nakfa": "ERN", "nakfa": "ERN", "kwanza": "AOA"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"operations": "Operations", "ops": "Operations", "dev": "Engineering", "Ops": "Operations", "biz dev": "Sales", "SALES": "Sales"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "4.596,42", "unit": "liters", "industry": "PropTech", "currency": "Argentine Peso", "is_active": "false", "status": "HIGH", "job_title": "  CHIEF EXECUTIVE OFFICER", "state": "Ne", "col_x": ""}, {"amount": "2.139,78", "unit": "kilometers", "industry": "   Utilities", "currency": "   argentine peso", "is_active": "0", "status": " HIGH ", "job_title": "Rep", "state": "New  Mexioc", "col_x": ""}, {"amount": "5.018,69", "unit": "LITERS", "industry": "energy", "currency": "   Argentine Peso", "is_active": "FALSE", "status": "low", "job_title": "C.E.O.", "state": "NE", "col_x": ""}, {"amount": "5.457,72", "unit": "L", "industry": "Property", "currency": "cabo verde escudo", "is_active": "Yes", "status": "low", "job_title": "prod manager", "state": "DC", "col_x": ""}, {"amount": "765,37", "unit": "L", "industry": " Telecommunications", "currency": "Argentine Peso ", "is_active": "N", "status": "Hihg  ", "job_title": "sales rep", "state": "Illinois", "col_x": ""}, {"amount": "$2,118.40", "unit": " LBS ", "industry": "Telecomumnications", "currency": "Argentine Peso  ", "is_active": "FALSE", "status": "P2", "job_title": "product manager", "state": " District of Columbia", "col_x": ""}, {"amount": "2.013,52", "unit": "Km", "industry": "   telecommuniations", "currency": "ars", "is_active": "0", "status": "Med", "job_title": "Sales Rp", "state": "Colorrado", "col_x": ""}, {"amount": "$3,553.85", "unit": "pounds", "industry": "RealEstate", "currency": "AMD   ", "is_active": "0", "status": "HIGH", "job_title": "Product Manager", "state": " NE", "col_x": ""}, {"amount": "$4,002.65", "unit": "l", "industry": "Telecom", "currency": "Armenian Dram", "is_active": "false", "status": "low", "job_title": "Rep", "state": "Colorado", "col_x": ""}, {"amount": "7.582,60", "unit": "lb", "industry": "RealEstate", "currency": "   Argentine Peso", "is_active": "N", "status": "medium  ", "job_title": "  Rep", "state": "illinois", "col_x": ""}, {"amount": "8.869,83", "unit": "L", "industry": "TELCO", "currency": "CZK", "is_active": "0", "status": "medium", "job_title": "CEO   ", "state": "New Mrxico", "col_x": ""}, {"amount": "None", "unit": "lbs", "industry": "property", "currency": "ARMENIAN  DRAM", "is_active": "1", "status": "m", "job_title": "prod manager", "state": "New Mexico", "col_x": ""}, {"amount": "$2,675.61", "unit": "lb", "industry": "Oil & Gas", "currency": "crbo verde escudo", "is_active": "1", "status": "Low", "job_title": "Prod Mgr", "state": "il", "col_x": ""}, {"amount": "TBD", "unit": "Kilometer", "industry": "ENERGY", "currency": "Argentine Peso", "is_active": "0", "status": " P3 ", "job_title": "Chief Executive Officer", "state": "IL", "col_x": ""}, {"amount": "3.031,64", "unit": "   lirte", "industry": "oil & gas", "currency": "CZK", "is_active": "N", "status": "High", "job_title": " PM", "state": "IL", "col_x": ""}, {"amount": "6.616,55", "unit": "LB", "industry": "Energy", "currency": "cve", "is_active": "T", "status": "p3", "job_title": "rep", "state": "Il", "col_x": ""}, {"amount": "null", "unit": "Liter", "industry": "Telecom", "currency": "Argentine Peso  ", "is_active": "TRUE", "status": "med", "job_title": "C.E.Q.", "state": "CO", "col_x": ""}, {"amount": "-", "unit": "Lb", "industry": "telecommunications", "currency": "CVE", "is_active": "No", "status": "P2", "job_title": "Prod Manager", "state": "New Mexico", "col_x": ""}, {"amount": "$4,677.62", "unit": "km", "industry": "Telecommunications", "currency": "CVE", "is_active": "FALSE", "status": "L", "job_title": "CEO", "state": "Colorado", "col_x": ""}, {"amount": "$8,866.23", "unit": " litre ", "industry": "Comms", "currency": "Armenian Dram   ", "is_active": "F", "status": "medium", "job_title": "Rep", "state": " NE ", "col_x": ""}, {"amount": "$7,564.91", "unit": "lb", "industry": "ENERGY", "currency": "  cve", "is_active": "N", "status": "p3", "job_title": "Prod Manager", "state": " nm", "col_x": ""}, {"amount": "3.477,55", "unit": "lbs", "industry": "telco", "currency": "Czech  Koruna", "is_active": "1", "status": "LOW", "job_title": "rep", "state": "COLORADO", "col_x": ""}, {"amount": "$3,747.83", "unit": "Km", "industry": "property", "currency": "Argentine Peso  ", "is_active": "FALSE", "status": "low", "job_title": "  PROD MANAGER", "state": " Ne ", "col_x": ""}, {"amount": "na", "unit": "Lb", "industry": "energy   ", "currency": " CZK", "is_active": "false", "status": "   l", "job_title": "  Product Mafager", "state": " CO ", "col_x": ""}, {"amount": "8.395,61", "unit": " Kilometers ", "industry": "telecommunications", "currency": "amd ", "is_active": "0", "status": "  h", "job_title": "   Chief Executive Offcer", "state": "co", "col_x": ""}, {"amount": "null", "unit": "L", "industry": "Oil & Gas", "currency": "CZECH KXRUNA", "is_active": "Yes", "status": " HIGH", "job_title": "c.e.o.", "state": "Illinois", "col_x": ""}, {"amount": "$3,886.26", "unit": "km", "industry": "PROPERTY", "currency": "Cabo Verde Escudo", "is_active": "1", "status": "MEDIUM", "job_title": "salesrepresentative", "state": " CO", "col_x": ""}, {"amount": "$363.67", "unit": "liter", "industry": "Real Estate", "currency": "ARS", "is_active": "T", "status": "l ", "job_title": "  C.E.O.", "state": " CO", "col_x": ""}, {"amount": "2.465,88", "unit": "L  ", "industry": "ProTech", "currency": "Czech Koruna", "is_active": "0", "status": "Low", "job_title": "Chief Executive Officer", "state": "Il", "col_x": ""}, {"amount": "3.964,10", "unit": " litre ", "industry": "PROPTECH", "currency": "Armenian Dram ", "is_active": "No", "status": "high", "job_title": "   Chief Executive Officer", "state": "DC", "col_x": ""}], "clean": [{"amount": 4596.42, "unit": "L", "industry": "Real Estate", "currency": "ARS", "is_active": false, "status": "High", "job_title": "Chief Executive Officer", "state": "Nebraska"}, {"amount": 2139.78, "unit": "km", "industry": "Energy", "currency": "ARS", "is_active": false, "status": "High", "job_title": "Sales Representative", "state": "New Mexico"}, {"amount": 5018.69, "unit": "L", "industry": "Energy", "currency": "ARS", "is_active": false, "status": "Low", "job_title": "Chief Executive Officer", "state": "Nebraska"}, {"amount": 5457.72, "unit": "L", "industry": "Real Estate", "currency": "CVE", "is_active": true, "status": "Low", "job_title": "Product Manager", "state": "District of Columbia"}, {"amount": 765.37, "unit": "L", "industry": "Telecommunications", "currency": "ARS", "is_active": false, "status": "High", "job_title": "Sales Representative", "state": "Illinois"}, {"amount": 2118.4, "unit": "lb", "industry": "Telecommunications", "currency": "ARS", "is_active": false, "status": "Medium", "job_title": "Product Manager", "state": "District of Columbia"}, {"amount": 2013.52, "unit": "km", "industry": "Telecommunications", "currency": "ARS", "is_active": false, "status": "Medium", "job_title": "Sales Representative", "state": "Colorado"}, {"amount": 3553.85, "unit": "lb", "industry": "Real Estate", "currency": "AMD", "is_active": false, "status": "High", "job_title": "Product Manager", "state": "Nebraska"}, {"amount": 4002.65, "unit": "L", "industry": "Telecommunications", "currency": "AMD", "is_active": false, "status": "Low", "job_title": "Sales Representative", "state": "Colorado"}, {"amount": 7582.6, "unit": "lb", "industry": "Real Estate", "currency": "ARS", "is_active": false, "status": "Medium", "job_title": "Sales Representative", "state": "Illinois"}, {"amount": 8869.83, "unit": "L", "industry": "Telecommunications", "currency": "CZK", "is_active": false, "status": "Medium", "job_title": "Chief Executive Officer", "state": "New Mexico"}, {"amount": NaN, "unit": "lb", "industry": "Real Estate", "currency": "AMD", "is_active": true, "status": "Medium", "job_title": "Product Manager", "state": "New Mexico"}, {"amount": 2675.61, "unit": "lb", "industry": "Energy", "currency": "CVE", "is_active": true, "status": "Low", "job_title": "Product Manager", "state": "Illinois"}, {"amount": NaN, "unit": "km", "industry": "Energy", "currency": "ARS", "is_active": false, "status": "Low", "job_title": "Chief Executive Officer", "state": "Illinois"}, {"amount": 3031.64, "unit": "L", "industry": "Energy", "currency": "CZK", "is_active": false, "status": "High", "job_title": "Product Manager", "state": "Illinois"}, {"amount": 6616.55, "unit": "lb", "industry": "Energy", "currency": "CVE", "is_active": true, "status": "Low", "job_title": "Sales Representative", "state": "Illinois"}, {"amount": NaN, "unit": "L", "industry": "Telecommunications", "currency": "ARS", "is_active": true, "status": "Medium", "job_title": "Chief Executive Officer", "state": "Colorado"}, {"amount": NaN, "unit": "lb", "industry": "Telecommunications", "currency": "CVE", "is_active": false, "status": "Medium", "job_title": "Product Manager", "state": "New Mexico"}, {"amount": 4677.62, "unit": "km", "industry": "Telecommunications", "currency": "CVE", "is_active": false, "status": "Low", "job_title": "Chief Executive Officer", "state": "Colorado"}, {"amount": 8866.23, "unit": "L", "industry": "Telecommunications", "currency": "AMD", "is_active": false, "status": "Medium", "job_title": "Sales Representative", "state": "Nebraska"}, {"amount": 7564.91, "unit": "lb", "industry": "Energy", "currency": "CVE", "is_active": false, "status": "Low", "job_title": "Product Manager", "state": "New Mexico"}, {"amount": 3477.55, "unit": "lb", "industry": "Telecommunications", "currency": "CZK", "is_active": true, "status": "Low", "job_title": "Sales Representative", "state": "Colorado"}, {"amount": 3747.83, "unit": "km", "industry": "Real Estate", "currency": "ARS", "is_active": false, "status": "Low", "job_title": "Product Manager", "state": "Nebraska"}, {"amount": NaN, "unit": "lb", "industry": "Energy", "currency": "CZK", "is_active": false, "status": "Low", "job_title": "Product Manager", "state": "Colorado"}, {"amount": 8395.61, "unit": "km", "industry": "Telecommunications", "currency": "AMD", "is_active": false, "status": "High", "job_title": "Chief Executive Officer", "state": "Colorado"}, {"amount": NaN, "unit": "L", "industry": "Energy", "currency": "CZK", "is_active": true, "status": "High", "job_title": "Chief Executive Officer", "state": "Illinois"}, {"amount": 3886.26, "unit": "km", "industry": "Real Estate", "currency": "CVE", "is_active": true, "status": "Medium", "job_title": "Sales Representative", "state": "Colorado"}, {"amount": 363.67, "unit": "L", "industry": "Real Estate", "currency": "ARS", "is_active": true, "status": "Low", "job_title": "Chief Executive Officer", "state": "Colorado"}, {"amount": 2465.88, "unit": "L", "industry": "Real Estate", "currency": "CZK", "is_active": false, "status": "Low", "job_title": "Chief Executive Officer", "state": "Illinois"}, {"amount": 3964.1, "unit": "L", "industry": "Real Estate", "currency": "AMD", "is_active": false, "status": "High", "job_title": "Chief Executive Officer", "state": "District of Columbia"}], "dirty_cols": ["amount", "unit", "industry", "currency", "is_active", "status", "job_title", "state", "col_x"], "clean_cols": ["amount", "unit", "industry", "currency", "is_active", "status", "job_title", "state"], "plan": {"dataset_summary": "30 rows × 9 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liters": "L", "kilometers": "km", "LITERS": "L", "LBS": "lb", "Km": "km", "pounds": "lb", "l": "L", "lbs": "lb", "Kilometer": "km", "lirte": "L", "LB": "lb", "Liter": "L", "Lb": "lb", "litre": "L", "Kilometers": "km", "liter": "L"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PropTech": "Real Estate", "Utilities": "Energy", "energy": "Energy", "Property": "Real Estate", "Telecomumnications": "Telecommunications", "telecommuniations": "Telecommunications", "RealEstate": "Real Estate", "Telecom": "Telecommunications", "TELCO": "Telecommunications", "property": "Real Estate", "Oil & Gas": "Energy", "ENERGY": "Energy", "oil & gas": "Energy", "telecommunications": "Telecommunications", "Comms": "Telecommunications", "telco": "Telecommunications", "PROPERTY": "Real Estate", "ProTech": "Real Estate", "PROPTECH": "Real Estate"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Argentine Peso": "ARS", "argentine peso": "ARS", "cabo verde escudo": "CVE", "ars": "ARS", "Armenian Dram": "AMD", "ARMENIAN DRAM": "AMD", "crbo verde escudo": "CVE", "cve": "CVE", "Czech Koruna": "CZK", "amd": "AMD", "CZECH KXRUNA": "CZK", "Cabo Verde Escudo": "CVE"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HIGH": "High", "low": "Low", "Hihg": "High", "P2": "Medium", "Med": "Medium", "medium": "Medium", "m": "Medium", "P3": "Low", "p3": "Low", "med": "Medium", "L": "Low", "LOW": "Low", "l": "Low", "h": "High", "MEDIUM": "Medium", "high": "High"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "Rep": "Sales Representative", "C.E.O.": "Chief Executive Officer", "prod manager": "Product Manager", "sales rep": "Sales Representative", "product manager": "Product Manager", "Sales Rp": "Sales Representative", "CEO": "Chief Executive Officer", "Prod Mgr": "Product Manager", "PM": "Product Manager", "rep": "Sales Representative", "C.E.Q.": "Chief Executive Officer", "Prod Manager": "Product Manager", "PROD MANAGER": "Product Manager", "Product Mafager": "Product Manager", "Chief Executive Offcer": "Chief Executive Officer", "c.e.o.": "Chief Executive Officer", "salesrepresentative": "Sales Representative"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ne": "Nebraska", "New Mexioc": "New Mexico", "NE": "Nebraska", "DC": "District of Columbia", "Colorrado": "Colorado", "illinois": "Illinois", "New Mrxico": "New Mexico", "il": "Illinois", "IL": "Illinois", "Il": "Illinois", "CO": "Colorado", "nm": "New Mexico", "COLORADO": "Colorado", "co": "Colorado"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"company": "  Initech", "status": "  Low", "department": " Product", "unit": "KILOMETERS", "currency": "guinean franc", "signup_date": "45190"}, {"company": "Wonka Co", "status": "L", "department": " ACCOUNTING ", "unit": "mi", "currency": "Bov", "signup_date": "8/14/2023"}, {"company": " Acme Inc", "status": "P3  ", "department": "Operations", "unit": "mi", "currency": "BAHRAINI DINAR", "signup_date": "28 Apr 2023"}, {"company": " Stark Industries", "status": "Medium", "department": " OPS", "unit": "MI", "currency": "BHD", "signup_date": "2023-01-13"}, {"company": " Cyberdyne", "status": "medium", "department": " Finance ", "unit": "kilometers", "currency": "Bahraini Dinar", "signup_date": "2023-10-04"}, {"company": "  Hooli", "status": "med", "department": " Operations ", "unit": "mi", "currency": "BHD", "signup_date": "2023-04-17"}, {"company": "Umbrella ", "status": "p2", "department": "finance", "unit": "miles", "currency": "BZD", "signup_date": "8/18/2023"}, {"company": "Wonka  Co", "status": "l", "department": "  ACCOUNTIG", "unit": "Kilometer", "currency": "   guinean franc", "signup_date": "19 Sep 2023"}, {"company": "Wonka Co", "status": "P2", "department": "Ops", "unit": "mi", "currency": " BahrainiDinar", "signup_date": "12/15/2023"}, {"company": "Umbrella", "status": "   P3", "department": "finance", "unit": "mi", "currency": "Belize Dollar", "signup_date": "2023-03-16"}, {"company": "  Stark Industries", "status": "M", "department": "PM", "unit": "km", "currency": "Gnf", "signup_date": "6/11/2023"}, {"company": "Wonka  Co", "status": " Low ", "department": "Product", "unit": "mi", "currency": "bahraini dinar", "signup_date": "9/3/2023"}, {"company": "Initech ", "status": "m", "department": "ops", "unit": "KILOMETERS", "currency": " BZD ", "signup_date": "45221"}, {"company": "Initech", "status": "p2", "department": "PM", "unit": " mile", "currency": "Bov", "signup_date": "2023-06-22"}, {"company": "Globex", "status": "L", "department": " Fniance ", "unit": " km ", "currency": "Belize Dollar", "signup_date": "2023-04-14"}, {"company": "Umbrella", "status": "LOW   ", "department": "Prod   ", "unit": "mile ", "currency": "Guinean Frac", "signup_date": "9 Feb 2023"}, {"company": "Wonka  Co", "status": "low", "department": "pm", "unit": "KILOMETERS", "currency": "BOV", "signup_date": "45002"}, {"company": "Stark  Industries", "status": "low", "department": "Ops", "unit": "mi", "currency": "bleize  dollar", "signup_date": "2023-01-18"}, {"company": " Vehement", "status": "LOW", "department": "operations", "unit": "kilometer", "currency": "Bahhraini Dinar", "signup_date": "2023-02-17"}, {"company": "Soylent Corp", "status": "low", "department": "OPS", "unit": "mile", "currency": "Belize Dollar", "signup_date": "6/3/2023"}, {"company": "Globex", "status": " medium", "department": "   Ops", "unit": "   MILE", "currency": "BZD   ", "signup_date": "2023-10-10"}, {"company": "Umbrella  ", "status": "med", "department": " Finance", "unit": "km", "currency": "guinean franc", "signup_date": "45200"}, {"company": "Acme Inc   ", "status": "low", "department": "   Prod", "unit": "Mi", "currency": "bov", "signup_date": "9 Sep 2023"}, {"company": "Umbrella  ", "status": "Medium", "department": "Prood", "unit": "  km", "currency": "Belize Dollar", "signup_date": "6/8/2023"}, {"company": " Initech", "status": " Medium ", "department": "Finance", "unit": "mi  ", "currency": "GUINEAN  FRANC", "signup_date": "27 Oct 2023"}, {"company": "Cyberdyne", "status": "medium", "department": "Operations", "unit": "MI", "currency": " BZD", "signup_date": "5/19/2023"}, {"company": "Umbrella", "status": "low", "department": "Ops", "unit": "kilometer", "currency": "GNF", "signup_date": "17 Oct 2023"}, {"company": "Initech", "status": "M", "department": "Operations", "unit": "mile", "currency": "belize dollar  ", "signup_date": "12/5/2023"}, {"company": "Umbrella", "status": "Low", "department": "ops", "unit": "KILOMETER   ", "currency": "bov", "signup_date": "11/12/2023"}, {"company": "Stark Industries", "status": "M", "department": "Ops", "unit": "kilometer", "currency": "GUINEAN FRANC", "signup_date": "18 Mar 2023"}, {"company": "Stark Industries", "status": "M", "department": "Ops", "unit": "kilometer", "currency": "GUINEAN FRANC", "signup_date": "18 Mar 2023"}, {"company": "", "status": "", "department": "", "unit": "", "currency": "", "signup_date": ""}], "clean": [{"company": "Initech", "status": "Low", "department": "Product", "unit": "km", "currency": "GNF", "signup_date": "2023-09-21"}, {"company": "Wonka Co", "status": "Low", "department": "Finance", "unit": "mi", "currency": "BOV", "signup_date": "2023-08-14"}, {"company": "Acme Inc", "status": "Low", "department": "Operations", "unit": "mi", "currency": "BHD", "signup_date": "2023-04-28"}, {"company": "Stark Industries", "status": "Medium", "department": "Operations", "unit": "mi", "currency": "BHD", "signup_date": "2023-01-13"}, {"company": "Cyberdyne", "status": "Medium", "department": "Finance", "unit": "km", "currency": "BHD", "signup_date": "2023-10-04"}, {"company": "Hooli", "status": "Medium", "department": "Operations", "unit": "mi", "currency": "BHD", "signup_date": "2023-04-17"}, {"company": "Umbrella", "status": "Medium", "department": "Finance", "unit": "mi", "currency": "BZD", "signup_date": "2023-08-18"}, {"company": "Wonka Co", "status": "Low", "department": "Finance", "unit": "km", "currency": "GNF", "signup_date": "2023-09-19"}, {"company": "Wonka Co", "status": "Medium", "department": "Operations", "unit": "mi", "currency": "BHD", "signup_date": "2023-12-15"}, {"company": "Umbrella", "status": "Low", "department": "Finance", "unit": "mi", "currency": "BZD", "signup_date": "2023-03-16"}, {"company": "Stark Industries", "status": "Medium", "department": "Product", "unit": "km", "currency": "GNF", "signup_date": "2023-06-11"}, {"company": "Wonka Co", "status": "Low", "department": "Product", "unit": "mi", "currency": "BHD", "signup_date": "2023-09-03"}, {"company": "Initech", "status": "Medium", "department": "Operations", "unit": "km", "currency": "BZD", "signup_date": "2023-10-22"}, {"company": "Initech", "status": "Medium", "department": "Product", "unit": "mi", "currency": "BOV", "signup_date": "2023-06-22"}, {"company": "Globex", "status": "Low", "department": "Finance", "unit": "km", "currency": "BZD", "signup_date": "2023-04-14"}, {"company": "Umbrella", "status": "Low", "department": "Product", "unit": "mi", "currency": "GNF", "signup_date": "2023-02-09"}, {"company": "Wonka Co", "status": "Low", "department": "Product", "unit": "km", "currency": "BOV", "signup_date": "2023-03-17"}, {"company": "Stark Industries", "status": "Low", "department": "Operations", "unit": "mi", "currency": "BZD", "signup_date": "2023-01-18"}, {"company": "Vehement", "status": "Low", "department": "Operations", "unit": "km", "currency": "BHD", "signup_date": "2023-02-17"}, {"company": "Soylent Corp", "status": "Low", "department": "Operations", "unit": "mi", "currency": "BZD", "signup_date": "2023-06-03"}, {"company": "Globex", "status": "Medium", "department": "Operations", "unit": "mi", "currency": "BZD", "signup_date": "2023-10-10"}, {"company": "Umbrella", "status": "Medium", "department": "Finance", "unit": "km", "currency": "GNF", "signup_date": "2023-10-01"}, {"company": "Acme Inc", "status": "Low", "department": "Product", "unit": "mi", "currency": "BOV", "signup_date": "2023-09-09"}, {"company": "Umbrella", "status": "Medium", "department": "Product", "unit": "km", "currency": "BZD", "signup_date": "2023-06-08"}, {"company": "Initech", "status": "Medium", "department": "Finance", "unit": "mi", "currency": "GNF", "signup_date": "2023-10-27"}, {"company": "Cyberdyne", "status": "Medium", "department": "Operations", "unit": "mi", "currency": "BZD", "signup_date": "2023-05-19"}, {"company": "Umbrella", "status": "Low", "department": "Operations", "unit": "km", "currency": "GNF", "signup_date": "2023-10-17"}, {"company": "Initech", "status": "Medium", "department": "Operations", "unit": "mi", "currency": "BZD", "signup_date": "2023-12-05"}, {"company": "Umbrella", "status": "Low", "department": "Operations", "unit": "km", "currency": "BOV", "signup_date": "2023-11-12"}, {"company": "Stark Industries", "status": "Medium", "department": "Operations", "unit": "km", "currency": "GNF", "signup_date": "2023-03-18"}], "dirty_cols": ["company", "status", "department", "unit", "currency", "signup_date"], "clean_cols": ["company", "status", "department", "unit", "currency", "signup_date"], "plan": {"dataset_summary": "32 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"L": "Low", "P3": "Low", "medium": "Medium", "med": "Medium", "p2": "Medium", "l": "Low", "P2": "Medium", "M": "Medium", "m": "Medium", "LOW": "Low", "low": "Low"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ACCOUNTING": "Finance", "OPS": "Operations", "finance": "Finance", "ACCOUNTIG": "Finance", "Ops": "Operations", "PM": "Product", "ops": "Operations", "Fniance": "Finance", "Prod": "Product", "pm": "Product", "operations": "Operations", "Prood": "Product"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KILOMETERS": "km", "MI": "mi", "kilometers": "km", "miles": "mi", "Kilometer": "km", "mile": "mi", "kilometer": "km", "MILE": "mi", "Mi": "mi", "KILOMETER": "km"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"guinean franc": "GNF", "Bov": "BOV", "BAHRAINI DINAR": "BHD", "Bahraini Dinar": "BHD", "BahrainiDinar": "BHD", "Belize Dollar": "BZD", "Gnf": "GNF", "bahraini dinar": "BHD", "Guinean Frac": "GNF", "bleize dollar": "BZD", "Bahhraini Dinar": "BHD", "bov": "BOV", "GUINEAN FRANC": "GNF", "belize dollar": "BZD"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"city": "Lins", "currency": "CDF", "country": "Nepal", "job_title": "V.P.", "name": "Grace Petrov", "status": "won", "unnamed": ""}, {"city": " lins ", "currency": "FJD", "country": "AT", "job_title": "eng mangaer", "name": "Karl  Cruz", "status": "open", "unnamed": ""}, {"city": "  jakomini", "currency": "FJD", "country": "Chrittmas Island", "job_title": "Account Executive", "name": "Carol  Wong", "status": "open", "unnamed": ""}, {"city": "jakomini", "currency": "Fiji Dollar", "country": "CXR", "job_title": "Vice Pres", "name": "Eve  Park", "status": "closed won   ", "unnamed": ""}, {"city": "Jakomini", "currency": "Fiji Dollar", "country": "Republic of Austria", "job_title": "Acct Exec  ", "name": "Judy Wong", "status": "  Ongoing", "unnamed": ""}, {"city": "JAKOMIEI", "currency": "cdf ", "country": " pcn ", "job_title": "acct exec", "name": "Grace Novak", "status": "WON", "unnamed": ""}, {"city": "Jakomini", "currency": "CDF", "country": "Fo", "job_title": "VP", "name": "Karl Petrov", "status": "WON", "unnamed": ""}, {"city": "Lins", "currency": "FJD", "country": "PCN", "job_title": "Administrative Assistant", "name": "Grace Novak   ", "status": "closed won", "unnamed": ""}, {"city": "lins", "currency": "  cdf", "country": "CXR", "job_title": "   ENG MGR", "name": "David Fischer", "status": "closed wn", "unnamed": ""}, {"city": "Jakomini", "currency": "CDF", "country": "np   ", "job_title": "Admin Assistant", "name": "Judy Ali", "status": "in-progress", "unnamed": ""}, {"city": "jakomini", "currency": " cdf", "country": "NP", "job_title": "Admin Assistant  ", "name": "Eve  Park", "status": "open   ", "unnamed": ""}, {"city": "jakomini", "currency": "CDF", "country": "FAROE IDLANDS", "job_title": "AE", "name": "David Khan", "status": "in-progress", "unnamed": ""}, {"city": "lins", "currency": "  FijiD ollar", "country": "Pitcairn", "job_title": "   VICE PRESIDENT", "name": "Lena Reyes", "status": "in-progress", "unnamed": ""}, {"city": "  jakomini", "currency": "  Fiji Dollar", "country": "FRO", "job_title": "eng manager", "name": "Eve Ali", "status": "ongoing", "unnamed": ""}, {"city": "Lins", "currency": " FJD", "country": "AUT", "job_title": "Vice  President", "name": "  Alice Smith", "status": "WON  ", "unnamed": ""}, {"city": "Jakomini", "currency": "Fiji Dollar", "country": "at", "job_title": "eng  manager", "name": "  Grace Diaz", "status": "   IN PROGRESS", "unnamed": ""}, {"city": "Lins ", "currency": "fiji dollar", "country": "CXR", "job_title": "engineering mgr", "name": "   Omar Lee", "status": "closed-won", "unnamed": ""}, {"city": " Jakomini ", "currency": " fjd ", "country": "CXR", "job_title": "eng  mgr", "name": "Lena  Fischer", "status": "In  Progress", "unnamed": ""}, {"city": "jakomini", "currency": "Fjd", "country": "AUT", "job_title": "Account Executive", "name": "Sara  Ali", "status": "   new", "unnamed": ""}, {"city": "Jauomini   ", "currency": "Congolese Franc", "country": " NP ", "job_title": "Engineeing  Mgr", "name": "Frank Moore", "status": "WON", "unnamed": ""}, {"city": "Jaakomini", "currency": " FJD ", "country": "nepal", "job_title": "acct exec", "name": "Eve  Johnson", "status": "WON", "unnamed": ""}, {"city": "Lins", "currency": "congolese franc", "country": "pcn", "job_title": "engineering mgr", "name": "Alice Wong", "status": "in-prosress", "unnamed": ""}, {"city": "lisn", "currency": "Congolese Franc", "country": "cx", "job_title": "Eng  Manger", "name": "Priya Moore", "status": "OPEN", "unnamed": ""}, {"city": " Jakomini", "currency": "CDF", "country": "Aut", "job_title": "VP   ", "name": "David Smith ", "status": "won", "unnamed": ""}, {"city": "Lins", "currency": " FJD ", "country": " aut ", "job_title": "Acct Exec", "name": "Mona  Novak", "status": "Oepn", "unnamed": ""}, {"city": "", "currency": "", "country": "", "job_title": "", "name": "", "status": "", "unnamed": ""}], "clean": [{"city": "Lins", "currency": "CDF", "country": "Nepal", "job_title": "Vice President", "name": "Grace Petrov", "status": "Won"}, {"city": "Lins", "currency": "FJD", "country": "Austria", "job_title": "Engineering Manager", "name": "Karl Cruz", "status": "Open"}, {"city": "Jakomini", "currency": "FJD", "country": "Christmas Island", "job_title": "Account Executive", "name": "Carol Wong", "status": "Open"}, {"city": "Jakomini", "currency": "FJD", "country": "Christmas Island", "job_title": "Vice President", "name": "Eve Park", "status": "Won"}, {"city": "Jakomini", "currency": "FJD", "country": "Austria", "job_title": "Account Executive", "name": "Judy Wong", "status": "In Progress"}, {"city": "Jakomini", "currency": "CDF", "country": "Pitcairn", "job_title": "Account Executive", "name": "Grace Novak", "status": "Won"}, {"city": "Jakomini", "currency": "CDF", "country": "Faroe Islands", "job_title": "Vice President", "name": "Karl Petrov", "status": "Won"}, {"city": "Lins", "currency": "FJD", "country": "Pitcairn", "job_title": "Administrative Assistant", "name": "Grace Novak", "status": "Won"}, {"city": "Lins", "currency": "CDF", "country": "Christmas Island", "job_title": "Engineering Manager", "name": "David Fischer", "status": "Won"}, {"city": "Jakomini", "currency": "CDF", "country": "Nepal", "job_title": "Administrative Assistant", "name": "Judy Ali", "status": "In Progress"}, {"city": "Jakomini", "currency": "CDF", "country": "Nepal", "job_title": "Administrative Assistant", "name": "Eve Park", "status": "Open"}, {"city": "Jakomini", "currency": "CDF", "country": "Faroe Islands", "job_title": "Account Executive", "name": "David Khan", "status": "In Progress"}, {"city": "Lins", "currency": "FJD", "country": "Pitcairn", "job_title": "Vice President", "name": "Lena Reyes", "status": "In Progress"}, {"city": "Jakomini", "currency": "FJD", "country": "Faroe Islands", "job_title": "Engineering Manager", "name": "Eve Ali", "status": "In Progress"}, {"city": "Lins", "currency": "FJD", "country": "Austria", "job_title": "Vice President", "name": "Alice Smith", "status": "Won"}, {"city": "Jakomini", "currency": "FJD", "country": "Austria", "job_title": "Engineering Manager", "name": "Grace Diaz", "status": "In Progress"}, {"city": "Lins", "currency": "FJD", "country": "Christmas Island", "job_title": "Engineering Manager", "name": "Omar Lee", "status": "Won"}, {"city": "Jakomini", "currency": "FJD", "country": "Christmas Island", "job_title": "Engineering Manager", "name": "Lena Fischer", "status": "In Progress"}, {"city": "Jakomini", "currency": "FJD", "country": "Austria", "job_title": "Account Executive", "name": "Sara Ali", "status": "Open"}, {"city": "Jakomini", "currency": "CDF", "country": "Nepal", "job_title": "Engineering Manager", "name": "Frank Moore", "status": "Won"}, {"city": "Jakomini", "currency": "FJD", "country": "Nepal", "job_title": "Account Executive", "name": "Eve Johnson", "status": "Won"}, {"city": "Lins", "currency": "CDF", "country": "Pitcairn", "job_title": "Engineering Manager", "name": "Alice Wong", "status": "In Progress"}, {"city": "Lins", "currency": "CDF", "country": "Christmas Island", "job_title": "Engineering Manager", "name": "Priya Moore", "status": "Open"}, {"city": "Jakomini", "currency": "CDF", "country": "Austria", "job_title": "Vice President", "name": "David Smith", "status": "Won"}, {"city": "Lins", "currency": "FJD", "country": "Austria", "job_title": "Account Executive", "name": "Mona Novak", "status": "Open"}], "dirty_cols": ["city", "currency", "country", "job_title", "name", "status", "unnamed"], "clean_cols": ["city", "currency", "country", "job_title", "name", "status"], "plan": {"dataset_summary": "26 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"lins": "Lins", "jakomini": "Jakomini", "JAKOMIEI": "Jakomini", "Jauomini": "Jakomini", "Jaakomini": "Jakomini", "lisn": "Lins"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fiji Dollar": "FJD", "cdf": "CDF", "FijiD ollar": "FJD", "fiji dollar": "FJD", "fjd": "FJD", "Fjd": "FJD", "Congolese Franc": "CDF", "congolese franc": "CDF"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AT": "Austria", "Chrittmas Island": "Christmas Island", "CXR": "Christmas Island", "Republic of Austria": "Austria", "pcn": "Pitcairn", "Fo": "Faroe Islands", "PCN": "Pitcairn", "np": "Nepal", "NP": "Nepal", "FAROE IDLANDS": "Faroe Islands", "FRO": "Faroe Islands", "AUT": "Austria", "at": "Austria", "nepal": "Nepal", "cx": "Christmas Island", "Aut": "Austria", "aut": "Austria"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"V.P.": "Vice President", "eng mangaer": "Engineering Manager", "Vice Pres": "Vice President", "Acct Exec": "Account Executive", "acct exec": "Account Executive", "VP": "Vice President", "ENG MGR": "Engineering Manager", "Admin Assistant": "Administrative Assistant", "AE": "Account Executive", "VICE PRESIDENT": "Vice President", "eng manager": "Engineering Manager", "engineering mgr": "Engineering Manager", "eng mgr": "Engineering Manager", "Engineeing Mgr": "Engineering Manager", "Eng Manger": "Engineering Manager"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"won": "Won", "open": "Open", "closed won": "Won", "Ongoing": "In Progress", "WON": "Won", "closed wn": "Won", "in-progress": "In Progress", "ongoing": "In Progress", "IN PROGRESS": "In Progress", "closed-won": "Won", "new": "Open", "in-prosress": "In Progress", "OPEN": "Open", "Oepn": "Open"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": " Operations", "phone": "5805464599", "email": " FMOGG@EXAMPLE.COM", "currency": "cve", "is_active": "Yes", "job_title": "Dir", "col_x": ""}, {"department": "Ops", "phone": "562.496.7488", "email": " FBEAB@EXAMPLE.COM", "currency": "aoa", "is_active": "Yes", "job_title": "Dir", "col_x": ""}, {"department": "PM", "phone": "540.668.9388", "email": " DCEMK@MAIL.COM ", "currency": "Cabo Verde Escudo", "is_active": "true", "job_title": "sales rep  ", "col_x": ""}, {"department": "PRODUCT", "phone": "5998344466", "email": "   OLNGP@TEST.ORG", "currency": "cve", "is_active": "TRUE", "job_title": "ACCT EXEC", "col_x": ""}, {"department": "Ops", "phone": "528-971-5361", "email": " HOEHDC@MAIL.COM", "currency": "Kwaanza", "is_active": "T", "job_title": "   Director", "col_x": ""}, {"department": "Prod", "phone": "573-258-3039", "email": " BKLBCI@EXAMPLE.COM", "currency": "Cabo Verde Escudo ", "is_active": "0", "job_title": "Account Executive", "col_x": ""}, {"department": "Prod", "phone": "524.194.7352", "email": " ogli@mail.com", "currency": "Kwanza", "is_active": "FALSE", "job_title": "Acct Exec", "col_x": ""}, {"department": "ops", "phone": "571-998-8719", "email": "  iapjl@mail.com", "currency": "CVE", "is_active": "Y", "job_title": "account executive", "col_x": ""}, {"department": "PM", "phone": "(552)358-2532", "email": " bmgeam@mail.com ", "currency": "Kwanza", "is_active": "F", "job_title": " Rep ", "col_x": ""}, {"department": "ops", "phone": "5985455300", "email": " PMFK@EXAMPLE.COM ", "currency": "CVE", "is_active": "F", "job_title": "Rep", "col_x": ""}, {"department": "Operations", "phone": "572-510-8713", "email": "eaphjhn@example.com", "currency": "kwanza", "is_active": "N", "job_title": "account executive", "col_x": ""}, {"department": "Ops   ", "phone": "524-840-2510", "email": "lgmj@example.com", "currency": "AOA  ", "is_active": "N", "job_title": "  Ae", "col_x": ""}, {"department": "operations", "phone": "515-950-4761", "email": "AIFLN@CORP.IO ", "currency": "CVE", "is_active": "No", "job_title": "sales representative  ", "col_x": ""}, {"department": "Ops", "phone": "551-594-5788", "email": "aipm@corp.io   ", "currency": "cabo verde escudo", "is_active": "No", "job_title": "sales representative", "col_x": ""}, {"department": "Operations", "phone": "539-313-1684", "email": "FGNEL@CORP.IO", "currency": "Cabo Verde Escudo", "is_active": "No", "job_title": "ae", "col_x": ""}, {"department": "PM", "phone": "573.223.7548", "email": "hbnlm@mail.com", "currency": " AOA", "is_active": "FALSE", "job_title": "Dierctor", "col_x": ""}, {"department": "pm", "phone": "(564)555-5628", "email": "acdko@test.org   ", "currency": "CVE", "is_active": "1", "job_title": "director", "col_x": ""}, {"department": "ops", "phone": "532.515.3822", "email": "MBONFFL@CORP.IO", "currency": "Cabo Verde Escudo", "is_active": "TRUE", "job_title": "DIR", "col_x": ""}, {"department": "Prrod", "phone": "(503)389-8166", "email": " LAIO@EXAMPLE.COM ", "currency": "cve", "is_active": "N", "job_title": "ae", "col_x": ""}, {"department": "Operations   ", "phone": "(506)217-6374", "email": "aggpjai@example.com", "currency": "Cabo Verde Escudo ", "is_active": "0", "job_title": "Dir.", "col_x": ""}, {"department": "Product", "phone": "5732733853", "email": "JGPDLIO@MAIL.COM   ", "currency": "AOA", "is_active": "false", "job_title": "Dir.   ", "col_x": ""}, {"department": "Ops", "phone": "5988026394", "email": "olckji@example.com", "currency": "AOA", "is_active": "FALSE", "job_title": "Acct Exec", "col_x": ""}, {"department": "Ops", "phone": "(595)306-2192", "email": "ilphhi@mail.com", "currency": "CVE", "is_active": "N", "job_title": "rep", "col_x": ""}, {"department": "pm", "phone": "5545014610", "email": " PPBFM@EXAMPLE.COM ", "currency": "kwanza  ", "is_active": "N", "job_title": "DIR.", "col_x": ""}, {"department": "operations", "phone": "5807905274", "email": "fepobc@test.org", "currency": "  cabo verde escudo", "is_active": "No", "job_title": "   Director", "col_x": ""}, {"department": "PRODUCT", "phone": "(558)962-1857", "email": "khgnknk@corp.io", "currency": "Kwanza", "is_active": "Y", "job_title": "REP", "col_x": ""}, {"department": "Ops", "phone": "579-379-6272", "email": "obaeoo@mail.com", "currency": "Kwanza", "is_active": "F", "job_title": "DIR", "col_x": ""}, {"department": "product", "phone": "5723611755", "email": " LKJCO@MAIL.COM", "currency": "   cabo verde escudo", "is_active": "false", "job_title": "Rep", "col_x": ""}, {"department": "PM", "phone": "576.834.6684", "email": "mjola@example.com", "currency": "cabo verde  escudo", "is_active": "true", "job_title": "Dir   ", "col_x": ""}, {"department": "Prod", "phone": "589.767.3496", "email": " fckimk@test.org", "currency": "AOA", "is_active": "TRUE", "job_title": "Dir", "col_x": ""}, {"department": "", "phone": "", "email": "", "currency": "", "is_active": "", "job_title": "", "col_x": ""}], "clean": [{"department": "Operations", "phone": "(580) 546-4599", "email": "fmogg@example.com", "currency": "CVE", "is_active": true, "job_title": "Director"}, {"department": "Operations", "phone": "(562) 496-7488", "email": "fbeab@example.com", "currency": "AOA", "is_active": true, "job_title": "Director"}, {"department": "Product", "phone": "(540) 668-9388", "email": "dcemk@mail.com", "currency": "CVE", "is_active": true, "job_title": "Sales Representative"}, {"department": "Product", "phone": "(599) 834-4466", "email": "olngp@test.org", "currency": "CVE", "is_active": true, "job_title": "Account Executive"}, {"department": "Operations", "phone": "(528) 971-5361", "email": "hoehdc@mail.com", "currency": "AOA", "is_active": true, "job_title": "Director"}, {"department": "Product", "phone": "(573) 258-3039", "email": "bklbci@example.com", "currency": "CVE", "is_active": false, "job_title": "Account Executive"}, {"department": "Product", "phone": "(524) 194-7352", "email": "ogli@mail.com", "currency": "AOA", "is_active": false, "job_title": "Account Executive"}, {"department": "Operations", "phone": "(571) 998-8719", "email": "iapjl@mail.com", "currency": "CVE", "is_active": true, "job_title": "Account Executive"}, {"department": "Product", "phone": "(552) 358-2532", "email": "bmgeam@mail.com", "currency": "AOA", "is_active": false, "job_title": "Sales Representative"}, {"department": "Operations", "phone": "(598) 545-5300", "email": "pmfk@example.com", "currency": "CVE", "is_active": false, "job_title": "Sales Representative"}, {"department": "Operations", "phone": "(572) 510-8713", "email": "eaphjhn@example.com", "currency": "AOA", "is_active": false, "job_title": "Account Executive"}, {"department": "Operations", "phone": "(524) 840-2510", "email": "lgmj@example.com", "currency": "AOA", "is_active": false, "job_title": "Account Executive"}, {"department": "Operations", "phone": "(515) 950-4761", "email": "aifln@corp.io", "currency": "CVE", "is_active": false, "job_title": "Sales Representative"}, {"department": "Operations", "phone": "(551) 594-5788", "email": "aipm@corp.io", "currency": "CVE", "is_active": false, "job_title": "Sales Representative"}, {"department": "Operations", "phone": "(539) 313-1684", "email": "fgnel@corp.io", "currency": "CVE", "is_active": false, "job_title": "Account Executive"}, {"department": "Product", "phone": "(573) 223-7548", "email": "hbnlm@mail.com", "currency": "AOA", "is_active": false, "job_title": "Director"}, {"department": "Product", "phone": "(564) 555-5628", "email": "acdko@test.org", "currency": "CVE", "is_active": true, "job_title": "Director"}, {"department": "Operations", "phone": "(532) 515-3822", "email": "mbonffl@corp.io", "currency": "CVE", "is_active": true, "job_title": "Director"}, {"department": "Product", "phone": "(503) 389-8166", "email": "laio@example.com", "currency": "CVE", "is_active": false, "job_title": "Account Executive"}, {"department": "Operations", "phone": "(506) 217-6374", "email": "aggpjai@example.com", "currency": "CVE", "is_active": false, "job_title": "Director"}, {"department": "Product", "phone": "(573) 273-3853", "email": "jgpdlio@mail.com", "currency": "AOA", "is_active": false, "job_title": "Director"}, {"department": "Operations", "phone": "(598) 802-6394", "email": "olckji@example.com", "currency": "AOA", "is_active": false, "job_title": "Account Executive"}, {"department": "Operations", "phone": "(595) 306-2192", "email": "ilphhi@mail.com", "currency": "CVE", "is_active": false, "job_title": "Sales Representative"}, {"department": "Product", "phone": "(554) 501-4610", "email": "ppbfm@example.com", "currency": "AOA", "is_active": false, "job_title": "Director"}, {"department": "Operations", "phone": "(580) 790-5274", "email": "fepobc@test.org", "currency": "CVE", "is_active": false, "job_title": "Director"}, {"department": "Product", "phone": "(558) 962-1857", "email": "khgnknk@corp.io", "currency": "AOA", "is_active": true, "job_title": "Sales Representative"}, {"department": "Operations", "phone": "(579) 379-6272", "email": "obaeoo@mail.com", "currency": "AOA", "is_active": false, "job_title": "Director"}, {"department": "Product", "phone": "(572) 361-1755", "email": "lkjco@mail.com", "currency": "CVE", "is_active": false, "job_title": "Sales Representative"}, {"department": "Product", "phone": "(576) 834-6684", "email": "mjola@example.com", "currency": "CVE", "is_active": true, "job_title": "Director"}, {"department": "Product", "phone": "(589) 767-3496", "email": "fckimk@test.org", "currency": "AOA", "is_active": true, "job_title": "Director"}], "dirty_cols": ["department", "phone", "email", "currency", "is_active", "job_title", "col_x"], "clean_cols": ["department", "phone", "email", "currency", "is_active", "job_title"], "plan": {"dataset_summary": "31 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ops": "Operations", "PM": "Product", "PRODUCT": "Product", "Prod": "Product", "ops": "Operations", "operations": "Operations", "pm": "Product", "Prrod": "Product", "product": "Product"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cve": "CVE", "aoa": "AOA", "Cabo Verde Escudo": "CVE", "Kwaanza": "AOA", "Kwanza": "AOA", "kwanza": "AOA", "cabo verde escudo": "CVE"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dir": "Director", "sales rep": "Sales Representative", "ACCT EXEC": "Account Executive", "Acct Exec": "Account Executive", "account executive": "Account Executive", "Rep": "Sales Representative", "Ae": "Account Executive", "sales representative": "Sales Representative", "ae": "Account Executive", "Dierctor": "Director", "director": "Director", "DIR": "Director", "Dir.": "Director", "rep": "Sales Representative", "DIR.": "Director", "REP": "Sales Representative"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"name": "Alice Khan", "company": "Wonka Co ", "phone": "5554395058", "extra": ""}, {"name": "Bob Wong", "company": "Hooli", "phone": "5307540279", "extra": ""}, {"name": "Carol Ali", "company": " Hooli ", "phone": "(547)868-3398", "extra": ""}, {"name": "Lena Fischer", "company": "Cyberdyne   ", "phone": "552-779-9145", "extra": ""}, {"name": "Frank Novak", "company": "Hooli", "phone": "(586)184-1378", "extra": ""}, {"name": "Lena Adams", "company": " Umbrella", "phone": "5180222880", "extra": ""}, {"name": "Mona Park   ", "company": "Hooli", "phone": "517.095.9998", "extra": ""}, {"name": " Bob Smith", "company": "Globex", "phone": "(515)192-6825", "extra": ""}, {"name": "Eve Novak ", "company": "Soylent  Corp", "phone": "5152858150", "extra": ""}, {"name": "David  Reyes", "company": "Stark Industries", "phone": "508.627.7795", "extra": ""}, {"name": "Mona Petrov", "company": "Cyberdyne   ", "phone": "5567841010", "extra": ""}, {"name": "Grace Diaz", "company": " Hooli", "phone": "500-211-2177", "extra": ""}, {"name": "Carol Lee", "company": "Acme Inc", "phone": "546-897-5080", "extra": ""}, {"name": "Eve Cruz", "company": "Hooli ", "phone": "5692425355", "extra": ""}, {"name": "Lena Lee  ", "company": "  Wonka Co", "phone": "580.271.8913", "extra": ""}, {"name": "Heidi Johnson", "company": "Stark Industries", "phone": "501.916.5197", "extra": ""}, {"name": "Grace  Johnson", "company": "Wonka Co", "phone": "5825621987", "extra": ""}, {"name": "Heidi Lee  ", "company": "Umbrella", "phone": "553.597.3972", "extra": ""}, {"name": "Ivan  Brandt", "company": "   Soylent Corp", "phone": "5502498441", "extra": ""}, {"name": "David Khan", "company": "  Acme Inc", "phone": "580.860.6928", "extra": ""}, {"name": " Eve Khan", "company": "  Globex", "phone": "546-353-2035", "extra": ""}, {"name": "Lena Khan", "company": " Cyberdyne ", "phone": "5039271838", "extra": ""}, {"name": "Judy Smith", "company": "   Vehement", "phone": "5163241035", "extra": ""}, {"name": " Bob Smith", "company": "Globex", "phone": "(515)192-6825", "extra": ""}, {"name": "Frank Novak", "company": "Hooli", "phone": "(586)184-1378", "extra": ""}], "clean": [{"name": "Alice Khan", "company": "Wonka Co", "phone": "(555) 439-5058"}, {"name": "Bob Wong", "company": "Hooli", "phone": "(530) 754-0279"}, {"name": "Carol Ali", "company": "Hooli", "phone": "(547) 868-3398"}, {"name": "Lena Fischer", "company": "Cyberdyne", "phone": "(552) 779-9145"}, {"name": "Frank Novak", "company": "Hooli", "phone": "(586) 184-1378"}, {"name": "Lena Adams", "company": "Umbrella", "phone": "(518) 022-2880"}, {"name": "Mona Park", "company": "Hooli", "phone": "(517) 095-9998"}, {"name": "Bob Smith", "company": "Globex", "phone": "(515) 192-6825"}, {"name": "Eve Novak", "company": "Soylent Corp", "phone": "(515) 285-8150"}, {"name": "David Reyes", "company": "Stark Industries", "phone": "(508) 627-7795"}, {"name": "Mona Petrov", "company": "Cyberdyne", "phone": "(556) 784-1010"}, {"name": "Grace Diaz", "company": "Hooli", "phone": "(500) 211-2177"}, {"name": "Carol Lee", "company": "Acme Inc", "phone": "(546) 897-5080"}, {"name": "Eve Cruz", "company": "Hooli", "phone": "(569) 242-5355"}, {"name": "Lena Lee", "company": "Wonka Co", "phone": "(580) 271-8913"}, {"name": "Heidi Johnson", "company": "Stark Industries", "phone": "(501) 916-5197"}, {"name": "Grace Johnson", "company": "Wonka Co", "phone": "(582) 562-1987"}, {"name": "Heidi Lee", "company": "Umbrella", "phone": "(553) 597-3972"}, {"name": "Ivan Brandt", "company": "Soylent Corp", "phone": "(550) 249-8441"}, {"name": "David Khan", "company": "Acme Inc", "phone": "(580) 860-6928"}, {"name": "Eve Khan", "company": "Globex", "phone": "(546) 353-2035"}, {"name": "Lena Khan", "company": "Cyberdyne", "phone": "(503) 927-1838"}, {"name": "Judy Smith", "company": "Vehement", "phone": "(516) 324-1035"}], "dirty_cols": ["name", "company", "phone", "extra"], "clean_cols": ["name", "company", "phone"], "plan": {"dataset_summary": "25 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"department": "Prod", "company": "  Wonka Co", "industry": "Hotels  & Travel", "country": " AX ", "col_x": ""}, {"department": "   Hr", "company": "Vehement", "industry": "Manufacturing", "country": "AX", "col_x": ""}, {"department": "biz dev", "company": "  Vehement", "industry": "INDUSTRIAL", "country": "Nepal", "col_x": ""}, {"department": "Product", "company": "Wonka Co", "industry": "Tourism", "country": "bd   ", "col_x": ""}, {"department": "Biz Dev", "company": "Hooli", "industry": "Technology", "country": " qat ", "col_x": ""}, {"department": "prod", "company": "Initech", "industry": "Mfg", "country": "qat", "col_x": ""}, {"department": "Sales", "company": "Wonka Co", "industry": "TRAVEL", "country": "BGD   ", "col_x": ""}, {"department": "Prod", "company": "  Soylent Corp", "industry": "   Industrial", "country": "Åland Islands", "col_x": ""}, {"department": " Information Technology", "company": "Vehement", "industry": " tech ", "country": "QAT", "col_x": ""}, {"department": "People Ops", "company": "Initech ", "industry": "Industrial", "country": "People's Republic of Bangladesh", "col_x": ""}, {"department": "Information Technology", "company": "Initech", "industry": "   Travel", "country": "AX", "col_x": ""}, {"department": "bz dev", "company": "Initech", "industry": "Industrial", "country": "ax", "col_x": ""}, {"department": "HR", "company": "Soylent Corp  ", "industry": "Hotels & Travel", "country": "Nepal", "col_x": ""}, {"department": "i.t.", "company": "  Vehement", "industry": "hotels & travel", "country": "PEOPLE'S REPUBLIC OF BANGLADESH", "col_x": ""}, {"department": "PM", "company": " Cyberdyne", "industry": "   MFG", "country": "federal democratic republic of nepal", "col_x": ""}, {"department": "information technology", "company": "Acme Inc", "industry": "Mfg", "country": "QA", "col_x": ""}, {"department": "Product", "company": "Globex", "industry": "Industrial", "country": "Bangladesh ", "col_x": ""}, {"department": "HUMAN RESOURCES   ", "company": " Soylent Corp", "industry": "Manufacturing", "country": "ax", "col_x": ""}, {"department": "i.t.", "company": "  Vehement", "industry": "hotels & travel", "country": "PEOPLE'S REPUBLIC OF BANGLADESH", "col_x": ""}], "clean": [{"department": "Product", "company": "Wonka Co", "industry": "Hospitality", "country": "Åland Islands"}, {"department": "Human Resources", "company": "Vehement", "industry": "Manufacturing", "country": "Åland Islands"}, {"department": "Sales", "company": "Vehement", "industry": "Manufacturing", "country": "Nepal"}, {"department": "Product", "company": "Wonka Co", "industry": "Hospitality", "country": "Bangladesh"}, {"department": "Sales", "company": "Hooli", "industry": "Technology", "country": "Qatar"}, {"department": "Product", "company": "Initech", "industry": "Manufacturing", "country": "Qatar"}, {"department": "Sales", "company": "Wonka Co", "industry": "Hospitality", "country": "Bangladesh"}, {"department": "Product", "company": "Soylent Corp", "industry": "Manufacturing", "country": "Åland Islands"}, {"department": "Information Technology", "company": "Vehement", "industry": "Technology", "country": "Qatar"}, {"department": "Human Resources", "company": "Initech", "industry": "Manufacturing", "country": "Bangladesh"}, {"department": "Information Technology", "company": "Initech", "industry": "Hospitality", "country": "Åland Islands"}, {"department": "Sales", "company": "Initech", "industry": "Manufacturing", "country": "Åland Islands"}, {"department": "Human Resources", "company": "Soylent Corp", "industry": "Hospitality", "country": "Nepal"}, {"department": "Information Technology", "company": "Vehement", "industry": "Hospitality", "country": "Bangladesh"}, {"department": "Product", "company": "Cyberdyne", "industry": "Manufacturing", "country": "Nepal"}, {"department": "Information Technology", "company": "Acme Inc", "industry": "Manufacturing", "country": "Qatar"}, {"department": "Product", "company": "Globex", "industry": "Manufacturing", "country": "Bangladesh"}, {"department": "Human Resources", "company": "Soylent Corp", "industry": "Manufacturing", "country": "Åland Islands"}], "dirty_cols": ["department", "company", "industry", "country", "col_x"], "clean_cols": ["department", "company", "industry", "country"], "plan": {"dataset_summary": "19 rows × 5 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Prod": "Product", "Hr": "Human Resources", "biz dev": "Sales", "Biz Dev": "Sales", "prod": "Product", "People Ops": "Human Resources", "bz dev": "Sales", "HR": "Human Resources", "i.t.": "Information Technology", "PM": "Product", "information technology": "Information Technology", "HUMAN RESOURCES": "Human Resources"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Hotels & Travel": "Hospitality", "INDUSTRIAL": "Manufacturing", "Tourism": "Hospitality", "Mfg": "Manufacturing", "TRAVEL": "Hospitality", "Industrial": "Manufacturing", "tech": "Technology", "Travel": "Hospitality", "hotels & travel": "Hospitality", "MFG": "Manufacturing"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AX": "Åland Islands", "bd": "Bangladesh", "qat": "Qatar", "BGD": "Bangladesh", "QAT": "Qatar", "People's Republic of Bangladesh": "Bangladesh", "ax": "Åland Islands", "PEOPLE'S REPUBLIC OF BANGLADESH": "Bangladesh", "federal democratic republic of nepal": "Nepal", "QA": "Qatar"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "$2,081.80", "is_active": "N", "phone": "(555)756-2678"}, {"amount": "4.652,43", "is_active": "FALSE", "phone": "5328865502"}, {"amount": "#N/A", "is_active": "false", "phone": "571.317.0657"}, {"amount": "5.065,86", "is_active": "No", "phone": "527-834-6707"}, {"amount": "$1,520.68", "is_active": "Y", "phone": "5143920746"}, {"amount": "--", "is_active": "F", "phone": "5986744795"}, {"amount": "1.178,87", "is_active": "N", "phone": "5420168120"}, {"amount": "null", "is_active": "true", "phone": "5975640242"}, {"amount": "505,87", "is_active": "false", "phone": "575.761.3061"}, {"amount": "1.271,21", "is_active": "0", "phone": "5519765722"}, {"amount": "$4,055.74", "is_active": "Yes", "phone": "590-199-2374"}, {"amount": "$3,232.26", "is_active": "Yes", "phone": "562-221-7996"}, {"amount": "$6,276.67", "is_active": "No", "phone": "(512)976-9682"}], "clean": [{"amount": 2081.8, "is_active": false, "phone": "(555) 756-2678"}, {"amount": 4652.43, "is_active": false, "phone": "(532) 886-5502"}, {"amount": NaN, "is_active": false, "phone": "(571) 317-0657"}, {"amount": 5065.86, "is_active": false, "phone": "(527) 834-6707"}, {"amount": 1520.68, "is_active": true, "phone": "(514) 392-0746"}, {"amount": NaN, "is_active": false, "phone": "(598) 674-4795"}, {"amount": 1178.87, "is_active": false, "phone": "(542) 016-8120"}, {"amount": NaN, "is_active": true, "phone": "(597) 564-0242"}, {"amount": 505.87, "is_active": false, "phone": "(575) 761-3061"}, {"amount": 1271.21, "is_active": false, "phone": "(551) 976-5722"}, {"amount": 4055.74, "is_active": true, "phone": "(590) 199-2374"}, {"amount": 3232.26, "is_active": true, "phone": "(562) 221-7996"}, {"amount": 6276.67, "is_active": false, "phone": "(512) 976-9682"}], "dirty_cols": ["amount", "is_active", "phone"], "clean_cols": ["amount", "is_active", "phone"], "plan": {"dataset_summary": "13 rows × 3 columns. 3 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"is_active": "1", "currency": "ERN", "status": " refund", "email": " ieda@mail.com ", "city": "Khanty-Mansiysk", "rate": "46.7%", "industry": "tech", "state": "De", "extra": ""}, {"is_active": "true", "currency": "GHS", "status": "   Pending", "email": "   agoflco@example.com", "city": "kahnty-mansiysk  ", "rate": "62.3%", "industry": "Oil & Gas", "state": "   de", "extra": ""}, {"is_active": "T", "currency": "hkd", "status": "pending", "email": "  EMGMLMJ@TEST.ORG", "city": "KHANTY-MANSIYSK", "rate": "82.3%", "industry": "Technoogy", "state": "DE", "extra": ""}, {"is_active": "false", "currency": "EGYPTIAN POUND", "status": " overdue", "email": "CEPOH@CORP.IO", "city": "VILLASIS", "rate": "82.1%", "industry": "tech", "state": "Delaware", "extra": ""}, {"is_active": "N", "currency": "Ghana Cedi", "status": "overdue", "email": "BEPG@TEST.ORG", "city": "khantymansiysk", "rate": "15.9%", "industry": "telecommunications", "state": "Nevada  ", "extra": ""}, {"is_active": "Yes", "currency": "ERN", "status": "late", "email": " HEPEHA@TEST.ORG ", "city": " KHANTY-OANSIYSK", "rate": "95.8%", "industry": "ENERGY", "state": "de", "extra": ""}, {"is_active": "false", "currency": "   HKD", "status": "refund", "email": "ibknio@test.org", "city": "  Khatny-Mansiysk", "rate": "58.9%", "industry": "tech", "state": "dtlaware   ", "extra": ""}, {"is_active": "0", "currency": "Ern   ", "status": " Late ", "email": "GJKLK@CORP.IO", "city": "Villasis", "rate": "83.0%", "industry": "   ED", "state": "nv", "extra": ""}, {"is_active": "0", "currency": "Egyptian Pound", "status": "pending", "email": "HJHNBN@TEST.ORG  ", "city": " VILLASIS ", "rate": "82.5%", "industry": "Telecommunications   ", "state": "delaware", "extra": ""}, {"is_active": "true", "currency": "ghs", "status": "  unpaiid", "email": "HPILJO@TEST.ORG", "city": "KhantyMansiysk  ", "rate": "71.4%", "industry": "Ed", "state": "NV", "extra": ""}, {"is_active": "1", "currency": "GHANA CEDI", "status": "UNPAID", "email": "mjhfeim@corp.io  ", "city": "villasis", "rate": "61.4%", "industry": "tech", "state": "DE", "extra": ""}, {"is_active": "Y", "currency": " Nakfa ", "status": " Refunded", "email": "IIDCCP@MAIL.COM", "city": "Khanty-Mnsiysk", "rate": "46.8%", "industry": "comms", "state": " Nevada ", "extra": ""}, {"is_active": "1", "currency": "Hong Kong Dollar", "status": "refund", "email": " jgikidi@corp.io ", "city": "   Villasis", "rate": "97.7%", "industry": "utilities", "state": "DELAWARE   ", "extra": ""}, {"is_active": "0", "currency": " ERN", "status": " PENDING ", "email": "pnmo@test.org   ", "city": "KhantyMansiysk", "rate": "32.3%", "industry": "education", "state": "Delaware", "extra": ""}, {"is_active": "true", "currency": "ERN", "status": "refunded", "email": "ldefo@example.com ", "city": "VILLAMIS", "rate": "29.0%", "industry": " ENERGY ", "state": " NV ", "extra": ""}, {"is_active": "1", "currency": "hong kong dollar", "status": "overdue", "email": "gfnfcf@mail.com ", "city": "VILLASIS", "rate": "25.7%", "industry": "Telecommunications", "state": "DE", "extra": ""}, {"is_active": "TRUE", "currency": "   Nakfa", "status": "due", "email": "bggi@example.com", "city": "Villasis", "rate": "45.0%", "industry": "EDTECH", "state": "Nv", "extra": ""}, {"is_active": "T", "currency": " ERN ", "status": "due", "email": "dghjlf@corp.io", "city": "Khanty-Mansiysk", "rate": "12.9%", "industry": "Telecommunications", "state": "delaware", "extra": ""}, {"is_active": "true", "currency": "Nakfa", "status": "REFUNDED", "email": "   MNDLM@CORP.IO", "city": "Villasis", "rate": "76.6%", "industry": "Telecom", "state": " DELAWARE ", "extra": ""}, {"is_active": "false", "currency": "ghana cedi", "status": "refunded", "email": "hnco@corp.io", "city": "KHANTY-MANSIYSK", "rate": "86.2%", "industry": "ED", "state": "NEVADA", "extra": ""}, {"is_active": "N", "currency": " egp ", "status": "Refunded", "email": " CLHGMKC@TEST.ORG ", "city": "Khanty-Mansiysk", "rate": "16.6%", "industry": "Energy   ", "state": "NV", "extra": ""}, {"is_active": "T", "currency": "Ern", "status": "late", "email": " iabag@test.org ", "city": "khnty-mansiysk   ", "rate": "56.0%", "industry": "  telecommunications", "state": "Delaware", "extra": ""}, {"is_active": "T", "currency": "ERN", "status": "late", "email": "   akabpg@test.org", "city": "khanty-mansiysk", "rate": "91.5%", "industry": "tech", "state": "DE", "extra": ""}, {"is_active": "false", "currency": "ghana cedi", "status": "Overdue", "email": "bape@example.com", "city": "   Khanty-Mansiysk", "rate": "42.3%", "industry": " EdTech", "state": "DLEAWARE", "extra": ""}, {"is_active": "0", "currency": "GHS", "status": "Overdve", "email": "kofmjh@test.org", "city": "villasis", "rate": "84.1%", "industry": "   Energy", "state": "Nevada", "extra": ""}, {"is_active": "No", "currency": "ERN", "status": "UNPAID", "email": "kgpid@mail.com", "city": "Villasis", "rate": "31.1%", "industry": " energy", "state": "NV", "extra": ""}, {"is_active": "0", "currency": "Ghana Cedi", "status": "Pending", "email": "DKMKJNK@TEST.ORG", "city": "khanty-mansiysk", "rate": "50.9%", "industry": "EdTech", "state": "de", "extra": ""}, {"is_active": "F", "currency": "GHS", "status": "pending", "email": "NAJAHB@EXAMPLE.COM   ", "city": "KhantyMansiysk", "rate": "5.6%", "industry": "UTILITIEES", "state": "delaware  ", "extra": ""}, {"is_active": "TRUE", "currency": "HONG BONG DOLLAR   ", "status": "refund", "email": "KKJHE@MAIL.COM", "city": "KHANTY-MANSIYSK", "rate": "77.7%", "industry": "energy", "state": "nevada", "extra": ""}, {"is_active": "0", "currency": "GHANA CEDI", "status": "refunded", "email": "edeaj@corp.io ", "city": "villasis", "rate": "6.8%", "industry": "It", "state": "delaware", "extra": ""}, {"is_active": "T", "currency": "hong kong dollar", "status": "refunded", "email": " MOKJPL@CORP.IO ", "city": "Villasis   ", "rate": "43.0%", "industry": "tech", "state": "Deaware", "extra": ""}, {"is_active": "1", "currency": "nakfa", "status": "PENDING", "email": "AHHM@EXAMPLE.COM", "city": "VILLASIS", "rate": "23.8%", "industry": "Oil & Gas", "state": "Nv", "extra": ""}, {"is_active": "T", "currency": "Nakfa", "status": "pending", "email": "jmgdbh@test.org", "city": " villasis ", "rate": "99.0%", "industry": "ed", "state": "DELAWARE", "extra": ""}, {"is_active": "false", "currency": "ERN", "status": "overdue", "email": "mabahco@mail.com", "city": "Villasis", "rate": "44.7%", "industry": "tech", "state": " nevada ", "extra": ""}, {"is_active": "true", "currency": "egyptian pound", "status": "due", "email": "abppkb@corp.io", "city": "khantymansiysk", "rate": "99.6%", "industry": "OIL & GAS", "state": "Nveada", "extra": ""}, {"is_active": "N", "currency": " hkd", "status": "overdue", "email": "fcacnd@example.com", "city": "villasis", "rate": "92.9%", "industry": "tech", "state": "DE   ", "extra": ""}, {"is_active": "N", "currency": "  Hong Kong Dollar", "status": " overdue ", "email": "FJMG@CORP.IO ", "city": "VILLASIS  ", "rate": "6.3%", "industry": "Comms", "state": "de", "extra": ""}, {"is_active": "N", "currency": "egp", "status": "refund", "email": "   HPMGPLO@EXAMPLE.COM", "city": " Khanty-Mansiysk ", "rate": "89.5%", "industry": "telco", "state": "NV", "extra": ""}, {"is_active": "TRUE", "currency": "GHANA CEDI", "status": "unpaid", "email": "ondono@example.com", "city": "Villasis", "rate": "17.5%", "industry": "comms", "state": "DE", "extra": ""}, {"is_active": "N", "currency": "EGP", "status": "rfeunded", "email": "  jjmd@test.org", "city": "Vjllasis", "rate": "67.9%", "industry": "ed", "state": " Nv ", "extra": ""}, {"is_active": "TRUE", "currency": "Hkd", "status": "reefunded", "email": " HBNGFNO@TEST.ORG ", "city": "villasis", "rate": "82.3%", "industry": "TELECOMMUNICATIONS", "state": "Nevada", "extra": ""}, {"is_active": "Y", "currency": "Nakfa", "status": "REFUNDED", "email": " jfpcf@corp.io ", "city": "Khanty-Mansiysk", "rate": "47.5%", "industry": "TECHNOLOGY", "state": "DELAWARE", "extra": ""}, {"is_active": "true", "currency": "Nakfa", "status": "reffunded", "email": "FOLBNA@EXAMPLE.COM", "city": "villasis", "rate": "79.0%", "industry": "tech", "state": "Nevada", "extra": ""}, {"is_active": "Y", "currency": "Egyptian Pound", "status": "pending", "email": "NDMCHJP@MAIL.COM", "city": "Khanty-Mansiysk", "rate": "22.7%", "industry": "edtech", "state": "nv  ", "extra": ""}, {"is_active": "", "currency": "", "status": "", "email": "", "city": "", "rate": "", "industry": "", "state": "", "extra": ""}], "clean": [{"is_active": true, "currency": "ERN", "status": "Refunded", "email": "ieda@mail.com", "city": "Khanty-Mansiysk", "rate": 0.467, "industry": "Technology", "state": "Delaware"}, {"is_active": true, "currency": "GHS", "status": "Pending", "email": "agoflco@example.com", "city": "Khanty-Mansiysk", "rate": 0.623, "industry": "Energy", "state": "Delaware"}, {"is_active": true, "currency": "HKD", "status": "Pending", "email": "emgmlmj@test.org", "city": "Khanty-Mansiysk", "rate": 0.823, "industry": "Technology", "state": "Delaware"}, {"is_active": false, "currency": "EGP", "status": "Overdue", "email": "cepoh@corp.io", "city": "Villasis", "rate": 0.821, "industry": "Technology", "state": "Delaware"}, {"is_active": false, "currency": "GHS", "status": "Overdue", "email": "bepg@test.org", "city": "Khanty-Mansiysk", "rate": 0.159, "industry": "Telecommunications", "state": "Nevada"}, {"is_active": true, "currency": "ERN", "status": "Overdue", "email": "hepeha@test.org", "city": "Khanty-Mansiysk", "rate": 0.958, "industry": "Energy", "state": "Delaware"}, {"is_active": false, "currency": "HKD", "status": "Refunded", "email": "ibknio@test.org", "city": "Khanty-Mansiysk", "rate": 0.589, "industry": "Technology", "state": "Delaware"}, {"is_active": false, "currency": "ERN", "status": "Overdue", "email": "gjklk@corp.io", "city": "Villasis", "rate": 0.83, "industry": "Education", "state": "Nevada"}, {"is_active": false, "currency": "EGP", "status": "Pending", "email": "hjhnbn@test.org", "city": "Villasis", "rate": 0.825, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": true, "currency": "GHS", "status": "Pending", "email": "hpiljo@test.org", "city": "Khanty-Mansiysk", "rate": 0.7140000000000001, "industry": "Education", "state": "Nevada"}, {"is_active": true, "currency": "GHS", "status": "Pending", "email": "mjhfeim@corp.io", "city": "Villasis", "rate": 0.614, "industry": "Technology", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Refunded", "email": "iidccp@mail.com", "city": "Khanty-Mansiysk", "rate": 0.46799999999999997, "industry": "Telecommunications", "state": "Nevada"}, {"is_active": true, "currency": "HKD", "status": "Refunded", "email": "jgikidi@corp.io", "city": "Villasis", "rate": 0.977, "industry": "Energy", "state": "Delaware"}, {"is_active": false, "currency": "ERN", "status": "Pending", "email": "pnmo@test.org", "city": "Khanty-Mansiysk", "rate": 0.32299999999999995, "industry": "Education", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Refunded", "email": "ldefo@example.com", "city": "Villasis", "rate": 0.29, "industry": "Energy", "state": "Nevada"}, {"is_active": true, "currency": "HKD", "status": "Overdue", "email": "gfnfcf@mail.com", "city": "Villasis", "rate": 0.257, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Pending", "email": "bggi@example.com", "city": "Villasis", "rate": 0.45, "industry": "Education", "state": "Nevada"}, {"is_active": true, "currency": "ERN", "status": "Pending", "email": "dghjlf@corp.io", "city": "Khanty-Mansiysk", "rate": 0.129, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Refunded", "email": "mndlm@corp.io", "city": "Villasis", "rate": 0.7659999999999999, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": false, "currency": "GHS", "status": "Refunded", "email": "hnco@corp.io", "city": "Khanty-Mansiysk", "rate": 0.862, "industry": "Education", "state": "Nevada"}, {"is_active": false, "currency": "EGP", "status": "Refunded", "email": "clhgmkc@test.org", "city": "Khanty-Mansiysk", "rate": 0.166, "industry": "Energy", "state": "Nevada"}, {"is_active": true, "currency": "ERN", "status": "Overdue", "email": "iabag@test.org", "city": "Khanty-Mansiysk", "rate": 0.56, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Overdue", "email": "akabpg@test.org", "city": "Khanty-Mansiysk", "rate": 0.915, "industry": "Technology", "state": "Delaware"}, {"is_active": false, "currency": "GHS", "status": "Overdue", "email": "bape@example.com", "city": "Khanty-Mansiysk", "rate": 0.423, "industry": "Education", "state": "Delaware"}, {"is_active": false, "currency": "GHS", "status": "Overdue", "email": "kofmjh@test.org", "city": "Villasis", "rate": 0.841, "industry": "Energy", "state": "Nevada"}, {"is_active": false, "currency": "ERN", "status": "Pending", "email": "kgpid@mail.com", "city": "Villasis", "rate": 0.311, "industry": "Energy", "state": "Nevada"}, {"is_active": false, "currency": "GHS", "status": "Pending", "email": "dkmkjnk@test.org", "city": "Khanty-Mansiysk", "rate": 0.509, "industry": "Education", "state": "Delaware"}, {"is_active": false, "currency": "GHS", "status": "Pending", "email": "najahb@example.com", "city": "Khanty-Mansiysk", "rate": 0.055999999999999994, "industry": "Energy", "state": "Delaware"}, {"is_active": true, "currency": "HKD", "status": "Refunded", "email": "kkjhe@mail.com", "city": "Khanty-Mansiysk", "rate": 0.777, "industry": "Energy", "state": "Nevada"}, {"is_active": false, "currency": "GHS", "status": "Refunded", "email": "edeaj@corp.io", "city": "Villasis", "rate": 0.068, "industry": "Technology", "state": "Delaware"}, {"is_active": true, "currency": "HKD", "status": "Refunded", "email": "mokjpl@corp.io", "city": "Villasis", "rate": 0.43, "industry": "Technology", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Pending", "email": "ahhm@example.com", "city": "Villasis", "rate": 0.23800000000000002, "industry": "Energy", "state": "Nevada"}, {"is_active": true, "currency": "ERN", "status": "Pending", "email": "jmgdbh@test.org", "city": "Villasis", "rate": 0.99, "industry": "Education", "state": "Delaware"}, {"is_active": false, "currency": "ERN", "status": "Overdue", "email": "mabahco@mail.com", "city": "Villasis", "rate": 0.447, "industry": "Technology", "state": "Nevada"}, {"is_active": true, "currency": "EGP", "status": "Pending", "email": "abppkb@corp.io", "city": "Khanty-Mansiysk", "rate": 0.996, "industry": "Energy", "state": "Nevada"}, {"is_active": false, "currency": "HKD", "status": "Overdue", "email": "fcacnd@example.com", "city": "Villasis", "rate": 0.929, "industry": "Technology", "state": "Delaware"}, {"is_active": false, "currency": "HKD", "status": "Overdue", "email": "fjmg@corp.io", "city": "Villasis", "rate": 0.063, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": false, "currency": "EGP", "status": "Refunded", "email": "hpmgplo@example.com", "city": "Khanty-Mansiysk", "rate": 0.895, "industry": "Telecommunications", "state": "Nevada"}, {"is_active": true, "currency": "GHS", "status": "Pending", "email": "ondono@example.com", "city": "Villasis", "rate": 0.175, "industry": "Telecommunications", "state": "Delaware"}, {"is_active": false, "currency": "EGP", "status": "Refunded", "email": "jjmd@test.org", "city": "Villasis", "rate": 0.679, "industry": "Education", "state": "Nevada"}, {"is_active": true, "currency": "HKD", "status": "Refunded", "email": "hbngfno@test.org", "city": "Villasis", "rate": 0.823, "industry": "Telecommunications", "state": "Nevada"}, {"is_active": true, "currency": "ERN", "status": "Refunded", "email": "jfpcf@corp.io", "city": "Khanty-Mansiysk", "rate": 0.475, "industry": "Technology", "state": "Delaware"}, {"is_active": true, "currency": "ERN", "status": "Refunded", "email": "folbna@example.com", "city": "Villasis", "rate": 0.79, "industry": "Technology", "state": "Nevada"}, {"is_active": true, "currency": "EGP", "status": "Pending", "email": "ndmchjp@mail.com", "city": "Khanty-Mansiysk", "rate": 0.22699999999999998, "industry": "Education", "state": "Nevada"}], "dirty_cols": ["is_active", "currency", "status", "email", "city", "rate", "industry", "state", "extra"], "clean_cols": ["is_active", "currency", "status", "email", "city", "rate", "industry", "state"], "plan": {"dataset_summary": "45 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"hkd": "HKD", "EGYPTIAN POUND": "EGP", "Ghana Cedi": "GHS", "Ern": "ERN", "Egyptian Pound": "EGP", "ghs": "GHS", "GHANA CEDI": "GHS", "Nakfa": "ERN", "Hong Kong Dollar": "HKD", "hong kong dollar": "HKD", "ghana cedi": "GHS", "egp": "EGP", "HONG BONG DOLLAR": "HKD", "nakfa": "ERN", "egyptian pound": "EGP", "Hkd": "HKD"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"refund": "Refunded", "pending": "Pending", "overdue": "Overdue", "late": "Overdue", "Late": "Overdue", "unpaiid": "Pending", "UNPAID": "Pending", "PENDING": "Pending", "refunded": "Refunded", "due": "Pending", "REFUNDED": "Refunded", "Overdve": "Overdue", "unpaid": "Pending", "rfeunded": "Refunded", "reefunded": "Refunded", "reffunded": "Refunded"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kahnty-mansiysk": "Khanty-Mansiysk", "KHANTY-MANSIYSK": "Khanty-Mansiysk", "VILLASIS": "Villasis", "khantymansiysk": "Khanty-Mansiysk", "KHANTY-OANSIYSK": "Khanty-Mansiysk", "Khatny-Mansiysk": "Khanty-Mansiysk", "KhantyMansiysk": "Khanty-Mansiysk", "villasis": "Villasis", "Khanty-Mnsiysk": "Khanty-Mansiysk", "VILLAMIS": "Villasis", "khnty-mansiysk": "Khanty-Mansiysk", "khanty-mansiysk": "Khanty-Mansiysk", "Vjllasis": "Villasis"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"tech": "Technology", "Oil & Gas": "Energy", "Technoogy": "Technology", "telecommunications": "Telecommunications", "ENERGY": "Energy", "ED": "Education", "Ed": "Education", "comms": "Telecommunications", "utilities": "Energy", "education": "Education", "EDTECH": "Education", "Telecom": "Telecommunications", "EdTech": "Education", "energy": "Energy", "UTILITIEES": "Energy", "It": "Technology", "ed": "Education", "OIL & GAS": "Energy", "Comms": "Telecommunications", "telco": "Telecommunications", "TELECOMMUNICATIONS": "Telecommunications", "TECHNOLOGY": "Technology", "edtech": "Education"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"De": "Delaware", "de": "Delaware", "DE": "Delaware", "dtlaware": "Delaware", "nv": "Nevada", "delaware": "Delaware", "NV": "Nevada", "DELAWARE": "Delaware", "Nv": "Nevada", "NEVADA": "Nevada", "DLEAWARE": "Delaware", "nevada": "Nevada", "Deaware": "Delaware", "Nveada": "Nevada"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "5.565,48", "phone": "5965876906", "unit": "miles", "is_active": "true", "job_title": "SALES REP", "name": "Mona Moore"}, {"amount": "6.656,28", "phone": "5470055887", "unit": "miles", "is_active": "true", "job_title": "SENIOR ENGINEER", "name": "Eve Reyes  "}, {"amount": "$2,824.80", "phone": "503.728.8559", "unit": "kilograms", "is_active": "FALSE", "job_title": "Sr Engineer", "name": "Ivan Moore"}, {"amount": "2.890,86", "phone": "566-335-8648", "unit": "kilograms", "is_active": "1", "job_title": "sales reprsentative", "name": "Frank Moore"}, {"amount": "na", "phone": "(536)787-5132", "unit": "km", "is_active": "No", "job_title": "Sr Engineer", "name": "   Grace Park"}, {"amount": "na", "phone": "(584)869-4797", "unit": "KG", "is_active": "T", "job_title": "Rep", "name": "Mona Reyes"}, {"amount": "3.436,80", "phone": "579-784-1305", "unit": "kg", "is_active": "0", "job_title": "Rep  ", "name": " Ivan Wong"}, {"amount": "$6,560.17", "phone": "511-091-6323", "unit": "miles", "is_active": "Yes", "job_title": "senior engineer", "name": "Judy Petrov "}, {"amount": "1.107,34", "phone": "(548)759-5078", "unit": "KILOMETER", "is_active": "Y", "job_title": "Senior Eng", "name": "   Judy Park"}, {"amount": "4.893,51", "phone": "(555)264-4443", "unit": "   km", "is_active": "Y", "job_title": "salesrepresentative", "name": "Judy Ali"}, {"amount": "4.328,07", "phone": "(551)053-7834", "unit": "KM", "is_active": "F", "job_title": "Sr Engineer", "name": "Judy Lee  "}, {"amount": "$6,323.18", "phone": "586.047.6819", "unit": "MI", "is_active": "T", "job_title": "Sales Representative", "name": "Sara Smith  "}, {"amount": "?", "phone": "(574)001-9219", "unit": " km", "is_active": "Y", "job_title": "Sales Representative", "name": "Karl Lee"}, {"amount": "$1,045.97", "phone": "(583)454-0178", "unit": "mile", "is_active": "Yes", "job_title": "sr. engineer", "name": "Lena Wong"}, {"amount": "7.041,37", "phone": "5986794174", "unit": "mi", "is_active": "1", "job_title": "sr. engineer", "name": "Frank Wong"}, {"amount": "$6,957.65", "phone": "528.358.3771", "unit": "Km", "is_active": "false", "job_title": "SNR ENGINEER", "name": "  Sara Ali"}, {"amount": "$1,961.59", "phone": "529-283-3082", "unit": "mi", "is_active": "No", "job_title": "SENIOR ENG", "name": "Heidi Moore"}, {"amount": "$1,804.14", "phone": "531-756-8945", "unit": "mile", "is_active": "Yes", "job_title": "Rep", "name": "David Fischer"}, {"amount": "3.725,77", "phone": "526-945-6049", "unit": "  km", "is_active": "F", "job_title": "Sales Representative", "name": "Alice  Lee"}, {"amount": "$2,538.96", "phone": "511-228-9808", "unit": "kg", "is_active": "false", "job_title": "sales representative", "name": "  Karl Adams"}, {"amount": "2.527,99", "phone": "557.165.6510", "unit": "KM   ", "is_active": "N", "job_title": "Sales Rep", "name": "Ivan Diaz"}, {"amount": "$4,288.40", "phone": "(526)869-6751", "unit": "KG", "is_active": "FALSE", "job_title": "senior eng", "name": "Carol Diaz"}, {"amount": "$2,479.49", "phone": "5087084856", "unit": "MVLES", "is_active": "true", "job_title": "Senior Enggineer", "name": "Bob Smith"}, {"amount": "$5,371.40", "phone": "5036185301", "unit": " Mi ", "is_active": "No", "job_title": "REP ", "name": "Bob Moore"}, {"amount": "4.585,61", "phone": "(565)192-7720", "unit": "kilometer", "is_active": "0", "job_title": "Senior Engineer", "name": "David Fischer"}, {"amount": "7.893,31", "phone": "(525)533-5533", "unit": "   MILE", "is_active": "No", "job_title": "Slaes Representative", "name": "Grace Cruz"}, {"amount": "$1,045.97", "phone": "(583)454-0178", "unit": "mile", "is_active": "Yes", "job_title": "sr. engineer", "name": "Lena Wong"}], "clean": [{"amount": 5565.48, "phone": "(596) 587-6906", "unit": "mi", "is_active": true, "job_title": "Sales Representative", "name": "Mona Moore"}, {"amount": 6656.28, "phone": "(547) 005-5887", "unit": "mi", "is_active": true, "job_title": "Senior Engineer", "name": "Eve Reyes"}, {"amount": 2824.8, "phone": "(503) 728-8559", "unit": "kg", "is_active": false, "job_title": "Senior Engineer", "name": "Ivan Moore"}, {"amount": 2890.86, "phone": "(566) 335-8648", "unit": "kg", "is_active": true, "job_title": "Sales Representative", "name": "Frank Moore"}, {"amount": NaN, "phone": "(536) 787-5132", "unit": "km", "is_active": false, "job_title": "Senior Engineer", "name": "Grace Park"}, {"amount": NaN, "phone": "(584) 869-4797", "unit": "kg", "is_active": true, "job_title": "Sales Representative", "name": "Mona Reyes"}, {"amount": 3436.8, "phone": "(579) 784-1305", "unit": "kg", "is_active": false, "job_title": "Sales Representative", "name": "Ivan Wong"}, {"amount": 6560.17, "phone": "(511) 091-6323", "unit": "mi", "is_active": true, "job_title": "Senior Engineer", "name": "Judy Petrov"}, {"amount": 1107.34, "phone": "(548) 759-5078", "unit": "km", "is_active": true, "job_title": "Senior Engineer", "name": "Judy Park"}, {"amount": 4893.51, "phone": "(555) 264-4443", "unit": "km", "is_active": true, "job_title": "Sales Representative", "name": "Judy Ali"}, {"amount": 4328.07, "phone": "(551) 053-7834", "unit": "km", "is_active": false, "job_title": "Senior Engineer", "name": "Judy Lee"}, {"amount": 6323.18, "phone": "(586) 047-6819", "unit": "mi", "is_active": true, "job_title": "Sales Representative", "name": "Sara Smith"}, {"amount": NaN, "phone": "(574) 001-9219", "unit": "km", "is_active": true, "job_title": "Sales Representative", "name": "Karl Lee"}, {"amount": 1045.97, "phone": "(583) 454-0178", "unit": "mi", "is_active": true, "job_title": "Senior Engineer", "name": "Lena Wong"}, {"amount": 7041.37, "phone": "(598) 679-4174", "unit": "mi", "is_active": true, "job_title": "Senior Engineer", "name": "Frank Wong"}, {"amount": 6957.65, "phone": "(528) 358-3771", "unit": "km", "is_active": false, "job_title": "Senior Engineer", "name": "Sara Ali"}, {"amount": 1961.59, "phone": "(529) 283-3082", "unit": "mi", "is_active": false, "job_title": "Senior Engineer", "name": "Heidi Moore"}, {"amount": 1804.14, "phone": "(531) 756-8945", "unit": "mi", "is_active": true, "job_title": "Sales Representative", "name": "David Fischer"}, {"amount": 3725.77, "phone": "(526) 945-6049", "unit": "km", "is_active": false, "job_title": "Sales Representative", "name": "Alice Lee"}, {"amount": 2538.96, "phone": "(511) 228-9808", "unit": "kg", "is_active": false, "job_title": "Sales Representative", "name": "Karl Adams"}, {"amount": 2527.99, "phone": "(557) 165-6510", "unit": "km", "is_active": false, "job_title": "Sales Representative", "name": "Ivan Diaz"}, {"amount": 4288.4, "phone": "(526) 869-6751", "unit": "kg", "is_active": false, "job_title": "Senior Engineer", "name": "Carol Diaz"}, {"amount": 2479.49, "phone": "(508) 708-4856", "unit": "mi", "is_active": true, "job_title": "Senior Engineer", "name": "Bob Smith"}, {"amount": 5371.4, "phone": "(503) 618-5301", "unit": "mi", "is_active": false, "job_title": "Sales Representative", "name": "Bob Moore"}, {"amount": 4585.61, "phone": "(565) 192-7720", "unit": "km", "is_active": false, "job_title": "Senior Engineer", "name": "David Fischer"}, {"amount": 7893.31, "phone": "(525) 533-5533", "unit": "mi", "is_active": false, "job_title": "Sales Representative", "name": "Grace Cruz"}], "dirty_cols": ["amount", "phone", "unit", "is_active", "job_title", "name"], "clean_cols": ["amount", "phone", "unit", "is_active", "job_title", "name"], "plan": {"dataset_summary": "27 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miles": "mi", "kilograms": "kg", "KG": "kg", "KILOMETER": "km", "KM": "km", "MI": "mi", "mile": "mi", "Km": "km", "MVLES": "mi", "Mi": "mi", "kilometer": "km", "MILE": "mi"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SALES REP": "Sales Representative", "SENIOR ENGINEER": "Senior Engineer", "Sr Engineer": "Senior Engineer", "sales reprsentative": "Sales Representative", "Rep": "Sales Representative", "senior engineer": "Senior Engineer", "Senior Eng": "Senior Engineer", "salesrepresentative": "Sales Representative", "sr. engineer": "Senior Engineer", "SNR ENGINEER": "Senior Engineer", "SENIOR ENG": "Senior Engineer", "sales representative": "Sales Representative", "Sales Rep": "Sales Representative", "senior eng": "Senior Engineer", "Senior Enggineer": "Senior Engineer", "REP": "Sales Representative", "Slaes Representative": "Sales Representative"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"email": "MFIM@EXAMPLE.COM", "job_title": "Director", "signup_date": "7 Dec 2023"}, {"email": "HJPCO@MAIL.COM", "job_title": "dir", "signup_date": "2023-12-17"}, {"email": "IHPMC@CORP.IO", "job_title": "director", "signup_date": "45155"}, {"email": "KJONGEH@TEST.ORG", "job_title": "pboduct manager", "signup_date": "15 Dec 2023"}, {"email": "   pejjl@corp.io", "job_title": "PM", "signup_date": "2023-10-15"}, {"email": "dcckk@test.org", "job_title": "Product Manager", "signup_date": "26 Feb 2023"}, {"email": " pjlfk@test.org ", "job_title": "ProdManager", "signup_date": "22 Apr 2023"}, {"email": "  MAAN@TEST.ORG", "job_title": "Prod Manager", "signup_date": "4 Aug 2023"}, {"email": "OBJCKM@MAIL.COM", "job_title": "Pm", "signup_date": "8/1/2023"}, {"email": " OKBCGK@MAIL.COM ", "job_title": "Dir", "signup_date": "4/11/2023"}, {"email": " oaajjo@test.org ", "job_title": "Pm", "signup_date": "2023-08-06"}, {"email": "   lmpfbie@example.com", "job_title": "   Dir", "signup_date": "4 Feb 2023"}, {"email": "lakp@example.com", "job_title": "pm", "signup_date": "45051"}, {"email": " BGGAOL@MAIL.COM ", "job_title": "DIRR.", "signup_date": "27 Feb 2023"}, {"email": "hljmeij@corp.io", "job_title": "PM", "signup_date": "6 Mar 2023"}, {"email": "cneem@mail.com", "job_title": "pm   ", "signup_date": "2023-07-03"}], "clean": [{"email": "mfim@example.com", "job_title": "Director", "signup_date": "2023-12-07"}, {"email": "hjpco@mail.com", "job_title": "Director", "signup_date": "2023-12-17"}, {"email": "ihpmc@corp.io", "job_title": "Director", "signup_date": "2023-08-17"}, {"email": "kjongeh@test.org", "job_title": "Product Manager", "signup_date": "2023-12-15"}, {"email": "pejjl@corp.io", "job_title": "Product Manager", "signup_date": "2023-10-15"}, {"email": "dcckk@test.org", "job_title": "Product Manager", "signup_date": "2023-02-26"}, {"email": "pjlfk@test.org", "job_title": "Product Manager", "signup_date": "2023-04-22"}, {"email": "maan@test.org", "job_title": "Product Manager", "signup_date": "2023-08-04"}, {"email": "objckm@mail.com", "job_title": "Product Manager", "signup_date": "2023-08-01"}, {"email": "okbcgk@mail.com", "job_title": "Director", "signup_date": "2023-04-11"}, {"email": "oaajjo@test.org", "job_title": "Product Manager", "signup_date": "2023-08-06"}, {"email": "lmpfbie@example.com", "job_title": "Director", "signup_date": "2023-02-04"}, {"email": "lakp@example.com", "job_title": "Product Manager", "signup_date": "2023-05-05"}, {"email": "bggaol@mail.com", "job_title": "Director", "signup_date": "2023-02-27"}, {"email": "hljmeij@corp.io", "job_title": "Product Manager", "signup_date": "2023-03-06"}, {"email": "cneem@mail.com", "job_title": "Product Manager", "signup_date": "2023-07-03"}], "dirty_cols": ["email", "job_title", "signup_date"], "clean_cols": ["email", "job_title", "signup_date"], "plan": {"dataset_summary": "16 rows × 3 columns. 3 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"dir": "Director", "director": "Director", "pboduct manager": "Product Manager", "PM": "Product Manager", "ProdManager": "Product Manager", "Prod Manager": "Product Manager", "Pm": "Product Manager", "Dir": "Director", "pm": "Product Manager", "DIRR.": "Director"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"job_title": "Engineering Mgr", "unit": "   INCH", "name": "Grace  Diaz", "state": " Idaho", "signup_date": "2023-10-23", "status": "active   "}, {"job_title": "Engineering Mgr  ", "unit": "inch", "name": "  David Fischer", "state": "ct", "signup_date": "2023-06-16", "status": "active   "}, {"job_title": "engineering mgr  ", "unit": "LITERS", "name": "  Judy Moore", "state": "United States Minor Outlying Islands", "signup_date": "2023-02-06", "status": "cancelled"}, {"job_title": "Eng Mgr", "unit": "   l", "name": "Judy Lee   ", "state": " UM ", "signup_date": "6/4/2023", "status": "cancelled"}, {"job_title": "Chief Technology OOfficer", "unit": "itch", "name": "Bob Johnson", "state": "CALIFORNIA", "signup_date": "2023-12-15", "status": "  canceled"}, {"job_title": "CHEF TECHNOLOGY OFFICER  ", "unit": "l", "name": "   Heidi Moore", "state": "CA", "signup_date": "1 May 2023", "status": "Active  "}, {"job_title": "chief technology officer", "unit": "l", "name": "Bob  Reyes", "state": "ca", "signup_date": "2023-02-01", "status": "cancelled"}, {"job_title": "Eng Mgr", "unit": "liter", "name": "Heidi Reyes  ", "state": "UM", "signup_date": "8/9/2023", "status": "canceled"}, {"job_title": "Eng Maager", "unit": "Inch", "name": "Lena Khan", "state": "Connectwcut", "signup_date": "12/5/2023", "status": "Active"}, {"job_title": "CTO", "unit": "inches", "name": "Karl  Reyes", "state": " Californya ", "signup_date": "15 Dec 2023", "status": " canceled "}, {"job_title": "   CHIEF TECHNOLOGY OFFICER", "unit": "INCH", "name": " Bob Fischer", "state": "id", "signup_date": "45166", "status": "active"}, {"job_title": "Eng Mgr", "unit": "litre", "name": "Lena Petrov", "state": "UM", "signup_date": "2023-04-01", "status": "ACTIVE"}, {"job_title": "Enineering  Mgr", "unit": "liter", "name": "Omar Petrov", "state": "Connecticut", "signup_date": "3 Mar 2023", "status": "active"}, {"job_title": "   Eng Mgr", "unit": "Liters", "name": "Priya Petrov", "state": "CONNECTICUT", "signup_date": "9 Feb 2023", "status": "ACTIVE"}, {"job_title": "chief technology officer", "unit": "l", "name": "Bob  Reyes", "state": "ca", "signup_date": "2023-02-01", "status": "cancelled"}, {"job_title": "Eng Mgr", "unit": "litre", "name": "Lena Petrov", "state": "UM", "signup_date": "2023-04-01", "status": "ACTIVE"}], "clean": [{"job_title": "Engineering Manager", "unit": "in", "name": "Grace Diaz", "state": "Idaho", "signup_date": "2023-10-23", "status": "Active"}, {"job_title": "Engineering Manager", "unit": "in", "name": "David Fischer", "state": "Connecticut", "signup_date": "2023-06-16", "status": "Active"}, {"job_title": "Engineering Manager", "unit": "L", "name": "Judy Moore", "state": "United States Minor Outlying Islands", "signup_date": "2023-02-06", "status": "Churned"}, {"job_title": "Engineering Manager", "unit": "L", "name": "Judy Lee", "state": "United States Minor Outlying Islands", "signup_date": "2023-06-04", "status": "Churned"}, {"job_title": "Chief Technology Officer", "unit": "in", "name": "Bob Johnson", "state": "California", "signup_date": "2023-12-15", "status": "Churned"}, {"job_title": "Chief Technology Officer", "unit": "L", "name": "Heidi Moore", "state": "California", "signup_date": "2023-05-01", "status": "Active"}, {"job_title": "Chief Technology Officer", "unit": "L", "name": "Bob Reyes", "state": "California", "signup_date": "2023-02-01", "status": "Churned"}, {"job_title": "Engineering Manager", "unit": "L", "name": "Heidi Reyes", "state": "United States Minor Outlying Islands", "signup_date": "2023-08-09", "status": "Churned"}, {"job_title": "Engineering Manager", "unit": "in", "name": "Lena Khan", "state": "Connecticut", "signup_date": "2023-12-05", "status": "Active"}, {"job_title": "Chief Technology Officer", "unit": "in", "name": "Karl Reyes", "state": "California", "signup_date": "2023-12-15", "status": "Churned"}, {"job_title": "Chief Technology Officer", "unit": "in", "name": "Bob Fischer", "state": "Idaho", "signup_date": "2023-08-28", "status": "Active"}, {"job_title": "Engineering Manager", "unit": "L", "name": "Lena Petrov", "state": "United States Minor Outlying Islands", "signup_date": "2023-04-01", "status": "Active"}, {"job_title": "Engineering Manager", "unit": "L", "name": "Omar Petrov", "state": "Connecticut", "signup_date": "2023-03-03", "status": "Active"}, {"job_title": "Engineering Manager", "unit": "L", "name": "Priya Petrov", "state": "Connecticut", "signup_date": "2023-02-09", "status": "Active"}], "dirty_cols": ["job_title", "unit", "name", "state", "signup_date", "status"], "clean_cols": ["job_title", "unit", "name", "state", "signup_date", "status"], "plan": {"dataset_summary": "16 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Engineering Mgr": "Engineering Manager", "engineering mgr": "Engineering Manager", "Eng Mgr": "Engineering Manager", "Chief Technology OOfficer": "Chief Technology Officer", "CHEF TECHNOLOGY OFFICER": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "Eng Maager": "Engineering Manager", "CTO": "Chief Technology Officer", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "Enineering Mgr": "Engineering Manager"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"INCH": "in", "inch": "in", "LITERS": "L", "l": "L", "itch": "in", "liter": "L", "Inch": "in", "inches": "in", "litre": "L", "Liters": "L"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ct": "Connecticut", "UM": "United States Minor Outlying Islands", "CALIFORNIA": "California", "CA": "California", "ca": "California", "Connectwcut": "Connecticut", "Californya": "California", "id": "Idaho", "CONNECTICUT": "Connecticut"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"active": "Active", "cancelled": "Churned", "canceled": "Churned", "ACTIVE": "Active"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"city": "Trenque Lauquen", "country": "Bharain", "email": "lfmgpo@example.com   ", "phone": "590.953.1270", "unit": "   pounds", "extra": ""}, {"city": "Marbach am Neckar", "country": "ID", "email": "iigk@test.org", "phone": "(585)356-1359", "unit": "  INCH", "extra": ""}, {"city": "TRENQUELAUQUEN", "country": " IDN ", "email": "   GOMLPON@EXAMPLE.COM", "phone": "587.219.4056", "unit": "mi", "extra": ""}, {"city": "Trenque Lauquen", "country": "zmb", "email": "gadl@mail.com", "phone": "(563)241-8889", "unit": "pound", "extra": ""}, {"city": "MARBACH AM NECKAR", "country": "bahrain", "email": " ohmckhk@mail.com ", "phone": "(528)611-2466", "unit": " inch ", "extra": ""}, {"city": "Marbach  am Neckar", "country": "Bahryin", "email": "CKMEIAF@CORP.IO  ", "phone": "(515)337-4379", "unit": "lbs", "extra": ""}, {"city": "LIKINO-DULEVO", "country": "Republic of Zambia", "email": "HFKNBK@MAIL.COM", "phone": "516.255.0241", "unit": "pounds", "extra": ""}, {"city": "Likino-Dulevo", "country": "id", "email": "FEHO@MAIL.COM   ", "phone": "(508)685-2653", "unit": "   Mi", "extra": ""}, {"city": "Marbach am Neckar", "country": "Republic of Indonesia", "email": "lebpaib@test.org", "phone": "(512)284-8937", "unit": "in", "extra": ""}, {"city": "trenque lauquen", "country": "IDN", "email": " knjbeni@corp.io ", "phone": "(511)267-0496", "unit": "MILE", "extra": ""}, {"city": "Trenque Lauquen", "country": "BH", "email": "dgbpeei@test.org", "phone": "549.509.4516", "unit": "lb", "extra": ""}, {"city": "   MARBACH AM NECKAR", "country": "ZM", "email": "MFNFB@EXAMPLE.COM", "phone": "(513)937-4325", "unit": "incch", "extra": ""}, {"city": "marbach  am neckar", "country": "zm", "email": "AJPPN@EXAMPLE.COM", "phone": "557-266-5542", "unit": "Miles", "extra": ""}, {"city": "Likino-Dulevo", "country": "Kingdom of Bahrain", "email": "AJLL@EXAMPLE.COM", "phone": "(543)698-1593", "unit": "pounds", "extra": ""}, {"city": "Likino-Dulevo", "country": "Republic of Indonesia ", "email": "CLEJLA@MAIL.COM", "phone": "579-726-8164", "unit": " Lb ", "extra": ""}, {"city": "trenque lauquen", "country": "zm  ", "email": "KOGHEBI@MAIL.COM", "phone": "596.523.5609", "unit": "LB", "extra": ""}, {"city": "   Trenque Lauquen", "country": "republic of zambia", "email": "aoelfdf@test.org", "phone": "537-980-9409", "unit": "mi", "extra": ""}, {"city": "Likino-Dulevo", "country": "  Kingdom of Bahrain", "email": "NCMDOCK@MAIL.COM  ", "phone": "(535)465-6749", "unit": "   MLES", "extra": ""}, {"city": "trenque lauquen", "country": "Kingdom of Bahrain", "email": "MBBGK@CORP.IO", "phone": "5951646191", "unit": "miles", "extra": ""}, {"city": "trenque lauquen", "country": "bhr", "email": "FEKOND@MAIL.COM", "phone": "578-099-9060", "unit": "ponds", "extra": ""}, {"city": "Marbach am  Neckar", "country": "republic of indonesia", "email": "DNLHAK@EXAMPLE.COM", "phone": "576.974.0269", "unit": "mi", "extra": ""}, {"city": "likino-dulevo", "country": "zamba", "email": "   hnfgpb@test.org", "phone": "502-549-0384", "unit": "inch", "extra": ""}, {"city": "Marbach  Am Neckar", "country": "ZMB", "email": "ibkjope@corp.io", "phone": "525-779-9132", "unit": "inches", "extra": ""}, {"city": "marbach am neckar", "country": "repubblic of indonesia", "email": " dclhpe@mail.com ", "phone": "5582791363", "unit": "in", "extra": ""}, {"city": "Trenque Lauquen", "country": "   idn", "email": "opicd@test.org", "phone": "5155687442", "unit": "IN", "extra": ""}, {"city": "likifo-dulevo", "country": "BH  ", "email": "BCDLLII@TEST.ORG", "phone": "584.981.3732", "unit": "lb", "extra": ""}, {"city": "trenque  lauquen", "country": "bh", "email": " BLPGAB@EXAMPLE.COM", "phone": "(559)502-3822", "unit": " \" ", "extra": ""}, {"city": "Marbach am Neckar   ", "country": "ID", "email": " bcngfd@example.com ", "phone": "5283748218", "unit": "\"", "extra": ""}, {"city": "TRENQUE LAUQUEN", "country": "IDN", "email": "IJON@TEST.ORG", "phone": "561.387.2067", "unit": "in  ", "extra": ""}, {"city": "Trenque  Lauquen", "country": "bh", "email": "adfdl@mail.com", "phone": "546-132-0856", "unit": "miles", "extra": ""}, {"city": "LikinoDulevo", "country": "Kingdom of Bahrain", "email": "llidgaa@example.com", "phone": "5070310267", "unit": " Mi", "extra": ""}, {"city": "Trenque Lauquen", "country": "republic of indonesia", "email": "DJFDF@EXAMPLE.COM", "phone": "5329130605", "unit": "In", "extra": ""}, {"city": "likino-dulevo", "country": " ZM ", "email": "   KGEKHD@MAIL.COM", "phone": "5906918368", "unit": "Mile", "extra": ""}, {"city": "marbach am neckar", "country": "IDN", "email": "ngobegg@corp.io", "phone": "5795788228", "unit": "\"", "extra": ""}, {"city": "TRENQUE LAUQUEN", "country": "Zambia", "email": "LFNHHIO@TEST.ORG", "phone": "(540)458-5782", "unit": "pounds", "extra": ""}, {"city": "marbac am neckar", "country": "Zambia", "email": " OAGJG@EXAMPLE.COM", "phone": "558.411.5109", "unit": "\"", "extra": ""}, {"city": "Marbach am Neckar", "country": "IDN", "email": "ICAGOH@CORP.IO", "phone": "538.036.9985", "unit": "inch", "extra": ""}, {"city": "marbach am neckar", "country": "Republic Of Zambia  ", "email": "lfnieg@example.com", "phone": "538-615-7596", "unit": "mi", "extra": ""}, {"city": "Mrbach am Neckar", "country": "idn   ", "email": "   IBFPKL@CORP.IO", "phone": "565-294-9984", "unit": "   inches", "extra": ""}, {"city": "marbach am neckar", "country": "Kingdom Of Bahrain", "email": "oicb@test.org", "phone": "518.895.7012", "unit": "lbs", "extra": ""}, {"city": "LIKIINO-DULEVO", "country": "Kingdom of Bahrain", "email": " blng@test.org ", "phone": "527-716-5728", "unit": "mile", "extra": ""}, {"city": "Trenque Lauquen", "country": "ZMMBIA", "email": " OINP@CORP.IO ", "phone": "532.278.3797", "unit": "miles", "extra": ""}, {"city": "marbach am neckar", "country": "INDNESIA ", "email": "djocg@example.com", "phone": "5270385031", "unit": "inch", "extra": ""}, {"city": "Likino-Dulevo", "country": "BHR", "email": "napnhne@example.com  ", "phone": "5805068540", "unit": "lb", "extra": ""}, {"city": "trenque lauquen", "country": "Zaambia", "email": "njnccgj@mail.com", "phone": "570-322-2267", "unit": "lb", "extra": ""}, {"city": "Marbach  am Neckar", "country": "kingdom of bahrain", "email": "lcap@example.com", "phone": "554.611.8042", "unit": "lb", "extra": ""}, {"city": "marbach am neckar", "country": "BHR", "email": "ANJPFF@TEST.ORG", "phone": "5466490120", "unit": "Mile", "extra": ""}, {"city": "Likino-Dulevo", "country": "BHR", "email": "GJON@MAIL.COM", "phone": "(592)512-1577", "unit": "   Lb", "extra": ""}, {"city": "Marbach am Neckar", "country": "bhr", "email": "GCLFNP@CORP.IO", "phone": "571.818.5346", "unit": "Inch", "extra": ""}, {"city": "LikinoDulevo", "country": "BH", "email": " LCODNDH@MAIL.COM ", "phone": "5595319797", "unit": "Mi", "extra": ""}, {"city": "marbach am neckar", "country": "  republic of indonesia", "email": "PENOH@MAIL.COM", "phone": "5791867285", "unit": "MILE", "extra": ""}, {"city": "Taenque Lauquen", "country": "Republic of Indonesia", "email": "ndddp@corp.io", "phone": "5629715955", "unit": "mile", "extra": ""}, {"city": " Marbach am Neckar", "country": "indonesia", "email": "fdnkbag@example.com", "phone": "513-825-0732", "unit": "   inch", "extra": ""}, {"city": "Likino-Dulevo", "country": "id", "email": "jjhhg@mail.com", "phone": "587.886.3866", "unit": "pounds", "extra": ""}, {"city": "LiinoDulevo", "country": "bahrain", "email": "kholdja@example.com", "phone": "5407079665", "unit": "pound", "extra": ""}, {"city": "marbach am neckar", "country": "ZAMBIA", "email": "amiipno@example.com", "phone": "(536)105-6793", "unit": "lbs", "extra": ""}, {"city": "MARBACH AM NECKAR", "country": "id", "email": "NBOMEE@MAIL.COM", "phone": "5525204068", "unit": " inches ", "extra": ""}, {"city": "LIKINODULEVO", "country": "zambia", "email": "dhmbpnf@corp.io", "phone": "523-612-6627", "unit": "\"", "extra": ""}, {"city": "Marbach am Neckar", "country": "ID", "email": "iigk@test.org", "phone": "(585)356-1359", "unit": "  INCH", "extra": ""}, {"city": "LikinoDulevo", "country": "Kingdom of Bahrain", "email": "llidgaa@example.com", "phone": "5070310267", "unit": " Mi", "extra": ""}, {"city": "", "country": "", "email": "", "phone": "", "unit": "", "extra": ""}], "clean": [{"city": "Trenque Lauquen", "country": "Bahrain", "email": "lfmgpo@example.com", "phone": "(590) 953-1270", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "iigk@test.org", "phone": "(585) 356-1359", "unit": "in"}, {"city": "Trenque Lauquen", "country": "Indonesia", "email": "gomlpon@example.com", "phone": "(587) 219-4056", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Zambia", "email": "gadl@mail.com", "phone": "(563) 241-8889", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Bahrain", "email": "ohmckhk@mail.com", "phone": "(528) 611-2466", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Bahrain", "email": "ckmeiaf@corp.io", "phone": "(515) 337-4379", "unit": "lb"}, {"city": "Likino-Dulevo", "country": "Zambia", "email": "hfknbk@mail.com", "phone": "(516) 255-0241", "unit": "lb"}, {"city": "Likino-Dulevo", "country": "Indonesia", "email": "feho@mail.com", "phone": "(508) 685-2653", "unit": "mi"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "lebpaib@test.org", "phone": "(512) 284-8937", "unit": "in"}, {"city": "Trenque Lauquen", "country": "Indonesia", "email": "knjbeni@corp.io", "phone": "(511) 267-0496", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Bahrain", "email": "dgbpeei@test.org", "phone": "(549) 509-4516", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Zambia", "email": "mfnfb@example.com", "phone": "(513) 937-4325", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Zambia", "email": "ajppn@example.com", "phone": "(557) 266-5542", "unit": "mi"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "ajll@example.com", "phone": "(543) 698-1593", "unit": "lb"}, {"city": "Likino-Dulevo", "country": "Indonesia", "email": "clejla@mail.com", "phone": "(579) 726-8164", "unit": "lb"}, {"city": "Trenque Lauquen", "country": "Zambia", "email": "koghebi@mail.com", "phone": "(596) 523-5609", "unit": "lb"}, {"city": "Trenque Lauquen", "country": "Zambia", "email": "aoelfdf@test.org", "phone": "(537) 980-9409", "unit": "mi"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "ncmdock@mail.com", "phone": "(535) 465-6749", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Bahrain", "email": "mbbgk@corp.io", "phone": "(595) 164-6191", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Bahrain", "email": "fekond@mail.com", "phone": "(578) 099-9060", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "dnlhak@example.com", "phone": "(576) 974-0269", "unit": "mi"}, {"city": "Likino-Dulevo", "country": "Zambia", "email": "hnfgpb@test.org", "phone": "(502) 549-0384", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Zambia", "email": "ibkjope@corp.io", "phone": "(525) 779-9132", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "dclhpe@mail.com", "phone": "(558) 279-1363", "unit": "in"}, {"city": "Trenque Lauquen", "country": "Indonesia", "email": "opicd@test.org", "phone": "(515) 568-7442", "unit": "in"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "bcdllii@test.org", "phone": "(584) 981-3732", "unit": "lb"}, {"city": "Trenque Lauquen", "country": "Bahrain", "email": "blpgab@example.com", "phone": "(559) 502-3822", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "bcngfd@example.com", "phone": "(528) 374-8218", "unit": "in"}, {"city": "Trenque Lauquen", "country": "Indonesia", "email": "ijon@test.org", "phone": "(561) 387-2067", "unit": "in"}, {"city": "Trenque Lauquen", "country": "Bahrain", "email": "adfdl@mail.com", "phone": "(546) 132-0856", "unit": "mi"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "llidgaa@example.com", "phone": "(507) 031-0267", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Indonesia", "email": "djfdf@example.com", "phone": "(532) 913-0605", "unit": "in"}, {"city": "Likino-Dulevo", "country": "Zambia", "email": "kgekhd@mail.com", "phone": "(590) 691-8368", "unit": "mi"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "ngobegg@corp.io", "phone": "(579) 578-8228", "unit": "in"}, {"city": "Trenque Lauquen", "country": "Zambia", "email": "lfnhhio@test.org", "phone": "(540) 458-5782", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Zambia", "email": "oagjg@example.com", "phone": "(558) 411-5109", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "icagoh@corp.io", "phone": "(538) 036-9985", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Zambia", "email": "lfnieg@example.com", "phone": "(538) 615-7596", "unit": "mi"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "ibfpkl@corp.io", "phone": "(565) 294-9984", "unit": "in"}, {"city": "Marbach am Neckar", "country": "Bahrain", "email": "oicb@test.org", "phone": "(518) 895-7012", "unit": "lb"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "blng@test.org", "phone": "(527) 716-5728", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Zambia", "email": "oinp@corp.io", "phone": "(532) 278-3797", "unit": "mi"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "djocg@example.com", "phone": "(527) 038-5031", "unit": "in"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "napnhne@example.com", "phone": "(580) 506-8540", "unit": "lb"}, {"city": "Trenque Lauquen", "country": "Zambia", "email": "njnccgj@mail.com", "phone": "(570) 322-2267", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Bahrain", "email": "lcap@example.com", "phone": "(554) 611-8042", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Bahrain", "email": "anjpff@test.org", "phone": "(546) 649-0120", "unit": "mi"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "gjon@mail.com", "phone": "(592) 512-1577", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Bahrain", "email": "gclfnp@corp.io", "phone": "(571) 818-5346", "unit": "in"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "lcodndh@mail.com", "phone": "(559) 531-9797", "unit": "mi"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "penoh@mail.com", "phone": "(579) 186-7285", "unit": "mi"}, {"city": "Trenque Lauquen", "country": "Indonesia", "email": "ndddp@corp.io", "phone": "(562) 971-5955", "unit": "mi"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "fdnkbag@example.com", "phone": "(513) 825-0732", "unit": "in"}, {"city": "Likino-Dulevo", "country": "Indonesia", "email": "jjhhg@mail.com", "phone": "(587) 886-3866", "unit": "lb"}, {"city": "Likino-Dulevo", "country": "Bahrain", "email": "kholdja@example.com", "phone": "(540) 707-9665", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Zambia", "email": "amiipno@example.com", "phone": "(536) 105-6793", "unit": "lb"}, {"city": "Marbach am Neckar", "country": "Indonesia", "email": "nbomee@mail.com", "phone": "(552) 520-4068", "unit": "in"}, {"city": "Likino-Dulevo", "country": "Zambia", "email": "dhmbpnf@corp.io", "phone": "(523) 612-6627", "unit": "in"}], "dirty_cols": ["city", "country", "email", "phone", "unit", "extra"], "clean_cols": ["city", "country", "email", "phone", "unit"], "plan": {"dataset_summary": "61 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TRENQUELAUQUEN": "Trenque Lauquen", "MARBACH AM NECKAR": "Marbach am Neckar", "Marbach  am Neckar": "Marbach am Neckar", "LIKINO-DULEVO": "Likino-Dulevo", "trenque lauquen": "Trenque Lauquen", "marbach am neckar": "Marbach am Neckar", "likino-dulevo": "Likino-Dulevo", "Marbach Am Neckar": "Marbach am Neckar", "likifo-dulevo": "Likino-Dulevo", "TRENQUE LAUQUEN": "Trenque Lauquen", "LikinoDulevo": "Likino-Dulevo", "marbac am neckar": "Marbach am Neckar", "Mrbach am Neckar": "Marbach am Neckar", "LIKIINO-DULEVO": "Likino-Dulevo", "Taenque Lauquen": "Trenque Lauquen", "LiinoDulevo": "Likino-Dulevo", "LIKINODULEVO": "Likino-Dulevo"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Bharain": "Bahrain", "ID": "Indonesia", "IDN": "Indonesia", "zmb": "Zambia", "bahrain": "Bahrain", "Bahryin": "Bahrain", "Republic of Zambia": "Zambia", "id": "Indonesia", "Republic of Indonesia": "Indonesia", "BH": "Bahrain", "ZM": "Zambia", "zm": "Zambia", "Kingdom of Bahrain": "Bahrain", "republic of zambia": "Zambia", "bhr": "Bahrain", "republic of indonesia": "Indonesia", "zamba": "Zambia", "ZMB": "Zambia", "repubblic of indonesia": "Indonesia", "idn": "Indonesia", "bh": "Bahrain", "Republic Of Zambia": "Zambia", "Kingdom Of Bahrain": "Bahrain", "ZMMBIA": "Zambia", "INDNESIA": "Indonesia", "BHR": "Bahrain", "Zaambia": "Zambia", "kingdom of bahrain": "Bahrain", "indonesia": "Indonesia", "ZAMBIA": "Zambia", "zambia": "Zambia"}, "rationale": "Unified 31 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"pounds": "lb", "INCH": "in", "pound": "lb", "inch": "in", "lbs": "lb", "Mi": "mi", "MILE": "mi", "incch": "in", "Miles": "mi", "Lb": "lb", "LB": "lb", "MLES": "mi", "miles": "mi", "ponds": "lb", "inches": "in", "IN": "in", "\"": "in", "In": "in", "Mile": "mi", "mile": "mi", "Inch": "in"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "CHILE  ", "industry": "property", "is_active": "Yes", "status": "medium", "company": "Cyberdyne", "email": "BIOEFI@EXAMPLE.COM", "phone": "5724083772", "department": "  Fin"}, {"country": "NLD", "industry": "Software", "is_active": "F", "status": "Medium   ", "company": "Acme Inc", "email": "MMKOKBP@EXAMPLE.COM", "phone": "592.846.4252", "department": "cs"}, {"country": "The Netherlands", "industry": "tech", "is_active": "true", "status": "low", "company": "Wonka Co ", "email": " fhaip@mail.com ", "phone": "5134398843", "department": "   R&D"}, {"country": "CL", "industry": " Mfg ", "is_active": "No", "status": "l", "company": " Cyberdyne", "email": "EGGAD@CORP.IO   ", "phone": "5129521784", "department": " operations "}, {"country": "chile", "industry": "property", "is_active": "TRUE", "status": "med  ", "company": " Initech ", "email": "jbie@test.org", "phone": "548-764-5242", "department": "Eng"}, {"country": "chl", "industry": "energy", "is_active": "F", "status": "   Medium", "company": "Acme Inc", "email": "   cpajg@mail.com", "phone": "549.463.5877", "department": "support"}, {"country": "Netherlands", "industry": "software", "is_active": "N", "status": "medium", "company": "Cyberdyne", "email": "dphde@test.org", "phone": "5957665047", "department": " spuport"}, {"country": "NLD", "industry": "ENERGY   ", "is_active": "N", "status": "MED", "company": "Wonka  Co", "email": "kjbcme@test.org", "phone": "(566)813-3694", "department": "cust support"}, {"country": "Kingdom of the Netherlands", "industry": "UTILITIIES", "is_active": "No", "status": "medium", "company": "Globex", "email": "IFOPAO@CORP.IO", "phone": "547.530.4297", "department": "DEV"}, {"country": "Nl", "industry": "Utilities", "is_active": "N", "status": "   l", "company": " Stark Industries", "email": "kjmnh@corp.io", "phone": "5611211591", "department": "customer support"}, {"country": " NL ", "industry": "Software", "is_active": "Yes", "status": "Low", "company": "Stark Industries", "email": "amhmakm@corp.io", "phone": "(565)810-2570", "department": "   OPS"}, {"country": "chl", "industry": "PopTech", "is_active": "T", "status": "medium", "company": " Umbrella ", "email": "   oabgjb@test.org", "phone": "5718086509", "department": "Ops"}, {"country": "CHL ", "industry": "PROPERTY", "is_active": "Y", "status": " low ", "company": "Umbrella", "email": "   JLOJK@CORP.IO", "phone": "598.290.1646", "department": "Ops"}, {"country": "Netherlands", "industry": "Property", "is_active": "Yes", "status": "M", "company": "   Cyberdyne", "email": "CKAFOE@TEST.ORG", "phone": "(517)172-0336", "department": "   Eng"}, {"country": "  CHL", "industry": "MFG   ", "is_active": "TRUE", "status": "medium", "company": "Acme Inc", "email": "   DFBKKDI@CORP.IO", "phone": "(503)213-1450", "department": "Operations"}, {"country": "nl", "industry": "PROPERTY", "is_active": "T", "status": "p2", "company": "Soylent Corp", "email": "EDFCH@MAIL.COM", "phone": "572-796-5881", "department": "Eng"}, {"country": "Netherlajds", "industry": "   Mfg", "is_active": "Yes", "status": "M   ", "company": " Wonka Co", "email": "djog@example.com", "phone": "5625450195", "department": "Finance"}, {"country": "The Netherlands", "industry": "Techvology", "is_active": "Y", "status": "M", "company": "Stark Industries", "email": " ldip@example.com ", "phone": "5458688035", "department": "Ops"}, {"country": " CHL ", "industry": "propech", "is_active": "T", "status": "low", "company": "Vehement", "email": "CKLJ@MAIL.COM", "phone": "579-287-9473", "department": "fin"}, {"country": "REPUBLIC OF CHILE", "industry": "Real Estate", "is_active": "Y", "status": "L", "company": "   Hooli", "email": "PKOA@TEST.ORG", "phone": "522-092-2822", "department": "Support"}, {"country": "Kingdom Of The Netherlands", "industry": "Tecvnology", "is_active": "T", "status": "P3", "company": "Cyberdyne", "email": "NLNM@TEST.ORG ", "phone": "5093203087", "department": "Support"}, {"country": " CL ", "industry": "Tehcnology", "is_active": "1", "status": "p3", "company": "Globex", "email": "pdgeccp@test.org", "phone": "(535)652-7545", "department": "Ops"}, {"country": "Republic  of Chile", "industry": "technology", "is_active": "FALSE", "status": "Medium", "company": "Vehement ", "email": " JBHDI@MAIL.COM", "phone": "584.024.3359", "department": "CUST SUPPORT"}, {"country": "Republic Of Chile  ", "industry": "software", "is_active": "FALSE", "status": "Low", "company": "Hooli", "email": "diaoglo@example.com", "phone": "5485548242", "department": "   Accounting"}, {"country": "NLD", "industry": "manufacturing", "is_active": "1", "status": "P3", "company": "   Hooli", "email": " jchg@corp.io ", "phone": "(545)809-2645", "department": "Cust Support"}, {"country": "republic of cchile", "industry": "Utilities", "is_active": "FALSE", "status": "l", "company": "Umbrella", "email": "   KCHKNCJ@MAIL.COM", "phone": "5974713473", "department": "   Engineering"}, {"country": "NL", "industry": "Eergy", "is_active": "No", "status": "L", "company": "  Hooli", "email": "aideggg@corp.io", "phone": "(590)273-6143", "department": "ops"}, {"country": "cl", "industry": " realestate ", "is_active": "false", "status": " medium ", "company": "Soylent  Corp", "email": "miigaa@corp.io", "phone": "5824198622", "department": " OPERATIONS "}, {"country": "CL", "industry": "Real Estate", "is_active": "false", "status": "low", "company": " Vehement ", "email": "OGBAKE@MAIL.COM", "phone": "503-580-7171", "department": "cs"}, {"country": "Republic Of Chile", "industry": " PropTech ", "is_active": "T", "status": "P2", "company": "Stark  Industries", "email": "cpgejgf@mail.com  ", "phone": "(594)432-0899", "department": "FINANCE  "}, {"country": "CL", "industry": "Industrial", "is_active": "0", "status": "medium", "company": "Wonka Co", "email": "KMCMP@CORP.IO", "phone": "(546)985-1456", "department": "R&D"}, {"country": "netherlands", "industry": "Energy", "is_active": "0", "status": "M  ", "company": "Wonka Co", "email": "  eamiglo@test.org", "phone": "5728708014", "department": "Eng."}, {"country": "nld", "industry": "energy", "is_active": "N", "status": "   L", "company": "Vehement", "email": "nfokhk@test.org", "phone": "5718158956", "department": "Cust Support"}, {"country": "chile", "industry": " realestate ", "is_active": "Y", "status": "l", "company": " Initech ", "email": "FAOFMNA@TEST.ORG", "phone": "(546)711-4546", "department": "Ops"}, {"country": "Nl", "industry": "Energy", "is_active": "FALSE", "status": "l", "company": "Acme  Inc", "email": "khbkd@test.org", "phone": "(544)091-1967", "department": "CS"}, {"country": "CHL   ", "industry": "Technology", "is_active": "FALSE", "status": " l", "company": "Wonka Co", "email": "CIGMMIP@CORP.IO ", "phone": "594-953-6100", "department": "fin"}, {"country": "cl", "industry": " realestate ", "is_active": "false", "status": " medium ", "company": "Soylent  Corp", "email": "miigaa@corp.io", "phone": "5824198622", "department": " OPERATIONS "}], "clean": [{"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Medium", "company": "Cyberdyne", "email": "bioefi@example.com", "phone": "(572) 408-3772", "department": "Finance"}, {"country": "Netherlands", "industry": "Technology", "is_active": false, "status": "Medium", "company": "Acme Inc", "email": "mmkokbp@example.com", "phone": "(592) 846-4252", "department": "Customer Support"}, {"country": "Netherlands", "industry": "Technology", "is_active": true, "status": "Low", "company": "Wonka Co", "email": "fhaip@mail.com", "phone": "(513) 439-8843", "department": "Engineering"}, {"country": "Chile", "industry": "Manufacturing", "is_active": false, "status": "Low", "company": "Cyberdyne", "email": "eggad@corp.io", "phone": "(512) 952-1784", "department": "Operations"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Medium", "company": "Initech", "email": "jbie@test.org", "phone": "(548) 764-5242", "department": "Engineering"}, {"country": "Chile", "industry": "Energy", "is_active": false, "status": "Medium", "company": "Acme Inc", "email": "cpajg@mail.com", "phone": "(549) 463-5877", "department": "Customer Support"}, {"country": "Netherlands", "industry": "Technology", "is_active": false, "status": "Medium", "company": "Cyberdyne", "email": "dphde@test.org", "phone": "(595) 766-5047", "department": "Customer Support"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Medium", "company": "Wonka Co", "email": "kjbcme@test.org", "phone": "(566) 813-3694", "department": "Customer Support"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Medium", "company": "Globex", "email": "ifopao@corp.io", "phone": "(547) 530-4297", "department": "Engineering"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Low", "company": "Stark Industries", "email": "kjmnh@corp.io", "phone": "(561) 121-1591", "department": "Customer Support"}, {"country": "Netherlands", "industry": "Technology", "is_active": true, "status": "Low", "company": "Stark Industries", "email": "amhmakm@corp.io", "phone": "(565) 810-2570", "department": "Operations"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Medium", "company": "Umbrella", "email": "oabgjb@test.org", "phone": "(571) 808-6509", "department": "Operations"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Low", "company": "Umbrella", "email": "jlojk@corp.io", "phone": "(598) 290-1646", "department": "Operations"}, {"country": "Netherlands", "industry": "Real Estate", "is_active": true, "status": "Medium", "company": "Cyberdyne", "email": "ckafoe@test.org", "phone": "(517) 172-0336", "department": "Engineering"}, {"country": "Chile", "industry": "Manufacturing", "is_active": true, "status": "Medium", "company": "Acme Inc", "email": "dfbkkdi@corp.io", "phone": "(503) 213-1450", "department": "Operations"}, {"country": "Netherlands", "industry": "Real Estate", "is_active": true, "status": "Medium", "company": "Soylent Corp", "email": "edfch@mail.com", "phone": "(572) 796-5881", "department": "Engineering"}, {"country": "Netherlands", "industry": "Manufacturing", "is_active": true, "status": "Medium", "company": "Wonka Co", "email": "djog@example.com", "phone": "(562) 545-0195", "department": "Finance"}, {"country": "Netherlands", "industry": "Technology", "is_active": true, "status": "Medium", "company": "Stark Industries", "email": "ldip@example.com", "phone": "(545) 868-8035", "department": "Operations"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Low", "company": "Vehement", "email": "cklj@mail.com", "phone": "(579) 287-9473", "department": "Finance"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Low", "company": "Hooli", "email": "pkoa@test.org", "phone": "(522) 092-2822", "department": "Customer Support"}, {"country": "Netherlands", "industry": "Technology", "is_active": true, "status": "Low", "company": "Cyberdyne", "email": "nlnm@test.org", "phone": "(509) 320-3087", "department": "Customer Support"}, {"country": "Chile", "industry": "Technology", "is_active": true, "status": "Low", "company": "Globex", "email": "pdgeccp@test.org", "phone": "(535) 652-7545", "department": "Operations"}, {"country": "Chile", "industry": "Technology", "is_active": false, "status": "Medium", "company": "Vehement", "email": "jbhdi@mail.com", "phone": "(584) 024-3359", "department": "Customer Support"}, {"country": "Chile", "industry": "Technology", "is_active": false, "status": "Low", "company": "Hooli", "email": "diaoglo@example.com", "phone": "(548) 554-8242", "department": "Finance"}, {"country": "Netherlands", "industry": "Manufacturing", "is_active": true, "status": "Low", "company": "Hooli", "email": "jchg@corp.io", "phone": "(545) 809-2645", "department": "Customer Support"}, {"country": "Chile", "industry": "Energy", "is_active": false, "status": "Low", "company": "Umbrella", "email": "kchkncj@mail.com", "phone": "(597) 471-3473", "department": "Engineering"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Low", "company": "Hooli", "email": "aideggg@corp.io", "phone": "(590) 273-6143", "department": "Operations"}, {"country": "Chile", "industry": "Real Estate", "is_active": false, "status": "Medium", "company": "Soylent Corp", "email": "miigaa@corp.io", "phone": "(582) 419-8622", "department": "Operations"}, {"country": "Chile", "industry": "Real Estate", "is_active": false, "status": "Low", "company": "Vehement", "email": "ogbake@mail.com", "phone": "(503) 580-7171", "department": "Customer Support"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Medium", "company": "Stark Industries", "email": "cpgejgf@mail.com", "phone": "(594) 432-0899", "department": "Finance"}, {"country": "Chile", "industry": "Manufacturing", "is_active": false, "status": "Medium", "company": "Wonka Co", "email": "kmcmp@corp.io", "phone": "(546) 985-1456", "department": "Engineering"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Medium", "company": "Wonka Co", "email": "eamiglo@test.org", "phone": "(572) 870-8014", "department": "Engineering"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Low", "company": "Vehement", "email": "nfokhk@test.org", "phone": "(571) 815-8956", "department": "Customer Support"}, {"country": "Chile", "industry": "Real Estate", "is_active": true, "status": "Low", "company": "Initech", "email": "faofmna@test.org", "phone": "(546) 711-4546", "department": "Operations"}, {"country": "Netherlands", "industry": "Energy", "is_active": false, "status": "Low", "company": "Acme Inc", "email": "khbkd@test.org", "phone": "(544) 091-1967", "department": "Customer Support"}, {"country": "Chile", "industry": "Technology", "is_active": false, "status": "Low", "company": "Wonka Co", "email": "cigmmip@corp.io", "phone": "(594) 953-6100", "department": "Finance"}], "dirty_cols": ["country", "industry", "is_active", "status", "company", "email", "phone", "department"], "clean_cols": ["country", "industry", "is_active", "status", "company", "email", "phone", "department"], "plan": {"dataset_summary": "37 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CHILE": "Chile", "NLD": "Netherlands", "The Netherlands": "Netherlands", "CL": "Chile", "chile": "Chile", "chl": "Chile", "Kingdom of the Netherlands": "Netherlands", "Nl": "Netherlands", "NL": "Netherlands", "CHL": "Chile", "nl": "Netherlands", "Netherlajds": "Netherlands", "REPUBLIC OF CHILE": "Chile", "Kingdom Of The Netherlands": "Netherlands", "Republic of Chile": "Chile", "Republic Of Chile": "Chile", "republic of cchile": "Chile", "cl": "Chile", "netherlands": "Netherlands", "nld": "Netherlands"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"property": "Real Estate", "Software": "Technology", "tech": "Technology", "Mfg": "Manufacturing", "energy": "Energy", "software": "Technology", "ENERGY": "Energy", "UTILITIIES": "Energy", "Utilities": "Energy", "PopTech": "Real Estate", "PROPERTY": "Real Estate", "Property": "Real Estate", "MFG": "Manufacturing", "Techvology": "Technology", "propech": "Real Estate", "Tecvnology": "Technology", "Tehcnology": "Technology", "technology": "Technology", "manufacturing": "Manufacturing", "Eergy": "Energy", "realestate": "Real Estate", "PropTech": "Real Estate", "Industrial": "Manufacturing"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"medium": "Medium", "low": "Low", "l": "Low", "med": "Medium", "MED": "Medium", "M": "Medium", "p2": "Medium", "L": "Low", "P3": "Low", "p3": "Low", "P2": "Medium"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Fin": "Finance", "cs": "Customer Support", "R&D": "Engineering", "operations": "Operations", "Eng": "Engineering", "support": "Customer Support", "spuport": "Customer Support", "cust support": "Customer Support", "DEV": "Engineering", "customer support": "Customer Support", "OPS": "Operations", "Ops": "Operations", "fin": "Finance", "Support": "Customer Support", "CUST SUPPORT": "Customer Support", "Accounting": "Finance", "Cust Support": "Customer Support", "ops": "Operations", "OPERATIONS": "Operations", "FINANCE": "Finance", "Eng.": "Engineering", "CS": "Customer Support"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": " IT ", "signup_date": "5 Jul 2023", "city": "HOUSTON", "phone": "5268364156", "email": "MMGII@CORP.IO", "extra": ""}, {"department": "IT", "signup_date": "10 Apr 2023", "city": "crespo", "phone": "5446118468", "email": " fmcekhg@corp.io ", "extra": ""}, {"department": "OPERATIONS", "signup_date": "13 Mar 2023", "city": "shanghai", "phone": "5581280881", "email": "GHCBLDE@EXAMPLE.COM", "extra": ""}, {"department": " IT ", "signup_date": "10/2/2023", "city": "Shanghai   ", "phone": "563-249-6776", "email": "OHPKDD@CORP.IO ", "extra": ""}, {"department": "Pesple Ops", "signup_date": "45123", "city": "  Shanghhi", "phone": "521.369.7237", "email": "GEGDHP@MAIL.COM  ", "extra": ""}, {"department": "HR", "signup_date": "45095", "city": "Shanghai", "phone": "599-004-1824", "email": "OMGJ@EXAMPLE.COM   ", "extra": ""}, {"department": "Information Technology", "signup_date": "2023-01-03", "city": "crespo", "phone": "538-984-1971", "email": "HBGBCD@MAIL.COM", "extra": ""}, {"department": " HR", "signup_date": "45187", "city": "Manila", "phone": "526.519.1209", "email": "cdoake@test.org", "extra": ""}, {"department": "I.T.", "signup_date": "2023-07-25", "city": "manila", "phone": "5320456033", "email": "aclfg@test.org", "extra": ""}, {"department": "Operations", "signup_date": "6/18/2023", "city": "Mumbi", "phone": "514-714-7082", "email": "OGHPE@CORP.IO", "extra": ""}, {"department": "Human Resources", "signup_date": "5/1/2023", "city": "Crepso", "phone": "511.584.9497", "email": "bcjdcme@mail.com", "extra": ""}, {"department": " Operations", "signup_date": "44989", "city": "manila", "phone": "5056658772", "email": "pmioeef@mail.com", "extra": ""}, {"department": "I.T.", "signup_date": "5 Sep 2023", "city": "Cresgo", "phone": "(512)715-6871", "email": "NHPMN@EXAMPLE.COM", "extra": ""}, {"department": "Information Technology", "signup_date": "22 Dec 2023", "city": "HOUSTON   ", "phone": "(536)151-8525", "email": "CLPKFNP@EXAMPLE.COM   ", "extra": ""}, {"department": "  Human Resources", "signup_date": "2023-04-08", "city": "  Manila", "phone": "5258353034", "email": "BOPH@EXAMPLE.COM", "extra": ""}, {"department": "human resources", "signup_date": "5 Oct 2023", "city": "Mumbai", "phone": "(542)001-8198", "email": " gnepo@corp.io", "extra": ""}, {"department": "it", "signup_date": "45243", "city": " SHANGGHAI ", "phone": "(560)429-8477", "email": " onhhil@example.com", "extra": ""}, {"department": "it", "signup_date": "3/16/2023", "city": "Manila", "phone": "(505)681-1195", "email": "lbaljoi@test.org", "extra": ""}, {"department": "  Opertions", "signup_date": "2023-06-24", "city": "Manila", "phone": "5043645022", "email": "nkahf@example.com", "extra": ""}, {"department": "IT", "signup_date": "19 Mar 2023", "city": "mumbai", "phone": "515-838-7933", "email": "KCCI@EXAMPLE.COM ", "extra": ""}, {"department": "HUMAN RESOURCES", "signup_date": "45054", "city": "SHANGHAI", "phone": "588.638.4651", "email": "kodmpaj@example.com", "extra": ""}, {"department": "  people", "signup_date": "45231", "city": "   crespo", "phone": "535.462.8469", "email": "ageeca@mail.com", "extra": ""}, {"department": "operations", "signup_date": "12/4/2023", "city": "crespo", "phone": "561-872-4373", "email": "PNJD@TEST.ORG  ", "extra": ""}, {"department": "i.t.", "signup_date": "2/14/2023", "city": "Mumbai", "phone": "544.702.3422", "email": "HOMOH@CORP.IO", "extra": ""}, {"department": "Human Resources", "signup_date": "2023-08-06", "city": "Houston", "phone": "5341533842", "email": "PNINCO@MAIL.COM   ", "extra": ""}, {"department": " i.t.", "signup_date": "1/15/2023", "city": "Houston ", "phone": "545-197-0905", "email": "GNPMGE@EXAMPLE.COM", "extra": ""}, {"department": "Operations", "signup_date": "2023-04-14", "city": " crespo ", "phone": "(512)357-0010", "email": "OKLHJ@TEST.ORG ", "extra": ""}, {"department": "people", "signup_date": "2023-06-02", "city": "Shanghai", "phone": "509.674.2987", "email": "naop@test.org   ", "extra": ""}, {"department": "ops", "signup_date": "2023-05-13", "city": "housion", "phone": "5774803019", "email": "   mgec@example.com", "extra": ""}, {"department": " Ops", "signup_date": "4 Aug 2023", "city": "Shanghai", "phone": "506.669.6957", "email": "ACDJDH@CORP.IO", "extra": ""}, {"department": "people ops", "signup_date": "2023-09-09", "city": "Manila", "phone": "576-909-1325", "email": "KDMDB@TEST.ORG", "extra": ""}, {"department": "Hr", "signup_date": "7/5/2023", "city": "Bombay", "phone": "5858529398", "email": "GFGK@CORP.IO  ", "extra": ""}, {"department": "People", "signup_date": "22 Jan 2023", "city": "houston", "phone": "(514)334-0158", "email": "   bbeoa@mail.com", "extra": ""}, {"department": "OPS", "signup_date": "2023-12-23", "city": "BOMBAY", "phone": "(598)327-4716", "email": "BAADOH@MAIL.COM", "extra": ""}, {"department": "People", "signup_date": "27 Jun 2023", "city": "Manila", "phone": "(505)242-7201", "email": " eplcn@test.org ", "extra": ""}, {"department": " INFORMATION TECHNOLOGY", "signup_date": "11/9/2023", "city": "manila", "phone": "5744583644", "email": "okcea@example.com", "extra": ""}, {"department": "Information Technology", "signup_date": "2023-12-18", "city": "houston", "phone": "(591)927-8998", "email": "FFFF@EXAMPLE.COM", "extra": ""}, {"department": "Ops", "signup_date": "45003", "city": " mumbai ", "phone": "5267248422", "email": "hlkkl@corp.io   ", "extra": ""}, {"department": "People", "signup_date": "2023-03-10", "city": "Shanghai", "phone": "(577)878-5740", "email": " HIICCMI@EXAMPLE.COM", "extra": ""}, {"department": "Infrmation  Technology", "signup_date": "2023-07-04", "city": "Bombaay", "phone": "(560)783-6845", "email": "   PJMPODI@CORP.IO", "extra": ""}, {"department": "IT", "signup_date": "45134", "city": " crespo ", "phone": "501-420-1458", "email": "FNMIGAH@EXAMPLE.COM", "extra": ""}, {"department": "Ops", "signup_date": "2023-08-27", "city": "Houston", "phone": "5837292646", "email": "EPHCGG@EXAMPLE.COM ", "extra": ""}, {"department": "Information Technology", "signup_date": "14 Oct 2023", "city": " bombay ", "phone": "(523)516-2167", "email": "ifgk@test.org", "extra": ""}, {"department": "  OPERATIONS", "signup_date": "11/21/2023", "city": "CRESPO", "phone": "595.978.9255", "email": " iidb@mail.com ", "extra": ""}, {"department": "  Opertions", "signup_date": "2023-06-24", "city": "Manila", "phone": "5043645022", "email": "nkahf@example.com", "extra": ""}, {"department": "", "signup_date": "", "city": "", "phone": "", "email": "", "extra": ""}], "clean": [{"department": "Information Technology", "signup_date": "2023-07-05", "city": "Houston", "phone": "(526) 836-4156", "email": "mmgii@corp.io"}, {"department": "Information Technology", "signup_date": "2023-04-10", "city": "Crespo", "phone": "(544) 611-8468", "email": "fmcekhg@corp.io"}, {"department": "Operations", "signup_date": "2023-03-13", "city": "Shanghai", "phone": "(558) 128-0881", "email": "ghcblde@example.com"}, {"department": "Information Technology", "signup_date": "2023-10-02", "city": "Shanghai", "phone": "(563) 249-6776", "email": "ohpkdd@corp.io"}, {"department": "Human Resources", "signup_date": "2023-07-16", "city": "Shanghai", "phone": "(521) 369-7237", "email": "gegdhp@mail.com"}, {"department": "Human Resources", "signup_date": "2023-06-18", "city": "Shanghai", "phone": "(599) 004-1824", "email": "omgj@example.com"}, {"department": "Information Technology", "signup_date": "2023-01-03", "city": "Crespo", "phone": "(538) 984-1971", "email": "hbgbcd@mail.com"}, {"department": "Human Resources", "signup_date": "2023-09-18", "city": "Manila", "phone": "(526) 519-1209", "email": "cdoake@test.org"}, {"department": "Information Technology", "signup_date": "2023-07-25", "city": "Manila", "phone": "(532) 045-6033", "email": "aclfg@test.org"}, {"department": "Operations", "signup_date": "2023-06-18", "city": "Mumbai", "phone": "(514) 714-7082", "email": "oghpe@corp.io"}, {"department": "Human Resources", "signup_date": "2023-05-01", "city": "Crespo", "phone": "(511) 584-9497", "email": "bcjdcme@mail.com"}, {"department": "Operations", "signup_date": "2023-03-04", "city": "Manila", "phone": "(505) 665-8772", "email": "pmioeef@mail.com"}, {"department": "Information Technology", "signup_date": "2023-09-05", "city": "Crespo", "phone": "(512) 715-6871", "email": "nhpmn@example.com"}, {"department": "Information Technology", "signup_date": "2023-12-22", "city": "Houston", "phone": "(536) 151-8525", "email": "clpkfnp@example.com"}, {"department": "Human Resources", "signup_date": "2023-04-08", "city": "Manila", "phone": "(525) 835-3034", "email": "boph@example.com"}, {"department": "Human Resources", "signup_date": "2023-10-05", "city": "Mumbai", "phone": "(542) 001-8198", "email": "gnepo@corp.io"}, {"department": "Information Technology", "signup_date": "2023-11-13", "city": "Shanghai", "phone": "(560) 429-8477", "email": "onhhil@example.com"}, {"department": "Information Technology", "signup_date": "2023-03-16", "city": "Manila", "phone": "(505) 681-1195", "email": "lbaljoi@test.org"}, {"department": "Operations", "signup_date": "2023-06-24", "city": "Manila", "phone": "(504) 364-5022", "email": "nkahf@example.com"}, {"department": "Information Technology", "signup_date": "2023-03-19", "city": "Mumbai", "phone": "(515) 838-7933", "email": "kcci@example.com"}, {"department": "Human Resources", "signup_date": "2023-05-08", "city": "Shanghai", "phone": "(588) 638-4651", "email": "kodmpaj@example.com"}, {"department": "Human Resources", "signup_date": "2023-11-01", "city": "Crespo", "phone": "(535) 462-8469", "email": "ageeca@mail.com"}, {"department": "Operations", "signup_date": "2023-12-04", "city": "Crespo", "phone": "(561) 872-4373", "email": "pnjd@test.org"}, {"department": "Information Technology", "signup_date": "2023-02-14", "city": "Mumbai", "phone": "(544) 702-3422", "email": "homoh@corp.io"}, {"department": "Human Resources", "signup_date": "2023-08-06", "city": "Houston", "phone": "(534) 153-3842", "email": "pninco@mail.com"}, {"department": "Information Technology", "signup_date": "2023-01-15", "city": "Houston", "phone": "(545) 197-0905", "email": "gnpmge@example.com"}, {"department": "Operations", "signup_date": "2023-04-14", "city": "Crespo", "phone": "(512) 357-0010", "email": "oklhj@test.org"}, {"department": "Human Resources", "signup_date": "2023-06-02", "city": "Shanghai", "phone": "(509) 674-2987", "email": "naop@test.org"}, {"department": "Operations", "signup_date": "2023-05-13", "city": "Houston", "phone": "(577) 480-3019", "email": "mgec@example.com"}, {"department": "Operations", "signup_date": "2023-08-04", "city": "Shanghai", "phone": "(506) 669-6957", "email": "acdjdh@corp.io"}, {"department": "Human Resources", "signup_date": "2023-09-09", "city": "Manila", "phone": "(576) 909-1325", "email": "kdmdb@test.org"}, {"department": "Human Resources", "signup_date": "2023-07-05", "city": "Mumbai", "phone": "(585) 852-9398", "email": "gfgk@corp.io"}, {"department": "Human Resources", "signup_date": "2023-01-22", "city": "Houston", "phone": "(514) 334-0158", "email": "bbeoa@mail.com"}, {"department": "Operations", "signup_date": "2023-12-23", "city": "Mumbai", "phone": "(598) 327-4716", "email": "baadoh@mail.com"}, {"department": "Human Resources", "signup_date": "2023-06-27", "city": "Manila", "phone": "(505) 242-7201", "email": "eplcn@test.org"}, {"department": "Information Technology", "signup_date": "2023-11-09", "city": "Manila", "phone": "(574) 458-3644", "email": "okcea@example.com"}, {"department": "Information Technology", "signup_date": "2023-12-18", "city": "Houston", "phone": "(591) 927-8998", "email": "ffff@example.com"}, {"department": "Operations", "signup_date": "2023-03-18", "city": "Mumbai", "phone": "(526) 724-8422", "email": "hlkkl@corp.io"}, {"department": "Human Resources", "signup_date": "2023-03-10", "city": "Shanghai", "phone": "(577) 878-5740", "email": "hiiccmi@example.com"}, {"department": "Information Technology", "signup_date": "2023-07-04", "city": "Mumbai", "phone": "(560) 783-6845", "email": "pjmpodi@corp.io"}, {"department": "Information Technology", "signup_date": "2023-07-27", "city": "Crespo", "phone": "(501) 420-1458", "email": "fnmigah@example.com"}, {"department": "Operations", "signup_date": "2023-08-27", "city": "Houston", "phone": "(583) 729-2646", "email": "ephcgg@example.com"}, {"department": "Information Technology", "signup_date": "2023-10-14", "city": "Mumbai", "phone": "(523) 516-2167", "email": "ifgk@test.org"}, {"department": "Operations", "signup_date": "2023-11-21", "city": "Crespo", "phone": "(595) 978-9255", "email": "iidb@mail.com"}], "dirty_cols": ["department", "signup_date", "city", "phone", "email", "extra"], "clean_cols": ["department", "signup_date", "city", "phone", "email"], "plan": {"dataset_summary": "46 rows × 6 columns. 5 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "OPERATIONS": "Operations", "Pesple Ops": "Human Resources", "HR": "Human Resources", "I.T.": "Information Technology", "human resources": "Human Resources", "it": "Information Technology", "Opertions": "Operations", "HUMAN RESOURCES": "Human Resources", "people": "Human Resources", "operations": "Operations", "i.t.": "Information Technology", "ops": "Operations", "Ops": "Operations", "people ops": "Human Resources", "Hr": "Human Resources", "People": "Human Resources", "OPS": "Operations", "INFORMATION TECHNOLOGY": "Information Technology", "Infrmation Technology": "Information Technology"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"HOUSTON": "Houston", "crespo": "Crespo", "shanghai": "Shanghai", "Shanghhi": "Shanghai", "manila": "Manila", "Mumbi": "Mumbai", "Crepso": "Crespo", "Cresgo": "Crespo", "SHANGGHAI": "Shanghai", "mumbai": "Mumbai", "SHANGHAI": "Shanghai", "housion": "Houston", "Bombay": "Mumbai", "houston": "Houston", "BOMBAY": "Mumbai", "Bombaay": "Mumbai", "bombay": "Mumbai", "CRESPO": "Crespo"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"state": "GU", "unit": "\"", "country": "Islamic Republic Of Afghanistan", "email": "PPLKFFA@CORP.IO", "status": "cancelled", "phone": "(559)418-7848"}, {"state": "Illinois", "unit": "centimeters", "country": "AFG", "email": "hpgf@example.com  ", "status": "cancelled", "phone": "5858810795"}, {"state": "Illinois", "unit": "in", "country": "Islamic Republic of Afghanistan", "email": "INPCOL@TEST.ORG  ", "status": "PAUSED", "phone": "5069159271"}, {"state": "GU", "unit": "Lb", "country": "SVK", "email": " cefe@example.com ", "status": "Active", "phone": "5064245704"}, {"state": "IL   ", "unit": "Inches", "country": "SK  ", "email": "  AJDKI@MAIL.COM", "status": " Active", "phone": "573.589.0532"}, {"state": "GU", "unit": "cm", "country": "Slovakia", "email": " henfkh@corp.io", "status": "TRIAL", "phone": "(599)966-4957"}, {"state": "  il", "unit": " Cm ", "country": "AF", "email": "LOHH@TEST.ORG", "status": "   cnceled", "phone": "5162937272"}, {"state": "gu", "unit": "Lb ", "country": "SLOVAK REPUBLIC  ", "email": "cioeaae@corp.io", "status": "churned", "phone": "535-500-1052"}, {"state": "il", "unit": "POUNDS", "country": "hu", "email": " CCJIM@TEST.ORG ", "status": " churned ", "phone": "507.389.6112"}, {"state": "Illinois", "unit": "\"", "country": "AFG  ", "email": "jailhh@example.com", "status": "free trial", "phone": "(598)304-2325"}, {"state": "Guam", "unit": "pound", "country": "HUN", "email": "   LBLICOF@EXAMPLE.COM", "status": "canceelled", "phone": "530-147-0521"}, {"state": "guam", "unit": "cm", "country": "Hungary", "email": "LGPFED@EXAMPLE.COM", "status": "  churned", "phone": "557.626.3897"}, {"state": "gu ", "unit": "Cm", "country": "Hungary", "email": "chkoil@example.com", "status": "Active", "phone": "5904101378"}, {"state": "il", "unit": "INCHES  ", "country": "Hungary", "email": "cohp@corp.io", "status": " canceled ", "phone": "5768861619"}, {"state": "ilinois", "unit": "pouunds", "country": "AFG", "email": " EPCJC@CORP.IO ", "status": "ACTIVE", "phone": "506-048-4939"}, {"state": " Guam", "unit": "CENTIMETERS", "country": "Slovakia", "email": "maipdep@example.com", "status": "free trial", "phone": "(576)205-4481"}, {"state": "GU", "unit": "centimetters", "country": "hu", "email": "icdm@example.com", "status": "ACTIVE", "phone": "594.528.5101"}, {"state": "GU", "unit": "CM", "country": "SVK", "email": "lfalkpf@test.org", "status": "   Trial", "phone": "565-849-8649"}, {"state": "gu", "unit": "inch", "country": "HU", "email": "jlhe@test.org", "status": " active", "phone": "(513)686-3995"}, {"state": "   illinois", "unit": "in", "country": " HU ", "email": " JMEEGC@MAIL.COM", "status": " canpeled ", "phone": "(543)186-2985"}, {"state": " IL ", "unit": "lbs", "country": "AFGHANISTAN", "email": " johc@mail.com ", "status": "on hold", "phone": "5829972016"}, {"state": "guam", "unit": " ceentimeters ", "country": "Afghanistan", "email": "kckofi@test.org", "status": "Churned", "phone": "550.752.2242"}, {"state": "illinois", "unit": "centimeters", "country": "Hungary", "email": "ljapl@corp.io", "status": "paused", "phone": "500-790-2745"}, {"state": "Illinois", "unit": "inches", "country": "SK", "email": "kdlnc@corp.io", "status": "on hold", "phone": "569-776-3355"}, {"state": "ILLINOIS", "unit": "cm", "country": "SK   ", "email": "GNKBD@MAIL.COM", "status": "churned", "phone": "(534)450-6776"}, {"state": "Illinois", "unit": "inches", "country": "SVK", "email": "GJGILNK@MAIL.COM", "status": "paused", "phone": "517.034.1069"}, {"state": "GUM", "unit": "inch", "country": "   Hu", "email": " AKPNE@MAIL.COM", "status": "active", "phone": "562.129.0498"}, {"state": "guam", "unit": "centimeter", "country": "sk", "email": "iigc@corp.io", "status": " cancelled", "phone": "(539)926-0197"}, {"state": "Guam", "unit": " centimeter ", "country": "hungary", "email": "  ckma@mail.com", "status": "active", "phone": "5651774702"}, {"state": "GU", "unit": "lb", "country": "AF", "email": "PEGKGEM@MAIL.COM", "status": "chuned ", "phone": "525.682.2388"}, {"state": " il", "unit": "lb", "country": "SLOVAK REPUBLIC", "email": "BLDLCGD@CORP.IO  ", "status": "CANCELED  ", "phone": "519.954.3299"}, {"state": "IL  ", "unit": "pgund", "country": " ISLAMIC REPUBLIC OF AFGHANISTAN", "email": "onlg@example.com", "status": "ACTIVE", "phone": "515.810.6668"}, {"state": "illinois", "unit": "\"", "country": "Hungary  ", "email": "caok@corp.io", "status": " CANCELED ", "phone": "548-445-1935"}, {"state": "Il", "unit": "centimeter", "country": "sk", "email": " jdoe@test.org", "status": "ACTIVE", "phone": "529-310-5342"}, {"state": "GUAM", "unit": "Centimeter", "country": "afghanistan", "email": "faakeec@mail.com ", "status": "paused", "phone": "563.625.6441"}, {"state": "il", "unit": " lbs ", "country": "svk", "email": "BJOOIA@EXAMPLE.COM", "status": "Churned", "phone": "575-021-4655"}, {"state": "GU", "unit": " INCH", "country": "SVK", "email": "phkia@test.org", "status": "churned", "phone": "5480365414"}, {"state": "GU", "unit": "cm ", "country": "hungary", "email": "CGHM@EXAMPLE.COM", "status": "ACTIVE", "phone": "5083384772"}, {"state": "Il", "unit": " inch ", "country": " SK ", "email": "CLJEDFL@CORP.IO", "status": " trial ", "phone": "563.755.2272"}], "clean": [{"state": "Guam", "unit": "in", "country": "Afghanistan", "email": "pplkffa@corp.io", "status": "Churned", "phone": "(559) 418-7848"}, {"state": "Illinois", "unit": "cm", "country": "Afghanistan", "email": "hpgf@example.com", "status": "Churned", "phone": "(585) 881-0795"}, {"state": "Illinois", "unit": "in", "country": "Afghanistan", "email": "inpcol@test.org", "status": "Paused", "phone": "(506) 915-9271"}, {"state": "Guam", "unit": "lb", "country": "Slovakia", "email": "cefe@example.com", "status": "Active", "phone": "(506) 424-5704"}, {"state": "Illinois", "unit": "in", "country": "Slovakia", "email": "ajdki@mail.com", "status": "Active", "phone": "(573) 589-0532"}, {"state": "Guam", "unit": "cm", "country": "Slovakia", "email": "henfkh@corp.io", "status": "Trial", "phone": "(599) 966-4957"}, {"state": "Illinois", "unit": "cm", "country": "Afghanistan", "email": "lohh@test.org", "status": "Churned", "phone": "(516) 293-7272"}, {"state": "Guam", "unit": "lb", "country": "Slovakia", "email": "cioeaae@corp.io", "status": "Churned", "phone": "(535) 500-1052"}, {"state": "Illinois", "unit": "lb", "country": "Hungary", "email": "ccjim@test.org", "status": "Churned", "phone": "(507) 389-6112"}, {"state": "Illinois", "unit": "in", "country": "Afghanistan", "email": "jailhh@example.com", "status": "Trial", "phone": "(598) 304-2325"}, {"state": "Guam", "unit": "lb", "country": "Hungary", "email": "lblicof@example.com", "status": "Churned", "phone": "(530) 147-0521"}, {"state": "Guam", "unit": "cm", "country": "Hungary", "email": "lgpfed@example.com", "status": "Churned", "phone": "(557) 626-3897"}, {"state": "Guam", "unit": "cm", "country": "Hungary", "email": "chkoil@example.com", "status": "Active", "phone": "(590) 410-1378"}, {"state": "Illinois", "unit": "in", "country": "Hungary", "email": "cohp@corp.io", "status": "Churned", "phone": "(576) 886-1619"}, {"state": "Illinois", "unit": "lb", "country": "Afghanistan", "email": "epcjc@corp.io", "status": "Active", "phone": "(506) 048-4939"}, {"state": "Guam", "unit": "cm", "country": "Slovakia", "email": "maipdep@example.com", "status": "Trial", "phone": "(576) 205-4481"}, {"state": "Guam", "unit": "cm", "country": "Hungary", "email": "icdm@example.com", "status": "Active", "phone": "(594) 528-5101"}, {"state": "Guam", "unit": "cm", "country": "Slovakia", "email": "lfalkpf@test.org", "status": "Trial", "phone": "(565) 849-8649"}, {"state": "Guam", "unit": "in", "country": "Hungary", "email": "jlhe@test.org", "status": "Active", "phone": "(513) 686-3995"}, {"state": "Illinois", "unit": "in", "country": "Hungary", "email": "jmeegc@mail.com", "status": "Churned", "phone": "(543) 186-2985"}, {"state": "Illinois", "unit": "lb", "country": "Afghanistan", "email": "johc@mail.com", "status": "Paused", "phone": "(582) 997-2016"}, {"state": "Guam", "unit": "cm", "country": "Afghanistan", "email": "kckofi@test.org", "status": "Churned", "phone": "(550) 752-2242"}, {"state": "Illinois", "unit": "cm", "country": "Hungary", "email": "ljapl@corp.io", "status": "Paused", "phone": "(500) 790-2745"}, {"state": "Illinois", "unit": "in", "country": "Slovakia", "email": "kdlnc@corp.io", "status": "Paused", "phone": "(569) 776-3355"}, {"state": "Illinois", "unit": "cm", "country": "Slovakia", "email": "gnkbd@mail.com", "status": "Churned", "phone": "(534) 450-6776"}, {"state": "Illinois", "unit": "in", "country": "Slovakia", "email": "gjgilnk@mail.com", "status": "Paused", "phone": "(517) 034-1069"}, {"state": "Guam", "unit": "in", "country": "Hungary", "email": "akpne@mail.com", "status": "Active", "phone": "(562) 129-0498"}, {"state": "Guam", "unit": "cm", "country": "Slovakia", "email": "iigc@corp.io", "status": "Churned", "phone": "(539) 926-0197"}, {"state": "Guam", "unit": "cm", "country": "Hungary", "email": "ckma@mail.com", "status": "Active", "phone": "(565) 177-4702"}, {"state": "Guam", "unit": "lb", "country": "Afghanistan", "email": "pegkgem@mail.com", "status": "Churned", "phone": "(525) 682-2388"}, {"state": "Illinois", "unit": "lb", "country": "Slovakia", "email": "bldlcgd@corp.io", "status": "Churned", "phone": "(519) 954-3299"}, {"state": "Illinois", "unit": "lb", "country": "Afghanistan", "email": "onlg@example.com", "status": "Active", "phone": "(515) 810-6668"}, {"state": "Illinois", "unit": "in", "country": "Hungary", "email": "caok@corp.io", "status": "Churned", "phone": "(548) 445-1935"}, {"state": "Illinois", "unit": "cm", "country": "Slovakia", "email": "jdoe@test.org", "status": "Active", "phone": "(529) 310-5342"}, {"state": "Guam", "unit": "cm", "country": "Afghanistan", "email": "faakeec@mail.com", "status": "Paused", "phone": "(563) 625-6441"}, {"state": "Illinois", "unit": "lb", "country": "Slovakia", "email": "bjooia@example.com", "status": "Churned", "phone": "(575) 021-4655"}, {"state": "Guam", "unit": "in", "country": "Slovakia", "email": "phkia@test.org", "status": "Churned", "phone": "(548) 036-5414"}, {"state": "Guam", "unit": "cm", "country": "Hungary", "email": "cghm@example.com", "status": "Active", "phone": "(508) 338-4772"}, {"state": "Illinois", "unit": "in", "country": "Slovakia", "email": "cljedfl@corp.io", "status": "Trial", "phone": "(563) 755-2272"}], "dirty_cols": ["state", "unit", "country", "email", "status", "phone"], "clean_cols": ["state", "unit", "country", "email", "status", "phone"], "plan": {"dataset_summary": "39 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GU": "Guam", "IL": "Illinois", "il": "Illinois", "gu": "Guam", "guam": "Guam", "ilinois": "Illinois", "illinois": "Illinois", "ILLINOIS": "Illinois", "GUM": "Guam", "Il": "Illinois", "GUAM": "Guam"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"\"": "in", "centimeters": "cm", "Lb": "lb", "Inches": "in", "Cm": "cm", "POUNDS": "lb", "pound": "lb", "INCHES": "in", "pouunds": "lb", "CENTIMETERS": "cm", "centimetters": "cm", "CM": "cm", "inch": "in", "lbs": "lb", "ceentimeters": "cm", "inches": "in", "centimeter": "cm", "pgund": "lb", "Centimeter": "cm", "INCH": "in"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Islamic Republic Of Afghanistan": "Afghanistan", "AFG": "Afghanistan", "Islamic Republic of Afghanistan": "Afghanistan", "SVK": "Slovakia", "SK": "Slovakia", "AF": "Afghanistan", "SLOVAK REPUBLIC": "Slovakia", "hu": "Hungary", "HUN": "Hungary", "HU": "Hungary", "AFGHANISTAN": "Afghanistan", "Hu": "Hungary", "sk": "Slovakia", "hungary": "Hungary", "ISLAMIC REPUBLIC OF AFGHANISTAN": "Afghanistan", "afghanistan": "Afghanistan", "svk": "Slovakia"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"cancelled": "Churned", "PAUSED": "Paused", "TRIAL": "Trial", "cnceled": "Churned", "churned": "Churned", "free trial": "Trial", "canceelled": "Churned", "canceled": "Churned", "ACTIVE": "Active", "active": "Active", "canpeled": "Churned", "on hold": "Paused", "paused": "Paused", "chuned": "Churned", "CANCELED": "Churned", "trial": "Trial"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"name": "Grace Smith", "rate": "70.5%", "amount": "$6,767.08", "department": "Marketing", "email": "INPCJGN@MAIL.COM", "company": "   Vehement", "status": "Closed  Won", "signup_date": "2/7/2023", "col_x": ""}, {"name": "Judy Cruz ", "rate": "63.7%", "amount": "2.448,62", "department": " growth ", "email": " ehdob@example.com ", "company": "   Acme Inc", "status": "closed won", "signup_date": "45067", "col_x": ""}, {"name": "   Priya Fischer", "rate": "2.8%", "amount": "1.465,31", "department": "Prod", "email": "mkdej@example.com", "company": " Hooli", "status": "   won", "signup_date": "45128", "col_x": ""}, {"name": "  Eve Fischer", "rate": "43.5%", "amount": "$4,804.92", "department": "Growth", "email": "  BKFGH@TEST.ORG", "company": "Umbrella ", "status": "   new", "signup_date": "21 May 2023", "col_x": ""}, {"name": "Heidi  Wong", "rate": "6.4%", "amount": "$6,441.58", "department": "ENGINEERING", "email": "EAHDPL@TEST.ORG", "company": " Initech ", "status": "closed-won", "signup_date": "5 Oct 2023", "col_x": ""}, {"name": "Carol Fischer ", "rate": "83.3%", "amount": "None", "department": "Sales", "email": "pihoi@example.com", "company": "Stark Industries ", "status": "new", "signup_date": "7/10/2023", "col_x": ""}, {"name": "Ivan  Park", "rate": "3.4%", "amount": "$2,795.06", "department": "PM", "email": "PKBN@CORP.IO", "company": "Wonka  Co", "status": "LOST  ", "signup_date": "2023-09-25", "col_x": ""}, {"name": "Sara Moore  ", "rate": "69.8%", "amount": "2.314,77", "department": "r&d", "email": "dldgd@mail.com", "company": "Acme Inc  ", "status": "in prrogress", "signup_date": "13 Nov 2023", "col_x": ""}, {"name": "   Ivan Moore", "rate": "13.5%", "amount": "$5,752.24", "department": "Mktg", "email": "anfioh@example.com", "company": "Globex", "status": "new", "signup_date": "13 Jul 2023", "col_x": ""}, {"name": " Judy Ali", "rate": "29.8%", "amount": "4.703,06", "department": "mktg", "email": "icoempf@example.com", "company": "Wonka  Co", "status": "OPEN", "signup_date": "7 Jan 2023", "col_x": ""}, {"name": "Lena Adams  ", "rate": "50.8%", "amount": "$5,650.19", "department": "Salse", "email": "DHFJPN@CORP.IO", "company": "Globex", "status": "open", "signup_date": "4/7/2023", "col_x": ""}, {"name": "Lena Brandt", "rate": "65.9%", "amount": "$5,315.45", "department": " Sales ", "email": "phjkpa@test.org", "company": "Initech ", "status": "won", "signup_date": "2023-04-16", "col_x": ""}, {"name": "Heidi Johnson", "rate": "41.2%", "amount": "8.996,30", "department": "Product  ", "email": "gfpi@mail.com", "company": "Hooli   ", "status": "  In Progress", "signup_date": "2023-04-23", "col_x": ""}, {"name": "Ivan Johnson", "rate": "92.7%", "amount": "2.866,83", "department": "SALES", "email": " pikllf@mail.com", "company": "  Acme Inc", "status": "  open", "signup_date": "2023-04-15", "col_x": ""}, {"name": "Ivan Cruz", "rate": "27.4%", "amount": "$7,134.06", "department": "Dev", "email": " jooki@test.org ", "company": "Acme Inc", "status": "   new", "signup_date": "12/6/2023", "col_x": ""}, {"name": "  David Reyes", "rate": "8.3%", "amount": "$7,994.23", "department": " mkt ", "email": "gfhadh@corp.io", "company": "Initech", "status": "In  Piogress", "signup_date": "3/15/2023", "col_x": ""}, {"name": "Priya Moore  ", "rate": "19.6%", "amount": "1.085,10", "department": "Mkt", "email": "hbhkjf@example.com", "company": "  Umbrella", "status": "new", "signup_date": "10 May 2023", "col_x": ""}, {"name": "Judy Reyes   ", "rate": "48.3%", "amount": "$8,452.10", "department": "PM", "email": "pbnhopf@corp.io", "company": "Initech", "status": "   Oppen", "signup_date": "44990", "col_x": ""}, {"name": "Omar Novak", "rate": "51.6%", "amount": "6.208,72", "department": "Mkt", "email": "cpbpg@example.com", "company": "Vehement", "status": "New", "signup_date": "9 Feb 2023", "col_x": ""}, {"name": "Sara Smith", "rate": "34.0%", "amount": "$1,809.85", "department": "Growjh ", "email": "FDOFDI@EXAMPLE.COM", "company": "   Vehement", "status": "WON   ", "signup_date": "2023-08-23", "col_x": ""}, {"name": "Frank Johnson", "rate": "88.7%", "amount": "5.126,78", "department": "prod", "email": " FICBP@MAIL.COM ", "company": "Umbrella", "status": "won", "signup_date": "3/25/2023", "col_x": ""}, {"name": "Frank Moore", "rate": "16.4%", "amount": "3.145,28", "department": "marketing", "email": "DCECC@MAIL.COM", "company": "Globex ", "status": "new", "signup_date": "2023-07-24", "col_x": ""}, {"name": "Grace  Novak", "rate": "90.7%", "amount": "$6,120.68", "department": " Prod ", "email": "mkdgj@mail.com", "company": "   Stark Industries", "status": " ongoing", "signup_date": "45265", "col_x": ""}, {"name": "   Heidi Brandt", "rate": "51.8%", "amount": "$4,764.72", "department": "  mkt", "email": "nadfmg@example.com", "company": "Stark Industries", "status": "lost", "signup_date": "24 Aug 2023", "col_x": ""}, {"name": "Grace Moore", "rate": "88.4%", "amount": "4.739,17", "department": " MTKG ", "email": "anhid@corp.io", "company": "  Cyberdyne", "status": "Lost", "signup_date": "2023-01-22", "col_x": ""}, {"name": "Grace Cruz", "rate": "96.1%", "amount": "217,30", "department": " Sales ", "email": "JNNN@MAIL.COM", "company": "  Initech", "status": "ongoing   ", "signup_date": "44959", "col_x": ""}, {"name": "Frank Ali", "rate": "96.3%", "amount": "4.492,21", "department": "biz dv", "email": "GCEIK@CORP.IO", "company": "Wonka Co  ", "status": "in-progress", "signup_date": "19 Jan 2023", "col_x": ""}, {"name": "Bob Novak", "rate": "7.8%", "amount": "$5,382.39", "department": "eng ", "email": "GCKNALF@CORP.IO", "company": "Stark Industries", "status": "losst", "signup_date": "45273", "col_x": ""}, {"name": "Sara Khan", "rate": "1.6%", "amount": "$4,795.17", "department": " Markeing", "email": "njkeag@mail.com", "company": "Vehement", "status": " in-progress", "signup_date": "5 Dec 2023", "col_x": ""}, {"name": "  Carol Park", "rate": "20.3%", "amount": "$1,659.69", "department": "Mktg", "email": "OOGFEE@TEST.ORG", "company": "Hooli", "status": "closed won", "signup_date": "44969", "col_x": ""}, {"name": "Lena Lee", "rate": "35.5%", "amount": "3.932,19", "department": "Prod", "email": " PDENCPK@CORP.IO ", "company": "Cyberdyne", "status": "ongoig", "signup_date": "2023-11-26", "col_x": ""}, {"name": "Lena Fischer ", "rate": "72.8%", "amount": "$7,862.61", "department": "R&D", "email": " cape@test.org ", "company": "Soylent Corp", "status": " closed-won", "signup_date": "2023-10-10", "col_x": ""}, {"name": "Carol Smith", "rate": "55.3%", "amount": "$7,547.44", "department": " engineering ", "email": " APGLB@EXAMPLE.COM ", "company": "Globex   ", "status": " WIP", "signup_date": "2023-09-10", "col_x": ""}, {"name": "Alice Brandt", "rate": "59.3%", "amount": "332,75", "department": "Mkt", "email": "enfkcfm@mail.com", "company": "Acme Inc", "status": "ongning", "signup_date": "2023-07-25", "col_x": ""}, {"name": "Judy Fischer", "rate": "54.5%", "amount": "$6,452.82", "department": "  Product", "email": "dcdldln@mail.com", "company": "Hooli", "status": "closed-won", "signup_date": "2023-04-17", "col_x": ""}, {"name": "Heidi Cruz", "rate": "7.5%", "amount": "$6,211.03", "department": "mkt", "email": "GFCH@EXAMPLE.COM", "company": "Soylent Corp", "status": "Open", "signup_date": "45193", "col_x": ""}, {"name": "   David Khan", "rate": "93.3%", "amount": "5.432,50", "department": "   PROD", "email": "KPBN@MAIL.COM   ", "company": "Acme Inc", "status": "WON", "signup_date": "45056", "col_x": ""}, {"name": "Frank Khan  ", "rate": "4.3%", "amount": "N/A", "department": "Engineering", "email": "BEDPJNH@MAIL.COM", "company": " Wonka Co", "status": "Closed Won", "signup_date": "2023-04-17", "col_x": ""}, {"name": "Frank  Johnson", "rate": "64.6%", "amount": "8.439,39", "department": "Dev", "email": "bgod@mail.com", "company": " Cyberdyne", "status": "Closed-Won", "signup_date": "2023-11-05", "col_x": ""}, {"name": "  Sara Cruz", "rate": "59.9%", "amount": "$7,448.08", "department": "engineering", "email": "dnjdba@example.com", "company": "Soylent Corp", "status": "  Oxen", "signup_date": "45189", "col_x": ""}, {"name": "Lena  Moore", "rate": "17.7%", "amount": "$301.22", "department": "Biz Dev", "email": "MLKK@EXAMPLE.COM", "company": "Stark  Industries", "status": "new", "signup_date": "26 Dec 2023", "col_x": ""}, {"name": "Priya Wong", "rate": "89.0%", "amount": "5.048,46", "department": "Saales", "email": "IFBJ@TEST.ORG", "company": "Soylent  Corp", "status": " ongoing ", "signup_date": "8/14/2023", "col_x": ""}, {"name": "Omar  Petrov", "rate": "96.7%", "amount": "$416.99", "department": "mktg ", "email": "GMHMNFA@TEST.ORG", "company": "Acme Inc", "status": "Closed Won", "signup_date": "2023-05-10", "col_x": ""}, {"name": "Judy Brandt", "rate": "43.5%", "amount": "8.445,96", "department": "product", "email": "NNOH@TEST.ORG", "company": "  Vehement", "status": "Open", "signup_date": "45055", "col_x": ""}, {"name": "Mona Park   ", "rate": "12.8%", "amount": "1.308,43", "department": " growth ", "email": " HCAPFK@CORP.IO", "company": "Stark Industries", "status": "won", "signup_date": "45240", "col_x": ""}, {"name": "Karl Wong", "rate": "83.3%", "amount": "$719.51", "department": "Product", "email": "EDOCA@TEST.ORG", "company": " Vehement ", "status": "new", "signup_date": "2023-01-21", "col_x": ""}, {"name": "Eve Park", "rate": "38.3%", "amount": "$2,540.66", "department": "Mktg", "email": "JALEKNF@MAIL.COM", "company": "Acme Inc", "status": "Lost", "signup_date": "6 Apr 2023", "col_x": ""}, {"name": "Bob Cruz", "rate": "42.4%", "amount": "1.351,48", "department": "prod", "email": "aidgoif@corp.io", "company": "Umbrella ", "status": "closed lost", "signup_date": "12/25/2023", "col_x": ""}, {"name": "Frank Adams", "rate": "92.1%", "amount": "$358.67", "department": "Dev ", "email": "PGPK@TEST.ORG", "company": " Cyberdyne", "status": "open", "signup_date": "11/19/2023", "col_x": ""}, {"name": "Sara Khan", "rate": "23.8%", "amount": "$6,211.15", "department": "Prod", "email": "nlpdl@test.org", "company": "Umbrella", "status": "Lost", "signup_date": "10/25/2023", "col_x": ""}, {"name": "Omar Wong  ", "rate": "5.2%", "amount": "$4,878.53", "department": "Mktg", "email": "   fjdg@test.org", "company": "   Globex", "status": "in progress", "signup_date": "2023-08-27", "col_x": ""}, {"name": "Mona  Adams", "rate": "53.9%", "amount": "8.265,55", "department": " sales ", "email": "lhjhg@corp.io", "company": "Globex", "status": " lost ", "signup_date": "7/21/2023", "col_x": ""}, {"name": "Grace Reyes", "rate": "6.0%", "amount": "7.847,04", "department": "Biz Dev", "email": "KCANJA@EXAMPLE.COM", "company": " Vehement ", "status": "WON", "signup_date": "24 Aug 2023", "col_x": ""}, {"name": "Mona  Adams", "rate": "53.9%", "amount": "8.265,55", "department": " sales ", "email": "lhjhg@corp.io", "company": "Globex", "status": " lost ", "signup_date": "7/21/2023", "col_x": ""}], "clean": [{"name": "Grace Smith", "rate": 0.705, "amount": 6767.08, "department": "Marketing", "email": "inpcjgn@mail.com", "company": "Vehement", "status": "Won", "signup_date": "2023-02-07"}, {"name": "Judy Cruz", "rate": 0.637, "amount": 2448.62, "department": "Marketing", "email": "ehdob@example.com", "company": "Acme Inc", "status": "Won", "signup_date": "2023-05-21"}, {"name": "Priya Fischer", "rate": 0.027999999999999997, "amount": 1465.31, "department": "Product", "email": "mkdej@example.com", "company": "Hooli", "status": "Won", "signup_date": "2023-07-21"}, {"name": "Eve Fischer", "rate": 0.435, "amount": 4804.92, "department": "Marketing", "email": "bkfgh@test.org", "company": "Umbrella", "status": "Open", "signup_date": "2023-05-21"}, {"name": "Heidi Wong", "rate": 0.064, "amount": 6441.58, "department": "Engineering", "email": "eahdpl@test.org", "company": "Initech", "status": "Won", "signup_date": "2023-10-05"}, {"name": "Carol Fischer", "rate": 0.833, "amount": NaN, "department": "Sales", "email": "pihoi@example.com", "company": "Stark Industries", "status": "Open", "signup_date": "2023-07-10"}, {"name": "Ivan Park", "rate": 0.034, "amount": 2795.06, "department": "Product", "email": "pkbn@corp.io", "company": "Wonka Co", "status": "Lost", "signup_date": "2023-09-25"}, {"name": "Sara Moore", "rate": 0.698, "amount": 2314.77, "department": "Engineering", "email": "dldgd@mail.com", "company": "Acme Inc", "status": "In Progress", "signup_date": "2023-11-13"}, {"name": "Ivan Moore", "rate": 0.135, "amount": 5752.24, "department": "Marketing", "email": "anfioh@example.com", "company": "Globex", "status": "Open", "signup_date": "2023-07-13"}, {"name": "Judy Ali", "rate": 0.298, "amount": 4703.06, "department": "Marketing", "email": "icoempf@example.com", "company": "Wonka Co", "status": "Open", "signup_date": "2023-01-07"}, {"name": "Lena Adams", "rate": 0.508, "amount": 5650.19, "department": "Sales", "email": "dhfjpn@corp.io", "company": "Globex", "status": "Open", "signup_date": "2023-04-07"}, {"name": "Lena Brandt", "rate": 0.659, "amount": 5315.45, "department": "Sales", "email": "phjkpa@test.org", "company": "Initech", "status": "Won", "signup_date": "2023-04-16"}, {"name": "Heidi Johnson", "rate": 0.41200000000000003, "amount": 8996.3, "department": "Product", "email": "gfpi@mail.com", "company": "Hooli", "status": "In Progress", "signup_date": "2023-04-23"}, {"name": "Ivan Johnson", "rate": 0.927, "amount": 2866.83, "department": "Sales", "email": "pikllf@mail.com", "company": "Acme Inc", "status": "Open", "signup_date": "2023-04-15"}, {"name": "Ivan Cruz", "rate": 0.27399999999999997, "amount": 7134.06, "department": "Engineering", "email": "jooki@test.org", "company": "Acme Inc", "status": "Open", "signup_date": "2023-12-06"}, {"name": "David Reyes", "rate": 0.083, "amount": 7994.23, "department": "Marketing", "email": "gfhadh@corp.io", "company": "Initech", "status": "In Progress", "signup_date": "2023-03-15"}, {"name": "Priya Moore", "rate": 0.196, "amount": 1085.1, "department": "Marketing", "email": "hbhkjf@example.com", "company": "Umbrella", "status": "Open", "signup_date": "2023-05-10"}, {"name": "Judy Reyes", "rate": 0.483, "amount": 8452.1, "department": "Product", "email": "pbnhopf@corp.io", "company": "Initech", "status": "Open", "signup_date": "2023-03-05"}, {"name": "Omar Novak", "rate": 0.516, "amount": 6208.72, "department": "Marketing", "email": "cpbpg@example.com", "company": "Vehement", "status": "Open", "signup_date": "2023-02-09"}, {"name": "Sara Smith", "rate": 0.34, "amount": 1809.85, "department": "Marketing", "email": "fdofdi@example.com", "company": "Vehement", "status": "Won", "signup_date": "2023-08-23"}, {"name": "Frank Johnson", "rate": 0.887, "amount": 5126.78, "department": "Product", "email": "ficbp@mail.com", "company": "Umbrella", "status": "Won", "signup_date": "2023-03-25"}, {"name": "Frank Moore", "rate": 0.16399999999999998, "amount": 3145.28, "department": "Marketing", "email": "dcecc@mail.com", "company": "Globex", "status": "Open", "signup_date": "2023-07-24"}, {"name": "Grace Novak", "rate": 0.907, "amount": 6120.68, "department": "Product", "email": "mkdgj@mail.com", "company": "Stark Industries", "status": "In Progress", "signup_date": "2023-12-05"}, {"name": "Heidi Brandt", "rate": 0.518, "amount": 4764.72, "department": "Marketing", "email": "nadfmg@example.com", "company": "Stark Industries", "status": "Lost", "signup_date": "2023-08-24"}, {"name": "Grace Moore", "rate": 0.884, "amount": 4739.17, "department": "Marketing", "email": "anhid@corp.io", "company": "Cyberdyne", "status": "Lost", "signup_date": "2023-01-22"}, {"name": "Grace Cruz", "rate": 0.961, "amount": 217.3, "department": "Sales", "email": "jnnn@mail.com", "company": "Initech", "status": "In Progress", "signup_date": "2023-02-02"}, {"name": "Frank Ali", "rate": 0.963, "amount": 4492.21, "department": "Sales", "email": "gceik@corp.io", "company": "Wonka Co", "status": "In Progress", "signup_date": "2023-01-19"}, {"name": "Bob Novak", "rate": 0.078, "amount": 5382.39, "department": "Engineering", "email": "gcknalf@corp.io", "company": "Stark Industries", "status": "Lost", "signup_date": "2023-12-13"}, {"name": "Sara Khan", "rate": 0.016, "amount": 4795.17, "department": "Marketing", "email": "njkeag@mail.com", "company": "Vehement", "status": "In Progress", "signup_date": "2023-12-05"}, {"name": "Carol Park", "rate": 0.203, "amount": 1659.69, "department": "Marketing", "email": "oogfee@test.org", "company": "Hooli", "status": "Won", "signup_date": "2023-02-12"}, {"name": "Lena Lee", "rate": 0.355, "amount": 3932.19, "department": "Product", "email": "pdencpk@corp.io", "company": "Cyberdyne", "status": "In Progress", "signup_date": "2023-11-26"}, {"name": "Lena Fischer", "rate": 0.728, "amount": 7862.61, "department": "Engineering", "email": "cape@test.org", "company": "Soylent Corp", "status": "Won", "signup_date": "2023-10-10"}, {"name": "Carol Smith", "rate": 0.5529999999999999, "amount": 7547.44, "department": "Engineering", "email": "apglb@example.com", "company": "Globex", "status": "In Progress", "signup_date": "2023-09-10"}, {"name": "Alice Brandt", "rate": 0.593, "amount": 332.75, "department": "Marketing", "email": "enfkcfm@mail.com", "company": "Acme Inc", "status": "In Progress", "signup_date": "2023-07-25"}, {"name": "Judy Fischer", "rate": 0.545, "amount": 6452.82, "department": "Product", "email": "dcdldln@mail.com", "company": "Hooli", "status": "Won", "signup_date": "2023-04-17"}, {"name": "Heidi Cruz", "rate": 0.075, "amount": 6211.03, "department": "Marketing", "email": "gfch@example.com", "company": "Soylent Corp", "status": "Open", "signup_date": "2023-09-24"}, {"name": "David Khan", "rate": 0.9329999999999999, "amount": 5432.5, "department": "Product", "email": "kpbn@mail.com", "company": "Acme Inc", "status": "Won", "signup_date": "2023-05-10"}, {"name": "Frank Khan", "rate": 0.043, "amount": NaN, "department": "Engineering", "email": "bedpjnh@mail.com", "company": "Wonka Co", "status": "Won", "signup_date": "2023-04-17"}, {"name": "Frank Johnson", "rate": 0.6459999999999999, "amount": 8439.39, "department": "Engineering", "email": "bgod@mail.com", "company": "Cyberdyne", "status": "Won", "signup_date": "2023-11-05"}, {"name": "Sara Cruz", "rate": 0.599, "amount": 7448.08, "department": "Engineering", "email": "dnjdba@example.com", "company": "Soylent Corp", "status": "Open", "signup_date": "2023-09-20"}, {"name": "Lena Moore", "rate": 0.177, "amount": 301.22, "department": "Sales", "email": "mlkk@example.com", "company": "Stark Industries", "status": "Open", "signup_date": "2023-12-26"}, {"name": "Priya Wong", "rate": 0.89, "amount": 5048.46, "department": "Sales", "email": "ifbj@test.org", "company": "Soylent Corp", "status": "In Progress", "signup_date": "2023-08-14"}, {"name": "Omar Petrov", "rate": 0.9670000000000001, "amount": 416.99, "department": "Marketing", "email": "gmhmnfa@test.org", "company": "Acme Inc", "status": "Won", "signup_date": "2023-05-10"}, {"name": "Judy Brandt", "rate": 0.435, "amount": 8445.96, "department": "Product", "email": "nnoh@test.org", "company": "Vehement", "status": "Open", "signup_date": "2023-05-09"}, {"name": "Mona Park", "rate": 0.128, "amount": 1308.43, "department": "Marketing", "email": "hcapfk@corp.io", "company": "Stark Industries", "status": "Won", "signup_date": "2023-11-10"}, {"name": "Karl Wong", "rate": 0.833, "amount": 719.51, "department": "Product", "email": "edoca@test.org", "company": "Vehement", "status": "Open", "signup_date": "2023-01-21"}, {"name": "Eve Park", "rate": 0.38299999999999995, "amount": 2540.66, "department": "Marketing", "email": "jaleknf@mail.com", "company": "Acme Inc", "status": "Lost", "signup_date": "2023-04-06"}, {"name": "Bob Cruz", "rate": 0.424, "amount": 1351.48, "department": "Product", "email": "aidgoif@corp.io", "company": "Umbrella", "status": "Lost", "signup_date": "2023-12-25"}, {"name": "Frank Adams", "rate": 0.9209999999999999, "amount": 358.67, "department": "Engineering", "email": "pgpk@test.org", "company": "Cyberdyne", "status": "Open", "signup_date": "2023-11-19"}, {"name": "Sara Khan", "rate": 0.23800000000000002, "amount": 6211.15, "department": "Product", "email": "nlpdl@test.org", "company": "Umbrella", "status": "Lost", "signup_date": "2023-10-25"}, {"name": "Omar Wong", "rate": 0.052000000000000005, "amount": 4878.53, "department": "Marketing", "email": "fjdg@test.org", "company": "Globex", "status": "In Progress", "signup_date": "2023-08-27"}, {"name": "Mona Adams", "rate": 0.539, "amount": 8265.55, "department": "Sales", "email": "lhjhg@corp.io", "company": "Globex", "status": "Lost", "signup_date": "2023-07-21"}, {"name": "Grace Reyes", "rate": 0.06, "amount": 7847.04, "department": "Sales", "email": "kcanja@example.com", "company": "Vehement", "status": "Won", "signup_date": "2023-08-24"}], "dirty_cols": ["name", "rate", "amount", "department", "email", "company", "status", "signup_date", "col_x"], "clean_cols": ["name", "rate", "amount", "department", "email", "company", "status", "signup_date"], "plan": {"dataset_summary": "54 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"growth": "Marketing", "Prod": "Product", "Growth": "Marketing", "ENGINEERING": "Engineering", "PM": "Product", "r&d": "Engineering", "Mktg": "Marketing", "mktg": "Marketing", "Salse": "Sales", "SALES": "Sales", "Dev": "Engineering", "mkt": "Marketing", "Mkt": "Marketing", "Growjh": "Marketing", "prod": "Product", "marketing": "Marketing", "MTKG": "Marketing", "biz dv": "Sales", "eng": "Engineering", "Markeing": "Marketing", "R&D": "Engineering", "engineering": "Engineering", "PROD": "Product", "Biz Dev": "Sales", "Saales": "Sales", "product": "Product", "sales": "Sales"}, "rationale": "Unified 27 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Closed Won": "Won", "closed won": "Won", "won": "Won", "new": "Open", "closed-won": "Won", "LOST": "Lost", "in prrogress": "In Progress", "OPEN": "Open", "open": "Open", "In Piogress": "In Progress", "Oppen": "Open", "New": "Open", "WON": "Won", "ongoing": "In Progress", "lost": "Lost", "in-progress": "In Progress", "losst": "Lost", "ongoig": "In Progress", "WIP": "In Progress", "ongning": "In Progress", "Closed-Won": "Won", "Oxen": "Open", "closed lost": "Lost", "in progress": "In Progress"}, "rationale": "Unified 24 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"department": "human  resources", "rate": "41.0%", "job_title": "   Dir.", "currency": "bov", "name": "Judy Diaz", "col_x": ""}, {"department": "  Fiannce", "rate": "4.1%", "job_title": "CHIEF TECHNOLOGY OFFICER  ", "currency": "  AWG", "name": "Karl Novak", "col_x": ""}, {"department": "Mkvg   ", "rate": "43.5%", "job_title": "Dri.   ", "currency": "bov", "name": "Ivan Park", "col_x": ""}, {"department": "HR", "rate": "91.7%", "job_title": "C.T.O.", "currency": "Bov", "name": " Priya Ali", "col_x": ""}, {"department": "mktg", "rate": "31.1%", "job_title": "PM", "currency": "mvdol ", "name": "Bob Adams   ", "col_x": ""}, {"department": "Peoplee Ops", "rate": "63.5%", "job_title": "Dir", "currency": "aruban florin", "name": "Sara Park ", "col_x": ""}, {"department": "fin", "rate": "36.7%", "job_title": "Dir   ", "currency": "ARUBAN FLORIN", "name": "Judy Park", "col_x": ""}, {"department": "People", "rate": "29.8%", "job_title": "Prod Manager", "currency": "Mvdol", "name": "Heidi Petrov", "col_x": ""}, {"department": "  Marketing", "rate": "7.9%", "job_title": "Product Manager", "currency": "Aruban Florin", "name": "Sara Wong", "col_x": ""}, {"department": " Mkt ", "rate": "2.9%", "job_title": "acct exec", "currency": "Mvdol   ", "name": "Frank  Johnson", "col_x": ""}, {"department": " Gorwth ", "rate": "24.1%", "job_title": "CHIEF TECHNOPOGY OFFICER", "currency": "Awg", "name": "Heidi Adams", "col_x": ""}, {"department": "peopleops", "rate": "30.1%", "job_title": "prod manager", "currency": "   aruban florin", "name": "David Reyes", "col_x": ""}, {"department": "people  ops", "rate": "7.8%", "job_title": "PRODUCT MANAGER", "currency": "ARUBAN FLORIN", "name": "Bob Brandt", "col_x": ""}, {"department": "growth", "rate": "57.2%", "job_title": "CTO", "currency": "Mvdol", "name": "Heidi  Diaz", "col_x": ""}, {"department": "People Ops", "rate": "0.1%", "job_title": "Director", "currency": "Aruban Florin   ", "name": "Grace Novak", "col_x": ""}, {"department": "Marketing", "rate": "34.7%", "job_title": "  Prod Manager", "currency": "Aruban Florin", "name": "Judy Adams", "col_x": ""}, {"department": "FINANCE", "rate": "91.6%", "job_title": " PM", "currency": "BOV  ", "name": " Eve Novak", "col_x": ""}, {"department": "People Ops", "rate": "19.0%", "job_title": " ae ", "currency": "   aruban florin", "name": "David Khan", "col_x": ""}, {"department": "mkt", "rate": "11.3%", "job_title": "PM", "currency": "  ARUBAN FLORIN", "name": "Carol Lee", "col_x": ""}, {"department": " growth ", "rate": "42.0%", "job_title": "AE", "currency": " bov ", "name": "David Lee", "col_x": ""}, {"department": " Mkt ", "rate": "99.9%", "job_title": "Acct  Exec", "currency": "MVDOL", "name": "   Sara Cruz", "col_x": ""}, {"department": "HUMAN RESOURCES", "rate": "22.2%", "job_title": "c.t.o.", "currency": "Bov", "name": "Frank Smith", "col_x": ""}, {"department": "Maarketing", "rate": "46.3%", "job_title": "Acct Exec", "currency": " Mvdol ", "name": "Mona Brandt", "col_x": ""}, {"department": "HR", "rate": "17.9%", "job_title": "c.t.o.", "currency": " bov", "name": "David Ali", "col_x": ""}, {"department": "PEOPLE OPS", "rate": "76.4%", "job_title": "Prod  Mgr", "currency": "BOV", "name": "  Frank Brandt", "col_x": ""}, {"department": " finance ", "rate": "72.7%", "job_title": "Dir  ", "currency": "awg", "name": "Priya Ali", "col_x": ""}, {"department": "HR", "rate": "98.2%", "job_title": "Pm", "currency": "aruban florin", "name": "Alice  Adams", "col_x": ""}, {"department": "Marketing", "rate": "24.5%", "job_title": "Di.", "currency": "Aruban Florin   ", "name": "Mona Park", "col_x": ""}, {"department": "Marketing", "rate": "39.5%", "job_title": "Account Executive", "currency": " BOV", "name": "Karl  Diaz", "col_x": ""}, {"department": "human resources", "rate": "31.8%", "job_title": " Director ", "currency": "ARUBAN FLORIN", "name": "Heidi Park", "col_x": ""}, {"department": "Accounting", "rate": "0.3%", "job_title": "  dir.", "currency": "  AWG", "name": "Priya Park", "col_x": ""}, {"department": "Fin", "rate": "95.7%", "job_title": "Driector", "currency": "  Mvdol", "name": "Grace Johnson   ", "col_x": ""}, {"department": "Fin   ", "rate": "99.3%", "job_title": "chief technology officer", "currency": "Aruban Florin", "name": "Heidi Smith   ", "col_x": ""}, {"department": " FIN ", "rate": "35.0%", "job_title": "Paod Mgr", "currency": "ARUBAN  FLORIN", "name": "Karl Wong", "col_x": ""}, {"department": "financce", "rate": "86.4%", "job_title": "Dir.", "currency": "Mvdol", "name": "Lena  Ali", "col_x": ""}, {"department": "Finace", "rate": "49.4%", "job_title": "Prod Mgr", "currency": "BOV", "name": "Eve Cruz", "col_x": ""}, {"department": "marketing", "rate": "20.5%", "job_title": "PM", "currency": "AWG", "name": "  Heidi Park", "col_x": ""}, {"department": "finance", "rate": "76.6%", "job_title": "Director", "currency": "BOV", "name": "  Bob Novak", "col_x": ""}, {"department": "people", "rate": "4.8%", "job_title": "   PROD MGR", "currency": "AWG", "name": "   Ivan Petrov", "col_x": ""}, {"department": "People", "rate": "72.5%", "job_title": "PM", "currency": "BOV", "name": "Grace Adams", "col_x": ""}, {"department": "fin", "rate": "49.5%", "job_title": "Account  Executive", "currency": "aruban florin", "name": "Judy Johnson   ", "col_x": ""}, {"department": "human resources  ", "rate": "3.9%", "job_title": "CTO", "currency": "  AWG", "name": "Omar Diaz", "col_x": ""}, {"department": "MKT", "rate": "54.7%", "job_title": "  CHIEF TECHNOLOGY OFFICER", "currency": "Aruban Florin", "name": "Priya Johnson   ", "col_x": ""}, {"department": "fin", "rate": "93.4%", "job_title": "product manager", "currency": "BOV", "name": "Alice Cruz  ", "col_x": ""}, {"department": "Fin", "rate": "76.7%", "job_title": "Prod Mgr", "currency": "BOV  ", "name": "Karl  Moore", "col_x": ""}, {"department": "Growth", "rate": "89.2%", "job_title": "Dir.", "currency": "Aruban  Florin", "name": "Lena Reyes", "col_x": ""}, {"department": "mkt", "rate": "65.5%", "job_title": " CTO ", "currency": "Mvdol", "name": "Mona Adams", "col_x": ""}, {"department": "finance", "rate": "76.6%", "job_title": "Director", "currency": "BOV", "name": "  Bob Novak", "col_x": ""}, {"department": "PEOPLE OPS", "rate": "76.4%", "job_title": "Prod  Mgr", "currency": "BOV", "name": "  Frank Brandt", "col_x": ""}], "clean": [{"department": "Human Resources", "rate": 0.41, "job_title": "Director", "currency": "BOV", "name": "Judy Diaz"}, {"department": "Finance", "rate": 0.040999999999999995, "job_title": "Chief Technology Officer", "currency": "AWG", "name": "Karl Novak"}, {"department": "Marketing", "rate": 0.435, "job_title": "Director", "currency": "BOV", "name": "Ivan Park"}, {"department": "Human Resources", "rate": 0.917, "job_title": "Chief Technology Officer", "currency": "BOV", "name": "Priya Ali"}, {"department": "Marketing", "rate": 0.311, "job_title": "Product Manager", "currency": "BOV", "name": "Bob Adams"}, {"department": "Human Resources", "rate": 0.635, "job_title": "Director", "currency": "AWG", "name": "Sara Park"}, {"department": "Finance", "rate": 0.36700000000000005, "job_title": "Director", "currency": "AWG", "name": "Judy Park"}, {"department": "Human Resources", "rate": 0.298, "job_title": "Product Manager", "currency": "BOV", "name": "Heidi Petrov"}, {"department": "Marketing", "rate": 0.079, "job_title": "Product Manager", "currency": "AWG", "name": "Sara Wong"}, {"department": "Marketing", "rate": 0.028999999999999998, "job_title": "Account Executive", "currency": "BOV", "name": "Frank Johnson"}, {"department": "Marketing", "rate": 0.24100000000000002, "job_title": "Chief Technology Officer", "currency": "AWG", "name": "Heidi Adams"}, {"department": "Human Resources", "rate": 0.301, "job_title": "Product Manager", "currency": "AWG", "name": "David Reyes"}, {"department": "Human Resources", "rate": 0.078, "job_title": "Product Manager", "currency": "AWG", "name": "Bob Brandt"}, {"department": "Marketing", "rate": 0.5720000000000001, "job_title": "Chief Technology Officer", "currency": "BOV", "name": "Heidi Diaz"}, {"department": "Human Resources", "rate": 0.001, "job_title": "Director", "currency": "AWG", "name": "Grace Novak"}, {"department": "Marketing", "rate": 0.34700000000000003, "job_title": "Product Manager", "currency": "AWG", "name": "Judy Adams"}, {"department": "Finance", "rate": 0.9159999999999999, "job_title": "Product Manager", "currency": "BOV", "name": "Eve Novak"}, {"department": "Human Resources", "rate": 0.19, "job_title": "Account Executive", "currency": "AWG", "name": "David Khan"}, {"department": "Marketing", "rate": 0.113, "job_title": "Product Manager", "currency": "AWG", "name": "Carol Lee"}, {"department": "Marketing", "rate": 0.42, "job_title": "Account Executive", "currency": "BOV", "name": "David Lee"}, {"department": "Marketing", "rate": 0.9990000000000001, "job_title": "Account Executive", "currency": "BOV", "name": "Sara Cruz"}, {"department": "Human Resources", "rate": 0.222, "job_title": "Chief Technology Officer", "currency": "BOV", "name": "Frank Smith"}, {"department": "Marketing", "rate": 0.46299999999999997, "job_title": "Account Executive", "currency": "BOV", "name": "Mona Brandt"}, {"department": "Human Resources", "rate": 0.179, "job_title": "Chief Technology Officer", "currency": "BOV", "name": "David Ali"}, {"department": "Human Resources", "rate": 0.764, "job_title": "Product Manager", "currency": "BOV", "name": "Frank Brandt"}, {"department": "Finance", "rate": 0.727, "job_title": "Director", "currency": "AWG", "name": "Priya Ali"}, {"department": "Human Resources", "rate": 0.982, "job_title": "Product Manager", "currency": "AWG", "name": "Alice Adams"}, {"department": "Marketing", "rate": 0.245, "job_title": "Director", "currency": "AWG", "name": "Mona Park"}, {"department": "Marketing", "rate": 0.395, "job_title": "Account Executive", "currency": "BOV", "name": "Karl Diaz"}, {"department": "Human Resources", "rate": 0.318, "job_title": "Director", "currency": "AWG", "name": "Heidi Park"}, {"department": "Finance", "rate": 0.003, "job_title": "Director", "currency": "AWG", "name": "Priya Park"}, {"department": "Finance", "rate": 0.9570000000000001, "job_title": "Director", "currency": "BOV", "name": "Grace Johnson"}, {"department": "Finance", "rate": 0.993, "job_title": "Chief Technology Officer", "currency": "AWG", "name": "Heidi Smith"}, {"department": "Finance", "rate": 0.35, "job_title": "Product Manager", "currency": "AWG", "name": "Karl Wong"}, {"department": "Finance", "rate": 0.8640000000000001, "job_title": "Director", "currency": "BOV", "name": "Lena Ali"}, {"department": "Finance", "rate": 0.494, "job_title": "Product Manager", "currency": "BOV", "name": "Eve Cruz"}, {"department": "Marketing", "rate": 0.205, "job_title": "Product Manager", "currency": "AWG", "name": "Heidi Park"}, {"department": "Finance", "rate": 0.7659999999999999, "job_title": "Director", "currency": "BOV", "name": "Bob Novak"}, {"department": "Human Resources", "rate": 0.048, "job_title": "Product Manager", "currency": "AWG", "name": "Ivan Petrov"}, {"department": "Human Resources", "rate": 0.725, "job_title": "Product Manager", "currency": "BOV", "name": "Grace Adams"}, {"department": "Finance", "rate": 0.495, "job_title": "Account Executive", "currency": "AWG", "name": "Judy Johnson"}, {"department": "Human Resources", "rate": 0.039, "job_title": "Chief Technology Officer", "currency": "AWG", "name": "Omar Diaz"}, {"department": "Marketing", "rate": 0.547, "job_title": "Chief Technology Officer", "currency": "AWG", "name": "Priya Johnson"}, {"department": "Finance", "rate": 0.934, "job_title": "Product Manager", "currency": "BOV", "name": "Alice Cruz"}, {"department": "Finance", "rate": 0.767, "job_title": "Product Manager", "currency": "BOV", "name": "Karl Moore"}, {"department": "Marketing", "rate": 0.892, "job_title": "Director", "currency": "AWG", "name": "Lena Reyes"}, {"department": "Marketing", "rate": 0.655, "job_title": "Chief Technology Officer", "currency": "BOV", "name": "Mona Adams"}], "dirty_cols": ["department", "rate", "job_title", "currency", "name", "col_x"], "clean_cols": ["department", "rate", "job_title", "currency", "name"], "plan": {"dataset_summary": "49 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"human resources": "Human Resources", "Fiannce": "Finance", "Mkvg": "Marketing", "HR": "Human Resources", "mktg": "Marketing", "Peoplee Ops": "Human Resources", "fin": "Finance", "People": "Human Resources", "Mkt": "Marketing", "Gorwth": "Marketing", "peopleops": "Human Resources", "people ops": "Human Resources", "growth": "Marketing", "People Ops": "Human Resources", "FINANCE": "Finance", "mkt": "Marketing", "HUMAN RESOURCES": "Human Resources", "Maarketing": "Marketing", "PEOPLE OPS": "Human Resources", "finance": "Finance", "Accounting": "Finance", "Fin": "Finance", "FIN": "Finance", "financce": "Finance", "Finace": "Finance", "marketing": "Marketing", "people": "Human Resources", "MKT": "Marketing", "Growth": "Marketing"}, "rationale": "Unified 29 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dir.": "Director", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "Dri.": "Director", "C.T.O.": "Chief Technology Officer", "PM": "Product Manager", "Dir": "Director", "Prod Manager": "Product Manager", "acct exec": "Account Executive", "CHIEF TECHNOPOGY OFFICER": "Chief Technology Officer", "prod manager": "Product Manager", "PRODUCT MANAGER": "Product Manager", "CTO": "Chief Technology Officer", "ae": "Account Executive", "AE": "Account Executive", "Acct Exec": "Account Executive", "c.t.o.": "Chief Technology Officer", "Prod Mgr": "Product Manager", "Pm": "Product Manager", "Di.": "Director", "dir.": "Director", "Driector": "Director", "chief technology officer": "Chief Technology Officer", "Paod Mgr": "Product Manager", "PROD MGR": "Product Manager", "product manager": "Product Manager"}, "rationale": "Unified 25 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bov": "BOV", "Bov": "BOV", "mvdol": "BOV", "aruban florin": "AWG", "ARUBAN FLORIN": "AWG", "Mvdol": "BOV", "Aruban Florin": "AWG", "Awg": "AWG", "MVDOL": "BOV", "awg": "AWG"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"is_active": "T", "country": "BIH", "phone": "(542)099-3598", "currency": "guinean franc", "amount": "6.181,30"}, {"is_active": "true", "country": "BA", "phone": "5946801412", "currency": "czk", "amount": "1.954,42"}, {"is_active": "FALSE", "country": " BIH", "phone": "505.650.9610", "currency": "   Cech Koruna", "amount": "$7,019.50"}, {"is_active": "false", "country": "saint kitts and nevis", "phone": "528-735-0068", "currency": "Guiuean Franc", "amount": "4.965,59"}, {"is_active": "TRUE", "country": "BA", "phone": "5049524946", "currency": "BBD", "amount": "$6,533.82"}, {"is_active": "false", "country": "Saint Kitts and Nevis", "phone": "(565)376-6848", "currency": "CZK", "amount": "4.354,77"}, {"is_active": "N", "country": "kn ", "phone": "545-420-6940", "currency": "BBD", "amount": "2.391,77"}, {"is_active": "TRUE", "country": " BIH ", "phone": "5161165659", "currency": "czk", "amount": "$5,867.45"}, {"is_active": "false", "country": "  Bosnia and Herzegovina", "phone": "501.047.7999", "currency": "BBD", "amount": "N/A"}, {"is_active": "F", "country": "Republic Of Bosnia And Herzegovina", "phone": "(533)205-0224", "currency": "  Guinean Franc", "amount": "$3,335.61"}, {"is_active": "F", "country": "KN", "phone": "575.685.9725", "currency": "Guinean Franc", "amount": "$5,960.57"}, {"is_active": "1", "country": "Bosnia and Herzegoina", "phone": "573-625-7242", "currency": "czk", "amount": "$881.31"}], "clean": [{"is_active": true, "country": "Bosnia and Herzegovina", "phone": "(542) 099-3598", "currency": "GNF", "amount": 6181.3}, {"is_active": true, "country": "Bosnia and Herzegovina", "phone": "(594) 680-1412", "currency": "CZK", "amount": 1954.42}, {"is_active": false, "country": "Bosnia and Herzegovina", "phone": "(505) 650-9610", "currency": "CZK", "amount": 7019.5}, {"is_active": false, "country": "Saint Kitts and Nevis", "phone": "(528) 735-0068", "currency": "GNF", "amount": 4965.59}, {"is_active": true, "country": "Bosnia and Herzegovina", "phone": "(504) 952-4946", "currency": "BBD", "amount": 6533.82}, {"is_active": false, "country": "Saint Kitts and Nevis", "phone": "(565) 376-6848", "currency": "CZK", "amount": 4354.77}, {"is_active": false, "country": "Saint Kitts and Nevis", "phone": "(545) 420-6940", "currency": "BBD", "amount": 2391.77}, {"is_active": true, "country": "Bosnia and Herzegovina", "phone": "(516) 116-5659", "currency": "CZK", "amount": 5867.45}, {"is_active": false, "country": "Bosnia and Herzegovina", "phone": "(501) 047-7999", "currency": "BBD", "amount": NaN}, {"is_active": false, "country": "Bosnia and Herzegovina", "phone": "(533) 205-0224", "currency": "GNF", "amount": 3335.61}, {"is_active": false, "country": "Saint Kitts and Nevis", "phone": "(575) 685-9725", "currency": "GNF", "amount": 5960.57}, {"is_active": true, "country": "Bosnia and Herzegovina", "phone": "(573) 625-7242", "currency": "CZK", "amount": 881.31}], "dirty_cols": ["is_active", "country", "phone", "currency", "amount"], "clean_cols": ["is_active", "country", "phone", "currency", "amount"], "plan": {"dataset_summary": "12 rows × 5 columns. 5 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BIH": "Bosnia and Herzegovina", "BA": "Bosnia and Herzegovina", "saint kitts and nevis": "Saint Kitts and Nevis", "kn": "Saint Kitts and Nevis", "Republic Of Bosnia And Herzegovina": "Bosnia and Herzegovina", "KN": "Saint Kitts and Nevis", "Bosnia and Herzegoina": "Bosnia and Herzegovina"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"guinean franc": "GNF", "czk": "CZK", "Cech Koruna": "CZK", "Guiuean Franc": "GNF", "Guinean Franc": "GNF"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": []}}
{"dirty": [{"unit": "L  ", "rate": "82.8%", "status": "paused", "signup_date": "2023-06-21", "currency": "egp   ", "job_title": "V..P.", "phone": "(532)374-3594", "is_active": "1"}, {"unit": "liter", "rate": "47.4%", "status": "free  trial", "signup_date": "25 May 2023", "currency": "EGP", "job_title": "   Vice Pres", "phone": "5432156857", "is_active": "true"}, {"unit": " milliliter ", "rate": "94.9%", "status": "Paused", "signup_date": "2023-02-21", "currency": " dominican peso", "job_title": "CTO", "phone": "5989112085", "is_active": "TRUE"}, {"unit": "ml", "rate": "93.8%", "status": "on hold", "signup_date": "44962", "currency": "Barbados Dollar", "job_title": "VP", "phone": "(514)675-6238", "is_active": "Y"}, {"unit": "l", "rate": "97.6%", "status": "Paused", "signup_date": "2023-06-03", "currency": "djf", "job_title": "PM", "phone": "514-703-8389", "is_active": "TRUE"}, {"unit": "Km   ", "rate": "98.8%", "status": "trial", "signup_date": "1 Aug 2023", "currency": "BBD ", "job_title": "VP", "phone": "(552)405-1775", "is_active": "FALSE"}, {"unit": "liter", "rate": "56.2%", "status": "on hold", "signup_date": "45086", "currency": "Barbados Dollar", "job_title": "Product Manuger", "phone": "(567)827-6921", "is_active": "TRUE"}, {"unit": "kilometers", "rate": "82.5%", "status": "   paused", "signup_date": "5/7/2023", "currency": "DOP", "job_title": "Vice Pres", "phone": "572.815.4292", "is_active": "N"}, {"unit": "liters", "rate": "21.6%", "status": "  on hold", "signup_date": "27 Mar 2023", "currency": "djibouti franc", "job_title": "vp", "phone": "557-767-2152", "is_active": "No"}, {"unit": "kilometer   ", "rate": "57.9%", "status": "ACTIVE", "signup_date": "44973", "currency": "DOP   ", "job_title": "V.P.", "phone": "513-354-1702", "is_active": "Yes"}, {"unit": "Km", "rate": "33.6%", "status": "TRIAL", "signup_date": "11/6/2023", "currency": "barbdaos dollar   ", "job_title": "   Chief Technology Officer", "phone": "500-594-1918", "is_active": "FALSE"}, {"unit": "cetnimeter", "rate": "49.0%", "status": "TRIAL", "signup_date": "7/17/2023", "currency": "dominican peso", "job_title": "V.P.", "phone": "527-265-7379", "is_active": "true"}, {"unit": "Km", "rate": "15.2%", "status": " ACTIVE ", "signup_date": "45267", "currency": "bbd", "job_title": "CTO", "phone": "540.288.0587", "is_active": "FALSE"}, {"unit": "centimeter", "rate": "35.4%", "status": "TRIAL", "signup_date": "2023-05-24", "currency": "DJF", "job_title": "CTO", "phone": "508-958-7054", "is_active": "N"}, {"unit": "centimeters   ", "rate": "39.0%", "status": "Paused", "signup_date": "5/13/2023", "currency": "Djibouti Franc", "job_title": " VP ", "phone": "568-277-4074", "is_active": "1"}, {"unit": "ml", "rate": "89.2%", "status": "acive", "signup_date": "2023-12-25", "currency": "BBD", "job_title": "Chief Technology Officer ", "phone": "5222022082", "is_active": "FALSE"}, {"unit": "Liter", "rate": "89.7%", "status": "ACTOVE", "signup_date": "45242", "currency": " DJF", "job_title": "pm", "phone": "5245333103", "is_active": "Yes"}, {"unit": "Centimeters", "rate": "16.3%", "status": "Paused", "signup_date": "45279", "currency": "Dominican  Peso", "job_title": "product manager", "phone": "5530594745", "is_active": "1"}, {"unit": "Km", "rate": "76.3%", "status": "trial", "signup_date": "7/6/2023", "currency": "Djiouti  Franc", "job_title": "Product Manager", "phone": "(526)030-2291", "is_active": "F"}, {"unit": "cm", "rate": "40.9%", "status": "paused ", "signup_date": "19 Oct 2023", "currency": "Bbd", "job_title": "V.P.   ", "phone": "572-404-5930", "is_active": "TRUE"}, {"unit": "mL", "rate": "51.4%", "status": "Trial", "signup_date": "45030", "currency": "dop", "job_title": "VICE PRESIDENT", "phone": "(530)581-5393", "is_active": "1"}, {"unit": "liter", "rate": "4.7%", "status": "Active", "signup_date": "11/27/2023", "currency": "DJF", "job_title": "   PRODUCT MANAGER", "phone": "573.161.8908", "is_active": "0"}, {"unit": "Centimeter", "rate": "82.1%", "status": "active", "signup_date": "3/19/2023", "currency": "  dominican peso", "job_title": "Prod Manager", "phone": "553.051.1328", "is_active": "0"}, {"unit": "km", "rate": "3.4%", "status": "Active", "signup_date": "2023-06-22", "currency": "Egyptian Pound", "job_title": "Vice Pres", "phone": "527-839-9434", "is_active": "N"}, {"unit": "cm", "rate": "13.7%", "status": "paused", "signup_date": "3/15/2023", "currency": "DJF", "job_title": "Chief Techology Officer", "phone": "593-541-0850", "is_active": "Yes"}, {"unit": "kilometers", "rate": "89.5%", "status": "free trial", "signup_date": "5 Dec 2023", "currency": "DJF", "job_title": "Vice Pres", "phone": "(545)605-3615", "is_active": "No"}, {"unit": "ml", "rate": "5.7%", "status": "trial", "signup_date": "45140", "currency": "Egp  ", "job_title": "prod manager", "phone": "565.250.6896", "is_active": "FALSE"}, {"unit": "ml", "rate": "63.4%", "status": "PAUSED", "signup_date": "6 Aug 2023", "currency": "Barbados Dollar", "job_title": "Chief Technology Officer", "phone": "(521)993-5847", "is_active": "false"}, {"unit": "  centimeters", "rate": "68.0%", "status": "trial", "signup_date": "17 Nov 2023", "currency": "Barbados Dollar", "job_title": "Chief Technology Officer", "phone": "562-147-3219", "is_active": "false"}, {"unit": "centimeters", "rate": "5.2%", "status": "  Active", "signup_date": "2/11/2023", "currency": "Bbd", "job_title": "C.T.O.", "phone": "5229263740", "is_active": "FALSE"}, {"unit": "centimeter", "rate": "99.7%", "status": "Activve", "signup_date": "44979", "currency": "DOMINICAN PESO   ", "job_title": "pm", "phone": "571-467-7414", "is_active": "TRUE"}, {"unit": "centimeter", "rate": "81.3%", "status": " Active", "signup_date": "45127", "currency": "dominican peso  ", "job_title": " vp ", "phone": "501-583-7145", "is_active": "No"}, {"unit": " kilometer ", "rate": "96.9%", "status": "paused", "signup_date": "1/11/2023", "currency": "BBD", "job_title": "CTO", "phone": "(598)161-9820", "is_active": "Y"}, {"unit": "liter", "rate": "99.4%", "status": "active", "signup_date": "2023-12-26", "currency": "djf ", "job_title": "cto", "phone": "574-073-7023", "is_active": "TRUE"}, {"unit": "  km", "rate": "85.1%", "status": "ACTIVE", "signup_date": "9/26/2023", "currency": "   dominican peso", "job_title": "Product Manager", "phone": "524.585.4289", "is_active": "false"}, {"unit": "L", "rate": "68.1%", "status": "ACTIVE", "signup_date": "2023-02-21", "currency": "DJF", "job_title": "chief technology officer", "phone": "544-310-4603", "is_active": "Yes"}, {"unit": "centimeters", "rate": "5.2%", "status": "  Active", "signup_date": "2/11/2023", "currency": "Bbd", "job_title": "C.T.O.", "phone": "5229263740", "is_active": "FALSE"}], "clean": [{"unit": "L", "rate": 0.828, "status": "Paused", "signup_date": "2023-06-21", "currency": "EGP", "job_title": "Vice President", "phone": "(532) 374-3594", "is_active": true}, {"unit": "L", "rate": 0.474, "status": "Trial", "signup_date": "2023-05-25", "currency": "EGP", "job_title": "Vice President", "phone": "(543) 215-6857", "is_active": true}, {"unit": "mL", "rate": 0.9490000000000001, "status": "Paused", "signup_date": "2023-02-21", "currency": "DOP", "job_title": "Chief Technology Officer", "phone": "(598) 911-2085", "is_active": true}, {"unit": "mL", "rate": 0.938, "status": "Paused", "signup_date": "2023-02-05", "currency": "BBD", "job_title": "Vice President", "phone": "(514) 675-6238", "is_active": true}, {"unit": "L", "rate": 0.976, "status": "Paused", "signup_date": "2023-06-03", "currency": "DJF", "job_title": "Product Manager", "phone": "(514) 703-8389", "is_active": true}, {"unit": "km", "rate": 0.988, "status": "Trial", "signup_date": "2023-08-01", "currency": "BBD", "job_title": "Vice President", "phone": "(552) 405-1775", "is_active": false}, {"unit": "L", "rate": 0.562, "status": "Paused", "signup_date": "2023-06-09", "currency": "BBD", "job_title": "Product Manager", "phone": "(567) 827-6921", "is_active": true}, {"unit": "km", "rate": 0.825, "status": "Paused", "signup_date": "2023-05-07", "currency": "DOP", "job_title": "Vice President", "phone": "(572) 815-4292", "is_active": false}, {"unit": "L", "rate": 0.21600000000000003, "status": "Paused", "signup_date": "2023-03-27", "currency": "DJF", "job_title": "Vice President", "phone": "(557) 767-2152", "is_active": false}, {"unit": "km", "rate": 0.579, "status": "Active", "signup_date": "2023-02-16", "currency": "DOP", "job_title": "Vice President", "phone": "(513) 354-1702", "is_active": true}, {"unit": "km", "rate": 0.336, "status": "Trial", "signup_date": "2023-11-06", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(500) 594-1918", "is_active": false}, {"unit": "cm", "rate": 0.49, "status": "Trial", "signup_date": "2023-07-17", "currency": "DOP", "job_title": "Vice President", "phone": "(527) 265-7379", "is_active": true}, {"unit": "km", "rate": 0.152, "status": "Active", "signup_date": "2023-12-07", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(540) 288-0587", "is_active": false}, {"unit": "cm", "rate": 0.354, "status": "Trial", "signup_date": "2023-05-24", "currency": "DJF", "job_title": "Chief Technology Officer", "phone": "(508) 958-7054", "is_active": false}, {"unit": "cm", "rate": 0.39, "status": "Paused", "signup_date": "2023-05-13", "currency": "DJF", "job_title": "Vice President", "phone": "(568) 277-4074", "is_active": true}, {"unit": "mL", "rate": 0.892, "status": "Active", "signup_date": "2023-12-25", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(522) 202-2082", "is_active": false}, {"unit": "L", "rate": 0.897, "status": "Active", "signup_date": "2023-11-12", "currency": "DJF", "job_title": "Product Manager", "phone": "(524) 533-3103", "is_active": true}, {"unit": "cm", "rate": 0.163, "status": "Paused", "signup_date": "2023-12-19", "currency": "DOP", "job_title": "Product Manager", "phone": "(553) 059-4745", "is_active": true}, {"unit": "km", "rate": 0.763, "status": "Trial", "signup_date": "2023-07-06", "currency": "DJF", "job_title": "Product Manager", "phone": "(526) 030-2291", "is_active": false}, {"unit": "cm", "rate": 0.409, "status": "Paused", "signup_date": "2023-10-19", "currency": "BBD", "job_title": "Vice President", "phone": "(572) 404-5930", "is_active": true}, {"unit": "mL", "rate": 0.514, "status": "Trial", "signup_date": "2023-04-14", "currency": "DOP", "job_title": "Vice President", "phone": "(530) 581-5393", "is_active": true}, {"unit": "L", "rate": 0.047, "status": "Active", "signup_date": "2023-11-27", "currency": "DJF", "job_title": "Product Manager", "phone": "(573) 161-8908", "is_active": false}, {"unit": "cm", "rate": 0.821, "status": "Active", "signup_date": "2023-03-19", "currency": "DOP", "job_title": "Product Manager", "phone": "(553) 051-1328", "is_active": false}, {"unit": "km", "rate": 0.034, "status": "Active", "signup_date": "2023-06-22", "currency": "EGP", "job_title": "Vice President", "phone": "(527) 839-9434", "is_active": false}, {"unit": "cm", "rate": 0.13699999999999998, "status": "Paused", "signup_date": "2023-03-15", "currency": "DJF", "job_title": "Chief Technology Officer", "phone": "(593) 541-0850", "is_active": true}, {"unit": "km", "rate": 0.895, "status": "Trial", "signup_date": "2023-12-05", "currency": "DJF", "job_title": "Vice President", "phone": "(545) 605-3615", "is_active": false}, {"unit": "mL", "rate": 0.057, "status": "Trial", "signup_date": "2023-08-02", "currency": "EGP", "job_title": "Product Manager", "phone": "(565) 250-6896", "is_active": false}, {"unit": "mL", "rate": 0.634, "status": "Paused", "signup_date": "2023-08-06", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(521) 993-5847", "is_active": false}, {"unit": "cm", "rate": 0.68, "status": "Trial", "signup_date": "2023-11-17", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(562) 147-3219", "is_active": false}, {"unit": "cm", "rate": 0.052000000000000005, "status": "Active", "signup_date": "2023-02-11", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(522) 926-3740", "is_active": false}, {"unit": "cm", "rate": 0.997, "status": "Active", "signup_date": "2023-02-22", "currency": "DOP", "job_title": "Product Manager", "phone": "(571) 467-7414", "is_active": true}, {"unit": "cm", "rate": 0.813, "status": "Active", "signup_date": "2023-07-20", "currency": "DOP", "job_title": "Vice President", "phone": "(501) 583-7145", "is_active": false}, {"unit": "km", "rate": 0.9690000000000001, "status": "Paused", "signup_date": "2023-01-11", "currency": "BBD", "job_title": "Chief Technology Officer", "phone": "(598) 161-9820", "is_active": true}, {"unit": "L", "rate": 0.9940000000000001, "status": "Active", "signup_date": "2023-12-26", "currency": "DJF", "job_title": "Chief Technology Officer", "phone": "(574) 073-7023", "is_active": true}, {"unit": "km", "rate": 0.851, "status": "Active", "signup_date": "2023-09-26", "currency": "DOP", "job_title": "Product Manager", "phone": "(524) 585-4289", "is_active": false}, {"unit": "L", "rate": 0.6809999999999999, "status": "Active", "signup_date": "2023-02-21", "currency": "DJF", "job_title": "Chief Technology Officer", "phone": "(544) 310-4603", "is_active": true}], "dirty_cols": ["unit", "rate", "status", "signup_date", "currency", "job_title", "phone", "is_active"], "clean_cols": ["unit", "rate", "status", "signup_date", "currency", "job_title", "phone", "is_active"], "plan": {"dataset_summary": "37 rows × 8 columns. 8 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liter": "L", "milliliter": "mL", "ml": "mL", "l": "L", "Km": "km", "kilometers": "km", "liters": "L", "kilometer": "km", "cetnimeter": "cm", "centimeter": "cm", "centimeters": "cm", "Liter": "L", "Centimeters": "cm", "Centimeter": "cm"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paused": "Paused", "free trial": "Trial", "on hold": "Paused", "trial": "Trial", "ACTIVE": "Active", "TRIAL": "Trial", "acive": "Active", "ACTOVE": "Active", "active": "Active", "PAUSED": "Paused", "Activve": "Active"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"egp": "EGP", "dominican peso": "DOP", "Barbados Dollar": "BBD", "djf": "DJF", "djibouti franc": "DJF", "barbdaos dollar": "BBD", "bbd": "BBD", "Djibouti Franc": "DJF", "Dominican Peso": "DOP", "Djiouti Franc": "DJF", "Bbd": "BBD", "dop": "DOP", "Egyptian Pound": "EGP", "Egp": "EGP", "DOMINICAN PESO": "DOP"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"V..P.": "Vice President", "Vice Pres": "Vice President", "CTO": "Chief Technology Officer", "VP": "Vice President", "PM": "Product Manager", "Product Manuger": "Product Manager", "vp": "Vice President", "V.P.": "Vice President", "pm": "Product Manager", "product manager": "Product Manager", "VICE PRESIDENT": "Vice President", "PRODUCT MANAGER": "Product Manager", "Prod Manager": "Product Manager", "Chief Techology Officer": "Chief Technology Officer", "prod manager": "Product Manager", "C.T.O.": "Chief Technology Officer", "cto": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"job_title": "C.E.O.", "department": "IT   ", "state": "GA   ", "currency": "GIP", "city": "Calasiao", "phone": "558.718.7346", "is_active": "Y", "name": "Ivan  Ali", "notes2": ""}, {"job_title": "Chief Technology Officer", "department": "cust support", "state": "ak", "currency": "   gip", "city": "Baarn", "phone": "508.383.7009", "is_active": "T", "name": "Alice Diaz", "notes2": ""}, {"job_title": "Chief Executive Officer", "department": "CS", "state": "ga", "currency": "KWANZA", "city": "managua", "phone": "509.667.4890", "is_active": "true", "name": "  Karl Adams", "notes2": ""}, {"job_title": "Eng Mrg", "department": "Support", "state": " ak ", "currency": "afn", "city": "los angeles", "phone": "5822686153", "is_active": "F", "name": "Carol  Novak", "notes2": ""}, {"job_title": "engineering  mgr", "department": " it ", "state": "Georgia", "currency": "Afghani", "city": "baarn", "phone": "501-081-8786", "is_active": "true", "name": "Judy  Novak", "notes2": ""}, {"job_title": "CEO", "department": " IT ", "state": "ALASKA   ", "currency": "Azerbaijan Manat", "city": "Baarn", "phone": "588.223.4175", "is_active": "1", "name": "Karl Fischer", "notes2": ""}, {"job_title": "C.E.O.", "department": "i.t.", "state": "alaska", "currency": " AZN", "city": "LA", "phone": "(569)952-6330", "is_active": "No", "name": "Mona Khan", "notes2": ""}, {"job_title": "C.E.O.", "department": "spport", "state": "AK", "currency": "  kwanza", "city": "Calasiao", "phone": "548-013-9361", "is_active": "F", "name": "Judy Lee", "notes2": ""}, {"job_title": "c.e.o.", "department": "Support", "state": "Alaska ", "currency": "AFN", "city": "hk", "phone": "528.119.0886", "is_active": "FALSE", "name": "Lena Brandt   ", "notes2": ""}, {"job_title": "eng manager", "department": "CS  ", "state": "Georgia", "currency": "Afghani", "city": "managua", "phone": "(517)355-4136", "is_active": "N", "name": "Heidi Adams", "notes2": ""}, {"job_title": "Engineering Mgr", "department": "informationtechnology", "state": "Georgia   ", "currency": "AZERBAIJAN MANAT", "city": "HK", "phone": "509-426-6353", "is_active": "No", "name": "Judy  Brandt", "notes2": ""}, {"job_title": "CEO", "department": "CUST SUPPORT", "state": "Aiaska", "currency": "Azerbaijan Manat", "city": "  MANAGUA", "phone": "566.349.5553", "is_active": "No", "name": "Grace Moore", "notes2": ""}, {"job_title": "CHIEF EXECUTIVE OFFICER", "department": "Cust Support", "state": " ga ", "currency": "Gibraltar Pound ", "city": "LA", "phone": "(535)471-3691", "is_active": "F", "name": "Bob Reyes", "notes2": ""}, {"job_title": "CEO", "department": "Support", "state": "georgia", "currency": "azerbaijan manat", "city": "Calasiao", "phone": "(525)330-4700", "is_active": "true", "name": "Omar Park", "notes2": ""}, {"job_title": "ENGINEERING MANAGER", "department": "I.T.", "state": "AK", "currency": "GIBRALTAR POUND", "city": "managua", "phone": "517-125-1494", "is_active": "Y", "name": "Lena Ali", "notes2": ""}, {"job_title": " ceo", "department": "   suppzrt", "state": "AK", "currency": "GIP", "city": "Baarn", "phone": "(584)611-3183", "is_active": "Y", "name": "Mona Park", "notes2": ""}, {"job_title": "C.T.O.", "department": "I.T.", "state": "Alaska", "currency": "Afghani", "city": " HK ", "phone": "582-320-5099", "is_active": "Yes", "name": "Mona Khan ", "notes2": ""}, {"job_title": "Chief Executive Officer", "department": "it", "state": "GA", "currency": "Gibraltar Pound", "city": " Calasiao ", "phone": "567-761-6645", "is_active": "Yes", "name": "Karl Smith", "notes2": ""}, {"job_title": "EZG MGR", "department": " SUUPPORT ", "state": "GA", "currency": "GIBRALTAR POUND   ", "city": "   baarn", "phone": "(528)409-9467", "is_active": "Y", "name": "Carol  Johnson", "notes2": ""}, {"job_title": "Chief Executive Officer", "department": "I.T.   ", "state": "Georgia", "currency": "AOA", "city": "Calasiao", "phone": "550.242.0590", "is_active": "N", "name": "Judy Cruz", "notes2": ""}, {"job_title": "eng mgr", "department": "CS", "state": "Ga  ", "currency": "AFN", "city": "hong kong", "phone": "585-229-9033", "is_active": "FALSE", "name": "Frank Fischer ", "notes2": ""}, {"job_title": " CEO ", "department": "custmer support", "state": "ak", "currency": "afhani", "city": "  Hong Kong", "phone": "587-512-3574", "is_active": "N", "name": "Alice Reyes ", "notes2": ""}, {"job_title": "C.E.O.  ", "department": "it", "state": "georgia", "currency": "Gip", "city": "Calasiao", "phone": "589.681.2467", "is_active": "T", "name": "Mona Cruz ", "notes2": ""}, {"job_title": "chief technology officer", "department": "Cust Support", "state": "   AK", "currency": "afn", "city": "HK", "phone": "541-587-7584", "is_active": "T", "name": "Grace Diaz", "notes2": ""}, {"job_title": "C.E.O.", "department": "Customer  Support", "state": " AK", "currency": "AZERBAIJAN MANAT", "city": "Los Angeles", "phone": "(589)130-7703", "is_active": "Y", "name": "David Smith", "notes2": ""}, {"job_title": "ceo", "department": "Customer Support  ", "state": "ak", "currency": "AZERBAIJAN MANAT", "city": "LOS ANGELES", "phone": "549.210.9729", "is_active": "Yes", "name": "Judy Johnson", "notes2": ""}, {"job_title": "eng mgr", "department": "Information Technology", "state": "Afaska", "currency": "Azerbaijan Manat", "city": "CALASIAO  ", "phone": "(580)781-8703", "is_active": "F", "name": "Lena Khan  ", "notes2": ""}, {"job_title": "eng mgr", "department": "I.K.", "state": "ak", "currency": "AFN", "city": "Calasiao", "phone": "5010677421", "is_active": "N", "name": "   Frank Novak", "notes2": ""}, {"job_title": "Chief Exceutive Officer", "department": "CS", "state": "GEORGIA", "currency": "Azerbaijan Manat", "city": "Managua", "phone": "5408264134", "is_active": "No", "name": "Heidi Fischer", "notes2": ""}, {"job_title": "engineering manager", "department": "CS", "state": "GA", "currency": "AZN   ", "city": "  L.A.", "phone": "(580)735-5752", "is_active": "T", "name": "Sara Novak   ", "notes2": ""}, {"job_title": "ENG  MGR", "department": "CS", "state": "Alaska", "currency": "AFGHANI", "city": "managua", "phone": "5929890648", "is_active": "FALSE", "name": "Heidi Ali", "notes2": ""}, {"job_title": "cto", "department": "CS", "state": "AK   ", "currency": " gip ", "city": "mnaagua", "phone": "544.384.8754", "is_active": "TRUE", "name": " Ivan Wong", "notes2": ""}, {"job_title": "engineering mgr", "department": "I.T.", "state": "alska", "currency": "gibraltar pound", "city": "MANAGUA", "phone": "514-593-4560", "is_active": "No", "name": "Ivan  Smith", "notes2": ""}, {"job_title": " CEO ", "department": "information  technology", "state": "Alaka", "currency": "AFN", "city": "HON GKONG", "phone": "528-812-0080", "is_active": "TRUE", "name": "David Johnson   ", "notes2": ""}, {"job_title": "ENG MANAGER", "department": "Information Technology  ", "state": " GA ", "currency": "Azn", "city": "LA", "phone": "525-069-8792", "is_active": "No", "name": "Eve  Moore", "notes2": ""}, {"job_title": "ChT.O.", "department": "Cust Support   ", "state": "GA", "currency": "Azn", "city": "los angeles", "phone": "591.115.1835", "is_active": "true", "name": "Carol  Petrov", "notes2": ""}, {"job_title": " CTO", "department": "CUSTOMER SUPPORT", "state": "AK", "currency": "AZN", "city": "MANAGUA", "phone": "5414910259", "is_active": "Yes", "name": "  Bob Reyes", "notes2": ""}, {"job_title": "CEO", "department": "Information Technology", "state": "Ga", "currency": "gibraltar pound   ", "city": "calasiao", "phone": "540.261.6271", "is_active": "true", "name": "Carol  Novak", "notes2": ""}, {"job_title": "CTO", "department": "   Support", "state": "Georgia", "currency": "GIP", "city": "Managua", "phone": "517.922.2355", "is_active": "FALSE", "name": "Mona Wong", "notes2": ""}, {"job_title": "ENG MANAGER", "department": "Customer Support  ", "state": "georgia", "currency": "AFN", "city": "managua", "phone": "5221920942", "is_active": "Yes", "name": "Carol Diaz", "notes2": ""}, {"job_title": "Enginering  Manager", "department": "i.t.", "state": "AK", "currency": "azn", "city": " HK ", "phone": "5352732584", "is_active": "N", "name": "David Adams", "notes2": ""}, {"job_title": "", "department": "", "state": "", "currency": "", "city": "", "phone": "", "is_active": "", "name": "", "notes2": ""}], "clean": [{"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Georgia", "currency": "GIP", "city": "Calasiao", "phone": "(558) 718-7346", "is_active": true, "name": "Ivan Ali"}, {"job_title": "Chief Technology Officer", "department": "Customer Support", "state": "Alaska", "currency": "GIP", "city": "Baarn", "phone": "(508) 383-7009", "is_active": true, "name": "Alice Diaz"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Georgia", "currency": "AOA", "city": "Managua", "phone": "(509) 667-4890", "is_active": true, "name": "Karl Adams"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Alaska", "currency": "AFN", "city": "Los Angeles", "phone": "(582) 268-6153", "is_active": false, "name": "Carol Novak"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Georgia", "currency": "AFN", "city": "Baarn", "phone": "(501) 081-8786", "is_active": true, "name": "Judy Novak"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Alaska", "currency": "AZN", "city": "Baarn", "phone": "(588) 223-4175", "is_active": true, "name": "Karl Fischer"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Alaska", "currency": "AZN", "city": "Los Angeles", "phone": "(569) 952-6330", "is_active": false, "name": "Mona Khan"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "AOA", "city": "Calasiao", "phone": "(548) 013-9361", "is_active": false, "name": "Judy Lee"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "AFN", "city": "Hong Kong", "phone": "(528) 119-0886", "is_active": false, "name": "Lena Brandt"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Georgia", "currency": "AFN", "city": "Managua", "phone": "(517) 355-4136", "is_active": false, "name": "Heidi Adams"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Georgia", "currency": "AZN", "city": "Hong Kong", "phone": "(509) 426-6353", "is_active": false, "name": "Judy Brandt"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "AZN", "city": "Managua", "phone": "(566) 349-5553", "is_active": false, "name": "Grace Moore"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Georgia", "currency": "GIP", "city": "Los Angeles", "phone": "(535) 471-3691", "is_active": false, "name": "Bob Reyes"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Georgia", "currency": "AZN", "city": "Calasiao", "phone": "(525) 330-4700", "is_active": true, "name": "Omar Park"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Alaska", "currency": "GIP", "city": "Managua", "phone": "(517) 125-1494", "is_active": true, "name": "Lena Ali"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "GIP", "city": "Baarn", "phone": "(584) 611-3183", "is_active": true, "name": "Mona Park"}, {"job_title": "Chief Technology Officer", "department": "Information Technology", "state": "Alaska", "currency": "AFN", "city": "Hong Kong", "phone": "(582) 320-5099", "is_active": true, "name": "Mona Khan"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Georgia", "currency": "GIP", "city": "Calasiao", "phone": "(567) 761-6645", "is_active": true, "name": "Karl Smith"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Georgia", "currency": "GIP", "city": "Baarn", "phone": "(528) 409-9467", "is_active": true, "name": "Carol Johnson"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Georgia", "currency": "AOA", "city": "Calasiao", "phone": "(550) 242-0590", "is_active": false, "name": "Judy Cruz"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Georgia", "currency": "AFN", "city": "Hong Kong", "phone": "(585) 229-9033", "is_active": false, "name": "Frank Fischer"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "AFN", "city": "Hong Kong", "phone": "(587) 512-3574", "is_active": false, "name": "Alice Reyes"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Georgia", "currency": "GIP", "city": "Calasiao", "phone": "(589) 681-2467", "is_active": true, "name": "Mona Cruz"}, {"job_title": "Chief Technology Officer", "department": "Customer Support", "state": "Alaska", "currency": "AFN", "city": "Hong Kong", "phone": "(541) 587-7584", "is_active": true, "name": "Grace Diaz"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "AZN", "city": "Los Angeles", "phone": "(589) 130-7703", "is_active": true, "name": "David Smith"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Alaska", "currency": "AZN", "city": "Los Angeles", "phone": "(549) 210-9729", "is_active": true, "name": "Judy Johnson"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Alaska", "currency": "AZN", "city": "Calasiao", "phone": "(580) 781-8703", "is_active": false, "name": "Lena Khan"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Alaska", "currency": "AFN", "city": "Calasiao", "phone": "(501) 067-7421", "is_active": false, "name": "Frank Novak"}, {"job_title": "Chief Executive Officer", "department": "Customer Support", "state": "Georgia", "currency": "AZN", "city": "Managua", "phone": "(540) 826-4134", "is_active": false, "name": "Heidi Fischer"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Georgia", "currency": "AZN", "city": "Los Angeles", "phone": "(580) 735-5752", "is_active": true, "name": "Sara Novak"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Alaska", "currency": "AFN", "city": "Managua", "phone": "(592) 989-0648", "is_active": false, "name": "Heidi Ali"}, {"job_title": "Chief Technology Officer", "department": "Customer Support", "state": "Alaska", "currency": "GIP", "city": "Managua", "phone": "(544) 384-8754", "is_active": true, "name": "Ivan Wong"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Alaska", "currency": "GIP", "city": "Managua", "phone": "(514) 593-4560", "is_active": false, "name": "Ivan Smith"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Alaska", "currency": "AFN", "city": "Hong Kong", "phone": "(528) 812-0080", "is_active": true, "name": "David Johnson"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Georgia", "currency": "AZN", "city": "Los Angeles", "phone": "(525) 069-8792", "is_active": false, "name": "Eve Moore"}, {"job_title": "Chief Technology Officer", "department": "Customer Support", "state": "Georgia", "currency": "AZN", "city": "Los Angeles", "phone": "(591) 115-1835", "is_active": true, "name": "Carol Petrov"}, {"job_title": "Chief Technology Officer", "department": "Customer Support", "state": "Alaska", "currency": "AZN", "city": "Managua", "phone": "(541) 491-0259", "is_active": true, "name": "Bob Reyes"}, {"job_title": "Chief Executive Officer", "department": "Information Technology", "state": "Georgia", "currency": "GIP", "city": "Calasiao", "phone": "(540) 261-6271", "is_active": true, "name": "Carol Novak"}, {"job_title": "Chief Technology Officer", "department": "Customer Support", "state": "Georgia", "currency": "GIP", "city": "Managua", "phone": "(517) 922-2355", "is_active": false, "name": "Mona Wong"}, {"job_title": "Engineering Manager", "department": "Customer Support", "state": "Georgia", "currency": "AFN", "city": "Managua", "phone": "(522) 192-0942", "is_active": true, "name": "Carol Diaz"}, {"job_title": "Engineering Manager", "department": "Information Technology", "state": "Alaska", "currency": "AZN", "city": "Hong Kong", "phone": "(535) 273-2584", "is_active": false, "name": "David Adams"}], "dirty_cols": ["job_title", "department", "state", "currency", "city", "phone", "is_active", "name", "notes2"], "clean_cols": ["job_title", "department", "state", "currency", "city", "phone", "is_active", "name"], "plan": {"dataset_summary": "42 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"C.E.O.": "Chief Executive Officer", "Eng Mrg": "Engineering Manager", "engineering mgr": "Engineering Manager", "CEO": "Chief Executive Officer", "c.e.o.": "Chief Executive Officer", "eng manager": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "ENGINEERING MANAGER": "Engineering Manager", "ceo": "Chief Executive Officer", "C.T.O.": "Chief Technology Officer", "EZG MGR": "Engineering Manager", "eng mgr": "Engineering Manager", "chief technology officer": "Chief Technology Officer", "Chief Exceutive Officer": "Chief Executive Officer", "engineering manager": "Engineering Manager", "ENG MGR": "Engineering Manager", "cto": "Chief Technology Officer", "ENG MANAGER": "Engineering Manager", "ChT.O.": "Chief Technology Officer", "CTO": "Chief Technology Officer", "Enginering Manager": "Engineering Manager"}, "rationale": "Unified 22 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Information Technology", "cust support": "Customer Support", "CS": "Customer Support", "Support": "Customer Support", "it": "Information Technology", "i.t.": "Information Technology", "spport": "Customer Support", "informationtechnology": "Information Technology", "CUST SUPPORT": "Customer Support", "Cust Support": "Customer Support", "I.T.": "Information Technology", "suppzrt": "Customer Support", "SUUPPORT": "Customer Support", "custmer support": "Customer Support", "I.K.": "Information Technology", "information technology": "Information Technology", "CUSTOMER SUPPORT": "Customer Support"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GA": "Georgia", "ak": "Alaska", "ga": "Georgia", "ALASKA": "Alaska", "alaska": "Alaska", "AK": "Alaska", "Aiaska": "Alaska", "georgia": "Georgia", "Ga": "Georgia", "Afaska": "Alaska", "GEORGIA": "Georgia", "alska": "Alaska", "Alaka": "Alaska"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"gip": "GIP", "KWANZA": "AOA", "afn": "AFN", "Afghani": "AFN", "Azerbaijan Manat": "AZN", "kwanza": "AOA", "AZERBAIJAN MANAT": "AZN", "Gibraltar Pound": "GIP", "azerbaijan manat": "AZN", "GIBRALTAR POUND": "GIP", "afhani": "AFN", "Gip": "GIP", "AFGHANI": "AFN", "gibraltar pound": "GIP", "Azn": "AZN", "azn": "AZN"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"managua": "Managua", "los angeles": "Los Angeles", "baarn": "Baarn", "LA": "Los Angeles", "hk": "Hong Kong", "HK": "Hong Kong", "MANAGUA": "Managua", "hong kong": "Hong Kong", "LOS ANGELES": "Los Angeles", "CALASIAO": "Calasiao", "L.A.": "Los Angeles", "mnaagua": "Managua", "HON GKONG": "Hong Kong", "calasiao": "Calasiao"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"amount": "$8,673.26", "status": "   in progress", "name": "Omar Novak", "department": "  Cust Support", "industry": "TECH", "rate": "10.8%", "company": " Globex", "city": "ChurchYonge Corridor"}, {"amount": "6.554,62", "status": "Closed Lost", "name": "Carol Lee", "department": "support", "industry": "Technoology  ", "rate": "86.5%", "company": " Acme Inc", "city": "Balod"}, {"amount": "3.232,92", "status": "open", "name": "Mona Ali", "department": "Customer Support", "industry": "software", "rate": "49.0%", "company": "Acme Inc", "city": "Shahrisabz"}, {"amount": "8.313,29", "status": "closed won", "name": "Omar Khan   ", "department": "Biz Dev ", "industry": "  Technology", "rate": "19.5%", "company": "Umbrella", "city": " stargard "}, {"amount": "None", "status": "ongoing", "name": "Mona Wong", "department": "Cust Support", "industry": "HOSPIITALITY", "rate": "14.1%", "company": "Stark Industries  ", "city": "CHURCH-YONGE CORRIDOR"}, {"amount": "3.387,26", "status": "NEW", "name": "Carol Brandt", "department": "Biw Dev", "industry": "Hospitality", "rate": "67.8%", "company": "Globex", "city": "   Balod"}, {"amount": "125,60", "status": "OPEN", "name": "   Frank Lee", "department": "Cust Support", "industry": "Tourism", "rate": "27.6%", "company": "Globex", "city": "Stargard  "}, {"amount": "None", "status": "Closed Lost", "name": "Mona Fischer", "department": "Biz  Dev", "industry": "Tourism", "rate": "70.3%", "company": "Stark Industries", "city": "SHAHRISABZ"}, {"amount": "$6,718.62", "status": "open", "name": "Lena Petrov", "department": "Biz Dev", "industry": "hospitality", "rate": "84.1%", "company": "Wonka  Co", "city": "Church-Yonge  Corridor"}, {"amount": "5.337,95", "status": "lost", "name": "Grace Reyes", "department": "Biz Dev", "industry": "Tourism", "rate": "34.6%", "company": "Hooli", "city": "Shahrisabz "}, {"amount": "6.690,85", "status": "Open   ", "name": "Alice  Moore", "department": "Sales", "industry": "IT", "rate": "44.3%", "company": "Stark  Industries", "city": "Balod"}, {"amount": "$4,254.36", "status": "open ", "name": " Sara Moore", "department": "Support ", "industry": " Tech ", "rate": "3.4%", "company": "Stark Industries", "city": "Church-Yonge  Corridor"}, {"amount": "7.562,10", "status": "lost", "name": "Omar Park", "department": "Sutport", "industry": "IT", "rate": "13.6%", "company": "Soylent  Corp", "city": " Shahrisabz "}, {"amount": "6.330,24", "status": "closedwon", "name": "Omar Cruz", "department": " Salps ", "industry": "It", "rate": "81.3%", "company": "Wonka  Co", "city": "churchyonge corridor"}, {"amount": "4.791,94", "status": " in progress", "name": "David Moore", "department": "Customer Support   ", "industry": "HOSPITALITY", "rate": "69.6%", "company": "Soylent Corp", "city": "STARGARD"}, {"amount": "6.238,35", "status": "Closed Lost", "name": "Lena Johnson", "department": "Sales", "industry": " tech", "rate": "0.9%", "company": "Vehement", "city": "stargard"}, {"amount": "N/A", "status": "new", "name": "David Novak", "department": "Customer Support", "industry": "TOURISM", "rate": "58.0%", "company": "Globex", "city": "Church-Yonge Corridor"}, {"amount": "$395.59", "status": "OPEN", "name": "   Ivan Cruz", "department": "CS", "industry": " toueism ", "rate": "29.0%", "company": "Initech  ", "city": "Stargard"}, {"amount": "--", "status": "lost", "name": "Frank Smith", "department": "CS   ", "industry": "   Tourism", "rate": "59.6%", "company": "Wonka Co", "city": "Balod"}, {"amount": "4.639,41", "status": "Closed Lost", "name": "  Frank Wong", "department": "sales ", "industry": "travel", "rate": "91.6%", "company": "Initech", "city": "Church-Yoge Corridor"}, {"amount": "5.205,05", "status": "closed-won", "name": "   Carol Petrov", "department": "CS", "industry": "Technology", "rate": "8.9%", "company": "Hooli", "city": "Church-Yonge Corridor "}, {"amount": "6.298,56", "status": "in progress   ", "name": "  Eve Wong", "department": "biz dev", "industry": "Hospitality", "rate": "59.9%", "company": "  Cyberdyne", "city": "Shahrisabz "}, {"amount": "$6,833.01", "status": "Lost", "name": " David Diaz", "department": "SUPPORT", "industry": "Tourism", "rate": "25.7%", "company": " Vehement ", "city": "Balod"}, {"amount": "TBD", "status": "WIP   ", "name": "Ivan Moore   ", "department": "CS", "industry": "hotels & travel   ", "rate": "17.6%", "company": "Acme Inc", "city": "  Stargard"}, {"amount": "260,57", "status": "Lost ", "name": "Bob Reyes", "department": "  BIZ DEV", "industry": "Tourism", "rate": "86.4%", "company": "Stark Industries", "city": " Blod "}, {"amount": "$2,430.37", "status": "WIP", "name": "Priya Khan ", "department": "Biz Dev", "industry": "Hospitality", "rate": "24.0%", "company": " Vehement ", "city": "Stargard"}, {"amount": "?", "status": "Closed Lost", "name": "Sara Moore   ", "department": " Biz Dev", "industry": "tourism", "rate": "29.3%", "company": "Soylent Corp", "city": "   Stargard"}, {"amount": "1.013,37", "status": "In Progress", "name": "Karl Adams", "department": "support", "industry": "hospitality   ", "rate": "22.8%", "company": "Acme Inc ", "city": "Stargadr"}, {"amount": "1.410,28", "status": "ONGOING", "name": "  David Smith", "department": "CS", "industry": "tech", "rate": "33.6%", "company": "Soylent Corp", "city": "Shahrisabz"}, {"amount": "$1,459.77", "status": "CLOSSED WON", "name": "Omar Fischer", "department": "Customer Support", "industry": "it", "rate": "78.0%", "company": "Acme Inc", "city": "SHAHRISABZ"}, {"amount": "$6,992.17", "status": "In Progress", "name": "Omar Petrov   ", "department": "Biz Dev", "industry": " IT ", "rate": "92.4%", "company": "Umbrella", "city": "church-yonge corridor"}, {"amount": "$5,482.47", "status": "CLOSED WON", "name": "Alice Smith", "department": "Customer Support", "industry": "hotels & travel", "rate": "99.7%", "company": "Wonka Co", "city": "shahrisabz "}, {"amount": "$6,263.29", "status": "WIP", "name": "Alice  Novak", "department": "BIZ DEV", "industry": "HOTELS & TRAVEL", "rate": "95.8%", "company": "Wonka Co", "city": " shahrisabz"}, {"amount": "1.549,64", "status": "open", "name": "Priya Diaz ", "department": "Sals", "industry": "Hotels & Travel", "rate": "21.3%", "company": "Acme Inc", "city": "balod"}, {"amount": "1.760,22", "status": "Closed Won", "name": "Omar Reyes", "department": " Sales ", "industry": "Hotels & Travel", "rate": "83.5%", "company": "Cyberdyne", "city": "Church-Yonge Corridor"}, {"amount": "1.521,25", "status": "In Progress", "name": "Bob  Lee", "department": "CUSTOMER SUPPORT", "industry": "Tech", "rate": "40.4%", "company": "  Vehement", "city": "stargard"}, {"amount": "$2,638.22", "status": "  InP-rogress", "name": "  Sara Reyes", "department": "Biz Dev", "industry": "technlogy", "rate": "31.9%", "company": "Hooli", "city": "Church-Yonge Corridor"}, {"amount": "N/A", "status": "Lost", "name": "Omar Park", "department": "Sales", "industry": "Tech", "rate": "60.4%", "company": "Acme Inc   ", "city": "Balod"}], "clean": [{"amount": 8673.26, "status": "In Progress", "name": "Omar Novak", "department": "Customer Support", "industry": "Technology", "rate": 0.10800000000000001, "company": "Globex", "city": "Church-Yonge Corridor"}, {"amount": 6554.62, "status": "Lost", "name": "Carol Lee", "department": "Customer Support", "industry": "Technology", "rate": 0.865, "company": "Acme Inc", "city": "Balod"}, {"amount": 3232.92, "status": "Open", "name": "Mona Ali", "department": "Customer Support", "industry": "Technology", "rate": 0.49, "company": "Acme Inc", "city": "Shahrisabz"}, {"amount": 8313.29, "status": "Won", "name": "Omar Khan", "department": "Sales", "industry": "Technology", "rate": 0.195, "company": "Umbrella", "city": "Stargard"}, {"amount": NaN, "status": "In Progress", "name": "Mona Wong", "department": "Customer Support", "industry": "Hospitality", "rate": 0.141, "company": "Stark Industries", "city": "Church-Yonge Corridor"}, {"amount": 3387.26, "status": "Open", "name": "Carol Brandt", "department": "Sales", "industry": "Hospitality", "rate": 0.6779999999999999, "company": "Globex", "city": "Balod"}, {"amount": 125.6, "status": "Open", "name": "Frank Lee", "department": "Customer Support", "industry": "Hospitality", "rate": 0.276, "company": "Globex", "city": "Stargard"}, {"amount": NaN, "status": "Lost", "name": "Mona Fischer", "department": "Sales", "industry": "Hospitality", "rate": 0.703, "company": "Stark Industries", "city": "Shahrisabz"}, {"amount": 6718.62, "status": "Open", "name": "Lena Petrov", "department": "Sales", "industry": "Hospitality", "rate": 0.841, "company": "Wonka Co", "city": "Church-Yonge Corridor"}, {"amount": 5337.95, "status": "Lost", "name": "Grace Reyes", "department": "Sales", "industry": "Hospitality", "rate": 0.34600000000000003, "company": "Hooli", "city": "Shahrisabz"}, {"amount": 6690.85, "status": "Open", "name": "Alice Moore", "department": "Sales", "industry": "Technology", "rate": 0.44299999999999995, "company": "Stark Industries", "city": "Balod"}, {"amount": 4254.36, "status": "Open", "name": "Sara Moore", "department": "Customer Support", "industry": "Technology", "rate": 0.034, "company": "Stark Industries", "city": "Church-Yonge Corridor"}, {"amount": 7562.1, "status": "Lost", "name": "Omar Park", "department": "Customer Support", "industry": "Technology", "rate": 0.136, "company": "Soylent Corp", "city": "Shahrisabz"}, {"amount": 6330.24, "status": "Won", "name": "Omar Cruz", "department": "Sales", "industry": "Technology", "rate": 0.813, "company": "Wonka Co", "city": "Church-Yonge Corridor"}, {"amount": 4791.94, "status": "In Progress", "name": "David Moore", "department": "Customer Support", "industry": "Hospitality", "rate": 0.696, "company": "Soylent Corp", "city": "Stargard"}, {"amount": 6238.35, "status": "Lost", "name": "Lena Johnson", "department": "Sales", "industry": "Technology", "rate": 0.009000000000000001, "company": "Vehement", "city": "Stargard"}, {"amount": NaN, "status": "Open", "name": "David Novak", "department": "Customer Support", "industry": "Hospitality", "rate": 0.58, "company": "Globex", "city": "Church-Yonge Corridor"}, {"amount": 395.59, "status": "Open", "name": "Ivan Cruz", "department": "Customer Support", "industry": "Hospitality", "rate": 0.29, "company": "Initech", "city": "Stargard"}, {"amount": NaN, "status": "Lost", "name": "Frank Smith", "department": "Customer Support", "industry": "Hospitality", "rate": 0.596, "company": "Wonka Co", "city": "Balod"}, {"amount": 4639.41, "status": "Lost", "name": "Frank Wong", "department": "Sales", "industry": "Hospitality", "rate": 0.9159999999999999, "company": "Initech", "city": "Church-Yonge Corridor"}, {"amount": 5205.05, "status": "Won", "name": "Carol Petrov", "department": "Customer Support", "industry": "Technology", "rate": 0.08900000000000001, "company": "Hooli", "city": "Church-Yonge Corridor"}, {"amount": 6298.56, "status": "In Progress", "name": "Eve Wong", "department": "Sales", "industry": "Hospitality", "rate": 0.599, "company": "Cyberdyne", "city": "Shahrisabz"}, {"amount": 6833.01, "status": "Lost", "name": "David Diaz", "department": "Customer Support", "industry": "Hospitality", "rate": 0.257, "company": "Vehement", "city": "Balod"}, {"amount": NaN, "status": "In Progress", "name": "Ivan Moore", "department": "Customer Support", "industry": "Hospitality", "rate": 0.17600000000000002, "company": "Acme Inc", "city": "Stargard"}, {"amount": 260.57, "status": "Lost", "name": "Bob Reyes", "department": "Sales", "industry": "Hospitality", "rate": 0.8640000000000001, "company": "Stark Industries", "city": "Balod"}, {"amount": 2430.37, "status": "In Progress", "name": "Priya Khan", "department": "Sales", "industry": "Hospitality", "rate": 0.24, "company": "Vehement", "city": "Stargard"}, {"amount": NaN, "status": "Lost", "name": "Sara Moore", "department": "Sales", "industry": "Hospitality", "rate": 0.293, "company": "Soylent Corp", "city": "Stargard"}, {"amount": 1013.37, "status": "In Progress", "name": "Karl Adams", "department": "Customer Support", "industry": "Hospitality", "rate": 0.228, "company": "Acme Inc", "city": "Stargard"}, {"amount": 1410.28, "status": "In Progress", "name": "David Smith", "department": "Customer Support", "industry": "Technology", "rate": 0.336, "company": "Soylent Corp", "city": "Shahrisabz"}, {"amount": 1459.77, "status": "Won", "name": "Omar Fischer", "department": "Customer Support", "industry": "Technology", "rate": 0.78, "company": "Acme Inc", "city": "Shahrisabz"}, {"amount": 6992.17, "status": "In Progress", "name": "Omar Petrov", "department": "Sales", "industry": "Technology", "rate": 0.924, "company": "Umbrella", "city": "Church-Yonge Corridor"}, {"amount": 5482.47, "status": "Won", "name": "Alice Smith", "department": "Customer Support", "industry": "Hospitality", "rate": 0.997, "company": "Wonka Co", "city": "Shahrisabz"}, {"amount": 6263.29, "status": "In Progress", "name": "Alice Novak", "department": "Sales", "industry": "Hospitality", "rate": 0.958, "company": "Wonka Co", "city": "Shahrisabz"}, {"amount": 1549.64, "status": "Open", "name": "Priya Diaz", "department": "Sales", "industry": "Hospitality", "rate": 0.213, "company": "Acme Inc", "city": "Balod"}, {"amount": 1760.22, "status": "Won", "name": "Omar Reyes", "department": "Sales", "industry": "Hospitality", "rate": 0.835, "company": "Cyberdyne", "city": "Church-Yonge Corridor"}, {"amount": 1521.25, "status": "In Progress", "name": "Bob Lee", "department": "Customer Support", "industry": "Technology", "rate": 0.40399999999999997, "company": "Vehement", "city": "Stargard"}, {"amount": 2638.22, "status": "In Progress", "name": "Sara Reyes", "department": "Sales", "industry": "Technology", "rate": 0.319, "company": "Hooli", "city": "Church-Yonge Corridor"}, {"amount": NaN, "status": "Lost", "name": "Omar Park", "department": "Sales", "industry": "Technology", "rate": 0.604, "company": "Acme Inc", "city": "Balod"}], "dirty_cols": ["amount", "status", "name", "department", "industry", "rate", "company", "city"], "clean_cols": ["amount", "status", "name", "department", "industry", "rate", "company", "city"], "plan": {"dataset_summary": "38 rows × 8 columns. 8 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"in progress": "In Progress", "Closed Lost": "Lost", "open": "Open", "closed won": "Won", "ongoing": "In Progress", "NEW": "Open", "OPEN": "Open", "lost": "Lost", "closedwon": "Won", "new": "Open", "closed-won": "Won", "WIP": "In Progress", "ONGOING": "In Progress", "CLOSSED WON": "Won", "CLOSED WON": "Won", "Closed Won": "Won", "InP-rogress": "In Progress"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cust Support": "Customer Support", "support": "Customer Support", "Biz Dev": "Sales", "Biw Dev": "Sales", "Support": "Customer Support", "Sutport": "Customer Support", "Salps": "Sales", "CS": "Customer Support", "sales": "Sales", "biz dev": "Sales", "SUPPORT": "Customer Support", "BIZ DEV": "Sales", "Sals": "Sales", "CUSTOMER SUPPORT": "Customer Support"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TECH": "Technology", "Technoology": "Technology", "software": "Technology", "HOSPIITALITY": "Hospitality", "Tourism": "Hospitality", "hospitality": "Hospitality", "IT": "Technology", "Tech": "Technology", "It": "Technology", "HOSPITALITY": "Hospitality", "tech": "Technology", "TOURISM": "Hospitality", "toueism": "Hospitality", "travel": "Hospitality", "hotels & travel": "Hospitality", "tourism": "Hospitality", "it": "Technology", "HOTELS & TRAVEL": "Hospitality", "Hotels & Travel": "Hospitality", "technlogy": "Technology"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ChurchYonge Corridor": "Church-Yonge Corridor", "stargard": "Stargard", "CHURCH-YONGE CORRIDOR": "Church-Yonge Corridor", "SHAHRISABZ": "Shahrisabz", "churchyonge corridor": "Church-Yonge Corridor", "STARGARD": "Stargard", "Church-Yoge Corridor": "Church-Yonge Corridor", "Blod": "Balod", "Stargadr": "Stargard", "church-yonge corridor": "Church-Yonge Corridor", "shahrisabz": "Shahrisabz", "balod": "Balod"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"industry": "   Utilities", "company": "Soylent Corp", "signup_date": "2023-09-11", "col_x": ""}, {"industry": "  telco", "company": " Stark Industries", "signup_date": "2023-07-16", "col_x": ""}, {"industry": "Utilities", "company": "Initech", "signup_date": "45243", "col_x": ""}, {"industry": "energy", "company": "Soylent Corp", "signup_date": "2023-12-09", "col_x": ""}, {"industry": "   Hospitality", "company": "Initech", "signup_date": "4 Feb 2023", "col_x": ""}, {"industry": "OIL & GAS", "company": "Cyberdyne", "signup_date": "2023-12-21", "col_x": ""}, {"industry": "  Hotels & Travel", "company": "  Soylent Corp", "signup_date": "4 Apr 2023", "col_x": ""}, {"industry": "Energy", "company": "  Acme Inc", "signup_date": "2023-09-26", "col_x": ""}, {"industry": " Tourism ", "company": " Cyberdyne ", "signup_date": "12/3/2023", "col_x": ""}, {"industry": "telecom", "company": "Hooli", "signup_date": "2023-02-08", "col_x": ""}, {"industry": "hosjitality", "company": "Soylent Corp", "signup_date": "9/13/2023", "col_x": ""}, {"industry": "  Energgy", "company": "Wonka Co", "signup_date": "2023-08-24", "col_x": ""}, {"industry": "oil &  gas", "company": "Soylent Corp", "signup_date": "45029", "col_x": ""}, {"industry": "Telecom  ", "company": " Initech ", "signup_date": "5/4/2023", "col_x": ""}, {"industry": "ENERGY", "company": "Acme  Inc", "signup_date": "2023-12-17", "col_x": ""}, {"industry": "Comms", "company": "Stark Industries", "signup_date": "1/20/2023", "col_x": ""}, {"industry": " Hotels & Travel", "company": "Wonka Co", "signup_date": "3/21/2023", "col_x": ""}, {"industry": "energy", "company": "   Globex", "signup_date": "2023-05-23", "col_x": ""}, {"industry": "hosjitality", "company": "Soylent Corp", "signup_date": "9/13/2023", "col_x": ""}, {"industry": "ENERGY", "company": "Acme  Inc", "signup_date": "2023-12-17", "col_x": ""}], "clean": [{"industry": "Energy", "company": "Soylent Corp", "signup_date": "2023-09-11"}, {"industry": "Telecommunications", "company": "Stark Industries", "signup_date": "2023-07-16"}, {"industry": "Energy", "company": "Initech", "signup_date": "2023-11-13"}, {"industry": "Energy", "company": "Soylent Corp", "signup_date": "2023-12-09"}, {"industry": "Hospitality", "company": "Initech", "signup_date": "2023-02-04"}, {"industry": "Energy", "company": "Cyberdyne", "signup_date": "2023-12-21"}, {"industry": "Hospitality", "company": "Soylent Corp", "signup_date": "2023-04-04"}, {"industry": "Energy", "company": "Acme Inc", "signup_date": "2023-09-26"}, {"industry": "Hospitality", "company": "Cyberdyne", "signup_date": "2023-12-03"}, {"industry": "Telecommunications", "company": "Hooli", "signup_date": "2023-02-08"}, {"industry": "Hospitality", "company": "Soylent Corp", "signup_date": "2023-09-13"}, {"industry": "Energy", "company": "Wonka Co", "signup_date": "2023-08-24"}, {"industry": "Energy", "company": "Soylent Corp", "signup_date": "2023-04-13"}, {"industry": "Telecommunications", "company": "Initech", "signup_date": "2023-05-04"}, {"industry": "Energy", "company": "Acme Inc", "signup_date": "2023-12-17"}, {"industry": "Telecommunications", "company": "Stark Industries", "signup_date": "2023-01-20"}, {"industry": "Hospitality", "company": "Wonka Co", "signup_date": "2023-03-21"}, {"industry": "Energy", "company": "Globex", "signup_date": "2023-05-23"}], "dirty_cols": ["industry", "company", "signup_date", "col_x"], "clean_cols": ["industry", "company", "signup_date"], "plan": {"dataset_summary": "20 rows × 4 columns. 3 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Utilities": "Energy", "telco": "Telecommunications", "energy": "Energy", "OIL & GAS": "Energy", "Hotels & Travel": "Hospitality", "Tourism": "Hospitality", "telecom": "Telecommunications", "hosjitality": "Hospitality", "Energgy": "Energy", "oil & gas": "Energy", "Telecom": "Telecommunications", "ENERGY": "Energy", "Comms": "Telecommunications"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"state": " NY ", "signup_date": "9 Feb 2023", "company": "Stark  Industries", "currency": "ARS", "col_x": ""}, {"state": " RI ", "signup_date": "2023-12-16", "company": "Globex   ", "currency": " LEK ", "col_x": ""}, {"state": "NY", "signup_date": "2023-08-16", "company": " Initech ", "currency": "Dominican Peso", "col_x": ""}, {"state": " ri ", "signup_date": "5/12/2023", "company": "Acme Inc", "currency": "DOP ", "col_x": ""}, {"state": "Rhode Island", "signup_date": "12/10/2023", "company": "Cyberdyne", "currency": "Dominican Peso", "col_x": ""}, {"state": "rhode islaand", "signup_date": "2023-08-04", "company": "Cyberdyne  ", "currency": "Lek", "col_x": ""}, {"state": "rhode island", "signup_date": "2023-07-08", "company": " Initech", "currency": "Dominican Peso", "col_x": ""}, {"state": " NEW YORK", "signup_date": "45246", "company": " Cyberdyne", "currency": "ARS", "col_x": ""}, {"state": "RI", "signup_date": "13 Oct 2023", "company": "Stark Industries", "currency": "Lek", "col_x": ""}, {"state": "ri", "signup_date": "25 Jun 2023", "company": "Soylent  Corp", "currency": "ARS", "col_x": ""}, {"state": "Rhode Island", "signup_date": "6 Aug 2023", "company": " Stark Industries", "currency": "Dominican Peeso", "col_x": ""}, {"state": "new york", "signup_date": "2/24/2023", "company": "Cyberdyne", "currency": "all", "col_x": ""}, {"state": "RI", "signup_date": "12 Sep 2023", "company": "   Stark Industries", "currency": "dop", "col_x": ""}, {"state": "ny", "signup_date": "2023-11-04", "company": "Cyberdyne", "currency": "DOP", "col_x": ""}, {"state": "NY", "signup_date": "10/11/2023", "company": "Acme  Inc", "currency": "Lek   ", "col_x": ""}, {"state": "Rhode Island", "signup_date": "2023-11-20", "company": "Umbrella ", "currency": "ALL", "col_x": ""}, {"state": "NY", "signup_date": "2023-01-09", "company": "Stark Industries  ", "currency": "ALL", "col_x": ""}, {"state": "   ri", "signup_date": "5 Jul 2023", "company": "Initech", "currency": "Lek", "col_x": ""}, {"state": " NY ", "signup_date": "9 Feb 2023", "company": "Stark  Industries", "currency": "ARS", "col_x": ""}, {"state": "", "signup_date": "", "company": "", "currency": "", "col_x": ""}], "clean": [{"state": "New York", "signup_date": "2023-02-09", "company": "Stark Industries", "currency": "ARS"}, {"state": "Rhode Island", "signup_date": "2023-12-16", "company": "Globex", "currency": "ALL"}, {"state": "New York", "signup_date": "2023-08-16", "company": "Initech", "currency": "DOP"}, {"state": "Rhode Island", "signup_date": "2023-05-12", "company": "Acme Inc", "currency": "DOP"}, {"state": "Rhode Island", "signup_date": "2023-12-10", "company": "Cyberdyne", "currency": "DOP"}, {"state": "Rhode Island", "signup_date": "2023-08-04", "company": "Cyberdyne", "currency": "ALL"}, {"state": "Rhode Island", "signup_date": "2023-07-08", "company": "Initech", "currency": "DOP"}, {"state": "New York", "signup_date": "2023-11-16", "company": "Cyberdyne", "currency": "ARS"}, {"state": "Rhode Island", "signup_date": "2023-10-13", "company": "Stark Industries", "currency": "ALL"}, {"state": "Rhode Island", "signup_date": "2023-06-25", "company": "Soylent Corp", "currency": "ARS"}, {"state": "Rhode Island", "signup_date": "2023-08-06", "company": "Stark Industries", "currency": "DOP"}, {"state": "New York", "signup_date": "2023-02-24", "company": "Cyberdyne", "currency": "ALL"}, {"state": "Rhode Island", "signup_date": "2023-09-12", "company": "Stark Industries", "currency": "DOP"}, {"state": "New York", "signup_date": "2023-11-04", "company": "Cyberdyne", "currency": "DOP"}, {"state": "New York", "signup_date": "2023-10-11", "company": "Acme Inc", "currency": "ALL"}, {"state": "Rhode Island", "signup_date": "2023-11-20", "company": "Umbrella", "currency": "ALL"}, {"state": "New York", "signup_date": "2023-01-09", "company": "Stark Industries", "currency": "ALL"}, {"state": "Rhode Island", "signup_date": "2023-07-05", "company": "Initech", "currency": "ALL"}], "dirty_cols": ["state", "signup_date", "company", "currency", "col_x"], "clean_cols": ["state", "signup_date", "company", "currency"], "plan": {"dataset_summary": "20 rows × 5 columns. 4 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NY": "New York", "RI": "Rhode Island", "ri": "Rhode Island", "rhode islaand": "Rhode Island", "rhode island": "Rhode Island", "NEW YORK": "New York", "new york": "New York", "ny": "New York"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LEK": "ALL", "Dominican Peso": "DOP", "Lek": "ALL", "Dominican Peeso": "DOP", "all": "ALL", "dop": "DOP"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "48.8%", "name": "Carol  Cruz", "job_title": "Chief Technology Officer", "state": " nevada ", "unit": "MI", "email": "kbjk@test.org"}, {"rate": "6.8%", "name": "  Karl Fischer", "job_title": "Vp", "state": "north dakota ", "unit": "liter", "email": "JCPHPKM@CORP.IO"}, {"rate": "91.9%", "name": "Bob Brandt", "job_title": "CTO", "state": "nevada", "unit": "MI", "email": "gndgfhm@example.com"}, {"rate": "35.2%", "name": "Grace Diaz", "job_title": "prod mgr", "state": "NV", "unit": "Mi", "email": "LPHAHCK@TEST.ORG   "}, {"rate": "83.2%", "name": "Judy Adams", "job_title": "Product Mnaager", "state": "Nevada", "unit": " mile", "email": "aoifoh@test.org"}, {"rate": "8.2%", "name": "Frank Johnson", "job_title": "PM", "state": "North Dakota", "unit": "LITERS", "email": " khhgeah@test.org"}, {"rate": "21.4%", "name": "Omar  Ali", "job_title": "PM", "state": "nevada", "unit": "mile", "email": "jjdnfgi@mail.com"}, {"rate": "61.5%", "name": "Frank Moore", "job_title": "product manager", "state": "ND", "unit": " Mi", "email": "jbgmio@corp.io"}, {"rate": "20.1%", "name": "Judy  Wong", "job_title": "c.t.o.", "state": "North Dakoat", "unit": "mi", "email": "BJBMG@CORP.IO"}, {"rate": "89.2%", "name": "Bob  Petrov", "job_title": "Product Manager", "state": "NV  ", "unit": "liters", "email": "PFPHKED@TEST.ORG  "}, {"rate": "7.5%", "name": "   Judy Brandt", "job_title": "prod manager ", "state": "NV", "unit": "mi", "email": "hioe@corp.io"}, {"rate": "5.3%", "name": " Priya Lee", "job_title": "PM", "state": "NORTH DAKOTA", "unit": "L", "email": " NNHCK@MAIL.COM"}, {"rate": "15.9%", "name": "   Bob Johnson", "job_title": "PM", "state": "Nevada", "unit": "liters", "email": "OMKP@CORP.IO"}, {"rate": "34.9%", "name": " Lena Khan", "job_title": "Product Manager", "state": "NEVADA", "unit": "mile ", "email": "   inij@mail.com"}, {"rate": "19.5%", "name": "Lena Brandt", "job_title": " CTO", "state": "Nevada", "unit": "Mi ", "email": " djofjf@example.com "}, {"rate": "85.7%", "name": "Lena  Lee", "job_title": "C.T.O.", "state": "NORTH DAGOTA", "unit": "liters", "email": "IKAB@MAIL.COM"}, {"rate": "69.9%", "name": "Eve Reyes", "job_title": " CTO", "state": " north dakota", "unit": "mi", "email": "dgood@test.org"}, {"rate": "75.8%", "name": "Karl Lee", "job_title": "CHIEF TECHNOLOGY OFFICER", "state": "nv", "unit": " Mi", "email": "icdbk@corp.io"}, {"rate": "29.5%", "name": " Alice Smith", "job_title": "   vice president", "state": "north dakota", "unit": " liter ", "email": "BBMKK@EXAMPLE.COM"}, {"rate": "93.6%", "name": "Judy  Khan", "job_title": "Prod Mgr", "state": "nd", "unit": " mple", "email": "MPCN@TEST.ORG"}, {"rate": "82.5%", "name": "Sara Diaz", "job_title": "Vice Pres", "state": "neoada", "unit": "liter  ", "email": "KNJFNJB@EXAMPLE.COM"}, {"rate": "82.6%", "name": "Judy Ali  ", "job_title": "Vice President", "state": "nd", "unit": "liters", "email": "dgaom@test.org"}, {"rate": "55.2%", "name": "Mona Moore   ", "job_title": "Chief Technology Officer", "state": "nv", "unit": "miile", "email": "  ediofl@mail.com"}, {"rate": "9.5%", "name": "Ivan Cruz", "job_title": "cto", "state": "North Dakota ", "unit": "liter", "email": "   nhdcooh@test.org"}, {"rate": "74.5%", "name": "Sara Brandt", "job_title": "   cto", "state": "NV", "unit": "mi   ", "email": "mddlm@corp.io "}, {"rate": "72.1%", "name": "Heidi Petrov", "job_title": "Vice President", "state": "ND", "unit": "mils", "email": "ecip@test.org"}, {"rate": "70.5%", "name": "Heidi  Cruz", "job_title": "PM", "state": "NV", "unit": "MILE", "email": "nehedag@corp.io"}, {"rate": "40.2%", "name": "David Moore", "job_title": "Vice  Pres", "state": "ND", "unit": "liters", "email": "IAHO@MAIL.COM"}, {"rate": "78.9%", "name": "Carol Ali", "job_title": "CTO", "state": "  North Dakota", "unit": "Liter", "email": "jadd@example.com"}, {"rate": "98.8%", "name": "Lena Moore", "job_title": "chief technology officer", "state": " NV ", "unit": "l", "email": "  apho@example.com"}, {"rate": "73.3%", "name": "Alice Khan", "job_title": "VP   ", "state": "NORTH DAKOTA", "unit": "litre", "email": "PAINB@TEST.ORG "}, {"rate": "40.4%", "name": "Judy Petrov", "job_title": "vice president", "state": "Nv", "unit": "l", "email": " namfl@test.org "}, {"rate": "66.0%", "name": "  Carol Diaz", "job_title": "VP", "state": "ND", "unit": "mi", "email": "NNCOHNK@CORP.IO"}, {"rate": "53.5%", "name": "Priya Park   ", "job_title": "v.p.", "state": "NEVAAD", "unit": " liters ", "email": "bjmide@test.org"}, {"rate": "49.3%", "name": "Karl Reyes", "job_title": "product manager", "state": "nv", "unit": "   liters", "email": "ijel@corp.io"}, {"rate": "55.2%", "name": "Ivan Smith", "job_title": "PROD MANAGER", "state": "Nevada ", "unit": " litre ", "email": "JIEFJPL@EXAMPLE.COM"}, {"rate": "45.3%", "name": "Mona  Fischer", "job_title": "  Cto", "state": "North Dakota", "unit": "miles", "email": " GFFLHL@EXAMPLE.COM"}, {"rate": "65.6%", "name": "Bob Fischer", "job_title": "   Prod Manager", "state": " NV ", "unit": "   Mi", "email": "JMIL@TEST.ORG"}, {"rate": "31.5%", "name": "Judy  Brandt", "job_title": "Prod Manager", "state": "North Dakota", "unit": "liter", "email": "dpjn@test.org"}, {"rate": "69.4%", "name": "Karl Lee   ", "job_title": "C.T.O.", "state": " ND ", "unit": " Mi", "email": "   GOCEBGL@TEST.ORG"}, {"rate": "48.3%", "name": "Carol Fischer", "job_title": "C.T.O.", "state": "NV", "unit": "mile", "email": "jlfdced@test.org"}, {"rate": "53.2%", "name": "Mona Petrov", "job_title": "chief technology officer", "state": "norvh dakota ", "unit": "liitre   ", "email": "ieih@test.org"}, {"rate": "31.4%", "name": "Priya  Ali", "job_title": "  Vice Pres", "state": "North Dakota", "unit": "liter", "email": "ddlfcke@corp.io"}, {"rate": "1.5%", "name": "Priya Adams", "job_title": "CTO", "state": "nd", "unit": "Mi", "email": "akkmekk@corp.io "}, {"rate": "74.2%", "name": "Heidi Reyes", "job_title": " Prod Mgr", "state": "Nevada", "unit": "mi", "email": "JDIOODE@MAIL.COM"}, {"rate": "8.2%", "name": "Frank Johnson", "job_title": "PM", "state": "North Dakota", "unit": "LITERS", "email": " khhgeah@test.org"}, {"rate": "91.9%", "name": "Bob Brandt", "job_title": "CTO", "state": "nevada", "unit": "MI", "email": "gndgfhm@example.com"}], "clean": [{"rate": 0.488, "name": "Carol Cruz", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "kbjk@test.org"}, {"rate": 0.068, "name": "Karl Fischer", "job_title": "Vice President", "state": "North Dakota", "unit": "L", "email": "jcphpkm@corp.io"}, {"rate": 0.919, "name": "Bob Brandt", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "gndgfhm@example.com"}, {"rate": 0.35200000000000004, "name": "Grace Diaz", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "lphahck@test.org"}, {"rate": 0.8320000000000001, "name": "Judy Adams", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "aoifoh@test.org"}, {"rate": 0.08199999999999999, "name": "Frank Johnson", "job_title": "Product Manager", "state": "North Dakota", "unit": "L", "email": "khhgeah@test.org"}, {"rate": 0.214, "name": "Omar Ali", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "jjdnfgi@mail.com"}, {"rate": 0.615, "name": "Frank Moore", "job_title": "Product Manager", "state": "North Dakota", "unit": "mi", "email": "jbgmio@corp.io"}, {"rate": 0.201, "name": "Judy Wong", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "mi", "email": "bjbmg@corp.io"}, {"rate": 0.892, "name": "Bob Petrov", "job_title": "Product Manager", "state": "Nevada", "unit": "L", "email": "pfphked@test.org"}, {"rate": 0.075, "name": "Judy Brandt", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "hioe@corp.io"}, {"rate": 0.053, "name": "Priya Lee", "job_title": "Product Manager", "state": "North Dakota", "unit": "L", "email": "nnhck@mail.com"}, {"rate": 0.159, "name": "Bob Johnson", "job_title": "Product Manager", "state": "Nevada", "unit": "L", "email": "omkp@corp.io"}, {"rate": 0.349, "name": "Lena Khan", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "inij@mail.com"}, {"rate": 0.195, "name": "Lena Brandt", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "djofjf@example.com"}, {"rate": 0.857, "name": "Lena Lee", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "L", "email": "ikab@mail.com"}, {"rate": 0.6990000000000001, "name": "Eve Reyes", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "mi", "email": "dgood@test.org"}, {"rate": 0.758, "name": "Karl Lee", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "icdbk@corp.io"}, {"rate": 0.295, "name": "Alice Smith", "job_title": "Vice President", "state": "North Dakota", "unit": "L", "email": "bbmkk@example.com"}, {"rate": 0.9359999999999999, "name": "Judy Khan", "job_title": "Product Manager", "state": "North Dakota", "unit": "mi", "email": "mpcn@test.org"}, {"rate": 0.825, "name": "Sara Diaz", "job_title": "Vice President", "state": "Nevada", "unit": "L", "email": "knjfnjb@example.com"}, {"rate": 0.826, "name": "Judy Ali", "job_title": "Vice President", "state": "North Dakota", "unit": "L", "email": "dgaom@test.org"}, {"rate": 0.552, "name": "Mona Moore", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "ediofl@mail.com"}, {"rate": 0.095, "name": "Ivan Cruz", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "L", "email": "nhdcooh@test.org"}, {"rate": 0.745, "name": "Sara Brandt", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "mddlm@corp.io"}, {"rate": 0.721, "name": "Heidi Petrov", "job_title": "Vice President", "state": "North Dakota", "unit": "mi", "email": "ecip@test.org"}, {"rate": 0.705, "name": "Heidi Cruz", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "nehedag@corp.io"}, {"rate": 0.402, "name": "David Moore", "job_title": "Vice President", "state": "North Dakota", "unit": "L", "email": "iaho@mail.com"}, {"rate": 0.789, "name": "Carol Ali", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "L", "email": "jadd@example.com"}, {"rate": 0.988, "name": "Lena Moore", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "L", "email": "apho@example.com"}, {"rate": 0.733, "name": "Alice Khan", "job_title": "Vice President", "state": "North Dakota", "unit": "L", "email": "painb@test.org"}, {"rate": 0.40399999999999997, "name": "Judy Petrov", "job_title": "Vice President", "state": "Nevada", "unit": "L", "email": "namfl@test.org"}, {"rate": 0.66, "name": "Carol Diaz", "job_title": "Vice President", "state": "North Dakota", "unit": "mi", "email": "nncohnk@corp.io"}, {"rate": 0.535, "name": "Priya Park", "job_title": "Vice President", "state": "Nevada", "unit": "L", "email": "bjmide@test.org"}, {"rate": 0.493, "name": "Karl Reyes", "job_title": "Product Manager", "state": "Nevada", "unit": "L", "email": "ijel@corp.io"}, {"rate": 0.552, "name": "Ivan Smith", "job_title": "Product Manager", "state": "Nevada", "unit": "L", "email": "jiefjpl@example.com"}, {"rate": 0.45299999999999996, "name": "Mona Fischer", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "mi", "email": "gfflhl@example.com"}, {"rate": 0.6559999999999999, "name": "Bob Fischer", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "jmil@test.org"}, {"rate": 0.315, "name": "Judy Brandt", "job_title": "Product Manager", "state": "North Dakota", "unit": "L", "email": "dpjn@test.org"}, {"rate": 0.6940000000000001, "name": "Karl Lee", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "mi", "email": "gocebgl@test.org"}, {"rate": 0.483, "name": "Carol Fischer", "job_title": "Chief Technology Officer", "state": "Nevada", "unit": "mi", "email": "jlfdced@test.org"}, {"rate": 0.532, "name": "Mona Petrov", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "L", "email": "ieih@test.org"}, {"rate": 0.314, "name": "Priya Ali", "job_title": "Vice President", "state": "North Dakota", "unit": "L", "email": "ddlfcke@corp.io"}, {"rate": 0.015, "name": "Priya Adams", "job_title": "Chief Technology Officer", "state": "North Dakota", "unit": "mi", "email": "akkmekk@corp.io"}, {"rate": 0.742, "name": "Heidi Reyes", "job_title": "Product Manager", "state": "Nevada", "unit": "mi", "email": "jdioode@mail.com"}], "dirty_cols": ["rate", "name", "job_title", "state", "unit", "email"], "clean_cols": ["rate", "name", "job_title", "state", "unit", "email"], "plan": {"dataset_summary": "47 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Vp": "Vice President", "CTO": "Chief Technology Officer", "prod mgr": "Product Manager", "Product Mnaager": "Product Manager", "PM": "Product Manager", "product manager": "Product Manager", "c.t.o.": "Chief Technology Officer", "prod manager": "Product Manager", "C.T.O.": "Chief Technology Officer", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "vice president": "Vice President", "Prod Mgr": "Product Manager", "Vice Pres": "Vice President", "cto": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "VP": "Vice President", "v.p.": "Vice President", "PROD MANAGER": "Product Manager", "Cto": "Chief Technology Officer", "Prod Manager": "Product Manager"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"nevada": "Nevada", "north dakota": "North Dakota", "NV": "Nevada", "ND": "North Dakota", "North Dakoat": "North Dakota", "NORTH DAKOTA": "North Dakota", "NEVADA": "Nevada", "NORTH DAGOTA": "North Dakota", "nv": "Nevada", "nd": "North Dakota", "neoada": "Nevada", "Nv": "Nevada", "NEVAAD": "Nevada", "norvh dakota": "North Dakota"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MI": "mi", "liter": "L", "Mi": "mi", "mile": "mi", "LITERS": "L", "liters": "L", "mple": "mi", "miile": "mi", "mils": "mi", "MILE": "mi", "Liter": "L", "l": "L", "litre": "L", "miles": "mi", "liitre": "L"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"industry": "finserv", "city": "Colne", "department": "rd", "is_active": "0", "signup_date": "23 May 2023", "extra": ""}, {"industry": "Industriial", "city": "Cole   ", "department": "Fin", "is_active": "Y", "signup_date": "2023-06-14", "extra": ""}, {"industry": "manufacturing", "city": "Colne", "department": "  Accounting", "is_active": "true", "signup_date": "2023-09-01", "extra": ""}, {"industry": "Mfg", "city": "Marsava", "department": "Mkt", "is_active": "No", "signup_date": "5 Aug 2023", "extra": ""}, {"industry": "Manufacturing ", "city": "marsala", "department": "Accouting", "is_active": "T", "signup_date": "45111", "extra": ""}, {"industry": "FINTECH", "city": "Alfafar", "department": "ENG", "is_active": "T", "signup_date": "12/19/2023", "extra": ""}, {"industry": " banking", "city": "Marrala", "department": "accounting", "is_active": "Y", "signup_date": "45155", "extra": ""}, {"industry": " idustrial ", "city": "Alfafar", "department": "Dev  ", "is_active": "N", "signup_date": "2023-12-05", "extra": ""}, {"industry": "   Finserv", "city": " Colne ", "department": "Growth", "is_active": "F", "signup_date": "26 Aug 2023", "extra": ""}, {"industry": "Mfg", "city": "Colne", "department": " ENG. ", "is_active": "No", "signup_date": "4 Mar 2023", "extra": ""}, {"industry": "BANKING", "city": "alfafar", "department": "marketing", "is_active": "1", "signup_date": "2023-06-08", "extra": ""}, {"industry": " Financial Services", "city": "colne", "department": "   Finance", "is_active": "No", "signup_date": "6/18/2023", "extra": ""}, {"industry": "Manuffacturing  ", "city": "marbala", "department": "ENG", "is_active": "No", "signup_date": "45078", "extra": ""}, {"industry": "FINSERV", "city": " Marsala ", "department": "Engineering", "is_active": "Y", "signup_date": "45267", "extra": ""}, {"industry": "Iddustrial", "city": " Marsala", "department": "Finance", "is_active": "No", "signup_date": "45084", "extra": ""}, {"industry": "Financial Services", "city": "Marsala", "department": "Accounting", "is_active": "F", "signup_date": "2/8/2023", "extra": ""}, {"industry": "Mfg", "city": "Marsava", "department": "Mkt", "is_active": "No", "signup_date": "5 Aug 2023", "extra": ""}, {"industry": "FINSERV", "city": " Marsala ", "department": "Engineering", "is_active": "Y", "signup_date": "45267", "extra": ""}], "clean": [{"industry": "Financial Services", "city": "Colne", "department": "Engineering", "is_active": false, "signup_date": "2023-05-23"}, {"industry": "Manufacturing", "city": "Colne", "department": "Finance", "is_active": true, "signup_date": "2023-06-14"}, {"industry": "Manufacturing", "city": "Colne", "department": "Finance", "is_active": true, "signup_date": "2023-09-01"}, {"industry": "Manufacturing", "city": "Marsala", "department": "Marketing", "is_active": false, "signup_date": "2023-08-05"}, {"industry": "Manufacturing", "city": "Marsala", "department": "Finance", "is_active": true, "signup_date": "2023-07-04"}, {"industry": "Financial Services", "city": "Alfafar", "department": "Engineering", "is_active": true, "signup_date": "2023-12-19"}, {"industry": "Financial Services", "city": "Marsala", "department": "Finance", "is_active": true, "signup_date": "2023-08-17"}, {"industry": "Manufacturing", "city": "Alfafar", "department": "Engineering", "is_active": false, "signup_date": "2023-12-05"}, {"industry": "Financial Services", "city": "Colne", "department": "Marketing", "is_active": false, "signup_date": "2023-08-26"}, {"industry": "Manufacturing", "city": "Colne", "department": "Engineering", "is_active": false, "signup_date": "2023-03-04"}, {"industry": "Financial Services", "city": "Alfafar", "department": "Marketing", "is_active": true, "signup_date": "2023-06-08"}, {"industry": "Financial Services", "city": "Colne", "department": "Finance", "is_active": false, "signup_date": "2023-06-18"}, {"industry": "Manufacturing", "city": "Marsala", "department": "Engineering", "is_active": false, "signup_date": "2023-06-01"}, {"industry": "Financial Services", "city": "Marsala", "department": "Engineering", "is_active": true, "signup_date": "2023-12-07"}, {"industry": "Manufacturing", "city": "Marsala", "department": "Finance", "is_active": false, "signup_date": "2023-06-07"}, {"industry": "Financial Services", "city": "Marsala", "department": "Finance", "is_active": false, "signup_date": "2023-02-08"}], "dirty_cols": ["industry", "city", "department", "is_active", "signup_date", "extra"], "clean_cols": ["industry", "city", "department", "is_active", "signup_date"], "plan": {"dataset_summary": "18 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"finserv": "Financial Services", "Industriial": "Manufacturing", "manufacturing": "Manufacturing", "Mfg": "Manufacturing", "FINTECH": "Financial Services", "banking": "Financial Services", "idustrial": "Manufacturing", "Finserv": "Financial Services", "BANKING": "Financial Services", "Manuffacturing": "Manufacturing", "FINSERV": "Financial Services", "Iddustrial": "Manufacturing"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cole": "Colne", "Marsava": "Marsala", "marsala": "Marsala", "Marrala": "Marsala", "alfafar": "Alfafar", "colne": "Colne", "marbala": "Marsala"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"rd": "Engineering", "Fin": "Finance", "Accounting": "Finance", "Mkt": "Marketing", "Accouting": "Finance", "ENG": "Engineering", "accounting": "Finance", "Dev": "Engineering", "Growth": "Marketing", "ENG.": "Engineering", "marketing": "Marketing"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"email": "LHFO@TEST.ORG", "department": "Cust  Support", "amount": "5.458,75", "rate": "22.4%", "status": "CHURNED", "name": "   Grace Smith", "notes2": ""}, {"email": "hemp@test.org", "department": "CS", "amount": "#N/A", "rate": "82.3%", "status": "ACTIEV", "name": "   Judy Wong", "notes2": ""}, {"email": " IHGKAE@CORP.IO ", "department": "PM", "amount": "538,09", "rate": "4.7%", "status": "canceled", "name": "Alice Diaz", "notes2": ""}, {"email": "oihmkcp@example.com", "department": "Cs", "amount": "2.449,74", "rate": "45.8%", "status": "   ACTIVE", "name": "David  Park", "notes2": ""}, {"email": " GJKJI@EXAMPLE.COM ", "department": "legal", "amount": "7.645,94", "rate": "38.8%", "status": "ACTIVE", "name": "Alice Lee   ", "notes2": ""}, {"email": "BBBA@CORP.IO", "department": "Product", "amount": "1.485,65", "rate": "45.0%", "status": " ACTIVE", "name": "Sara Khan", "notes2": ""}, {"email": "nidol@corp.io", "department": " leal ", "amount": "4.094,86", "rate": "27.7%", "status": "active", "name": "   Frank Cruz", "notes2": ""}, {"email": "fcba@mail.com", "department": "PM", "amount": "4.946,25", "rate": "77.8%", "status": "Active", "name": "Priya Novak   ", "notes2": ""}, {"email": "  JPCIMEL@EXAMPLE.COM", "department": " Product ", "amount": "$5,121.72", "rate": "15.0%", "status": " Active", "name": "Ivan Park", "notes2": ""}, {"email": "dhkp@corp.io", "department": "pm", "amount": "3.127,38", "rate": "95.8%", "status": "Active", "name": "  Carol Lee", "notes2": ""}, {"email": " hdgi@example.com ", "department": " SUPPORT", "amount": "2.810,14", "rate": "55.2%", "status": "churned", "name": " Omar Moore", "notes2": ""}, {"email": "obamgl@example.com", "department": "PM", "amount": "1.415,57", "rate": "57.1%", "status": "Active", "name": "Priya  Petrov", "notes2": ""}, {"email": " fbgdd@mail.com", "department": "PROD", "amount": "7.887,84", "rate": "1.7%", "status": "Churned", "name": "David Wong", "notes2": ""}, {"email": "  MIBAG@MAIL.COM", "department": "CS", "amount": "na", "rate": "89.5%", "status": "Active", "name": "Bob Reyes   ", "notes2": ""}, {"email": "ogch@corp.io", "department": "PM", "amount": "$5,344.48", "rate": "14.5%", "status": "CHURNED", "name": "Alice Johnson ", "notes2": ""}, {"email": "cpmob@corp.io  ", "department": "CS", "amount": "1.532,86", "rate": "35.4%", "status": "CANCELED", "name": "Alice Wong  ", "notes2": ""}, {"email": " NPFOH@TEST.ORG ", "department": "PM", "amount": "1.528,56", "rate": "39.6%", "status": "Active", "name": "Carol Reyes   ", "notes2": ""}, {"email": " epihigj@example.com ", "department": "Supzort", "amount": "4.334,71", "rate": "7.4%", "status": "churned", "name": "Karl Cruz", "notes2": ""}, {"email": "fhdk@example.com", "department": " Legaal ", "amount": "609,98", "rate": "94.0%", "status": "Churned", "name": "Judy Khan", "notes2": ""}, {"email": "cial@mail.com", "department": "CUST SUPPORT", "amount": "$8,020.60", "rate": "80.9%", "status": "active", "name": "Judy Diaz", "notes2": ""}, {"email": "coack@example.com", "department": "legal & compliance", "amount": "$1,934.75", "rate": "41.0%", "status": "canceled ", "name": "Ivan Cruz", "notes2": ""}, {"email": "  fgpa@mail.com", "department": "LEGAL & COMPLIANCE   ", "amount": "8.802,33", "rate": "41.1%", "status": " ACTIVE ", "name": "Alice Lee", "notes2": ""}, {"email": "  cppjn@mail.com", "department": "cs", "amount": "8.507,30", "rate": "22.8%", "status": "ACTIHE", "name": "Judy Smith", "notes2": ""}, {"email": "OGDBBJ@EXAMPLE.COM", "department": "Leagl", "amount": "#N/A", "rate": "25.1%", "status": "Churned", "name": "Lena Cruz", "notes2": ""}, {"email": "fbbb@example.com", "department": "pm", "amount": "$5,826.00", "rate": "95.1%", "status": "canceled", "name": "Alice Adams", "notes2": ""}, {"email": " hdgi@example.com ", "department": " SUPPORT", "amount": "2.810,14", "rate": "55.2%", "status": "churned", "name": " Omar Moore", "notes2": ""}, {"email": "", "department": "", "amount": "", "rate": "", "status": "", "name": "", "notes2": ""}], "clean": [{"email": "lhfo@test.org", "department": "Customer Support", "amount": 5458.75, "rate": 0.22399999999999998, "status": "Churned", "name": "Grace Smith"}, {"email": "hemp@test.org", "department": "Customer Support", "amount": NaN, "rate": 0.823, "status": "Active", "name": "Judy Wong"}, {"email": "ihgkae@corp.io", "department": "Product", "amount": 538.09, "rate": 0.047, "status": "Churned", "name": "Alice Diaz"}, {"email": "oihmkcp@example.com", "department": "Customer Support", "amount": 2449.74, "rate": 0.45799999999999996, "status": "Active", "name": "David Park"}, {"email": "gjkji@example.com", "department": "Legal", "amount": 7645.94, "rate": 0.38799999999999996, "status": "Active", "name": "Alice Lee"}, {"email": "bbba@corp.io", "department": "Product", "amount": 1485.65, "rate": 0.45, "status": "Active", "name": "Sara Khan"}, {"email": "nidol@corp.io", "department": "Legal", "amount": 4094.86, "rate": 0.27699999999999997, "status": "Active", "name": "Frank Cruz"}, {"email": "fcba@mail.com", "department": "Product", "amount": 4946.25, "rate": 0.778, "status": "Active", "name": "Priya Novak"}, {"email": "jpcimel@example.com", "department": "Product", "amount": 5121.72, "rate": 0.15, "status": "Active", "name": "Ivan Park"}, {"email": "dhkp@corp.io", "department": "Product", "amount": 3127.38, "rate": 0.958, "status": "Active", "name": "Carol Lee"}, {"email": "hdgi@example.com", "department": "Customer Support", "amount": 2810.14, "rate": 0.552, "status": "Churned", "name": "Omar Moore"}, {"email": "obamgl@example.com", "department": "Product", "amount": 1415.57, "rate": 0.5710000000000001, "status": "Active", "name": "Priya Petrov"}, {"email": "fbgdd@mail.com", "department": "Product", "amount": 7887.84, "rate": 0.017, "status": "Churned", "name": "David Wong"}, {"email": "mibag@mail.com", "department": "Customer Support", "amount": NaN, "rate": 0.895, "status": "Active", "name": "Bob Reyes"}, {"email": "ogch@corp.io", "department": "Product", "amount": 5344.48, "rate": 0.145, "status": "Churned", "name": "Alice Johnson"}, {"email": "cpmob@corp.io", "department": "Customer Support", "amount": 1532.86, "rate": 0.354, "status": "Churned", "name": "Alice Wong"}, {"email": "npfoh@test.org", "department": "Product", "amount": 1528.56, "rate": 0.396, "status": "Active", "name": "Carol Reyes"}, {"email": "epihigj@example.com", "department": "Customer Support", "amount": 4334.71, "rate": 0.07400000000000001, "status": "Churned", "name": "Karl Cruz"}, {"email": "fhdk@example.com", "department": "Legal", "amount": 609.98, "rate": 0.94, "status": "Churned", "name": "Judy Khan"}, {"email": "cial@mail.com", "department": "Customer Support", "amount": 8020.6, "rate": 0.809, "status": "Active", "name": "Judy Diaz"}, {"email": "coack@example.com", "department": "Legal", "amount": 1934.75, "rate": 0.41, "status": "Churned", "name": "Ivan Cruz"}, {"email": "fgpa@mail.com", "department": "Legal", "amount": 8802.33, "rate": 0.41100000000000003, "status": "Active", "name": "Alice Lee"}, {"email": "cppjn@mail.com", "department": "Customer Support", "amount": 8507.3, "rate": 0.228, "status": "Active", "name": "Judy Smith"}, {"email": "ogdbbj@example.com", "department": "Legal", "amount": NaN, "rate": 0.251, "status": "Churned", "name": "Lena Cruz"}, {"email": "fbbb@example.com", "department": "Product", "amount": 5826.0, "rate": 0.951, "status": "Churned", "name": "Alice Adams"}], "dirty_cols": ["email", "department", "amount", "rate", "status", "name", "notes2"], "clean_cols": ["email", "department", "amount", "rate", "status", "name"], "plan": {"dataset_summary": "27 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cust Support": "Customer Support", "CS": "Customer Support", "PM": "Product", "Cs": "Customer Support", "legal": "Legal", "leal": "Legal", "pm": "Product", "SUPPORT": "Customer Support", "PROD": "Product", "Supzort": "Customer Support", "Legaal": "Legal", "CUST SUPPORT": "Customer Support", "legal & compliance": "Legal", "LEGAL & COMPLIANCE": "Legal", "cs": "Customer Support", "Leagl": "Legal"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CHURNED": "Churned", "ACTIEV": "Active", "canceled": "Churned", "ACTIVE": "Active", "active": "Active", "churned": "Churned", "CANCELED": "Churned", "ACTIHE": "Active"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"industry": "IT  ", "phone": "(545)047-3759", "name": "Eve Park   ", "job_title": "   Dir.", "unit": "liter", "city": "Nova Andradina", "amount": "$4,716.88", "status": "Low  ", "unnamed": ""}, {"industry": "utilitiees", "phone": "508.563.3579", "name": "Heidi Ali", "job_title": "Administrative Assistant", "unit": "kg", "city": "Aonla", "amount": "2.072,93", "status": " P2", "unnamed": ""}, {"industry": "comms", "phone": "5583373978", "name": "Karl Brandt", "job_title": "Admin Assistant", "unit": "l   ", "city": "de meern", "amount": "5.551,53", "status": " medium", "unnamed": ""}, {"industry": " telecom ", "phone": "(596)871-1039", "name": "Priya Lee", "job_title": "Dir.", "unit": " kg", "city": "NOVA ANDRADINA", "amount": "$7,618.49", "status": "Medium", "unnamed": ""}, {"industry": "IT", "phone": "5156600373", "name": "Heidi  Lee", "job_title": "Engineering Mgr", "unit": " l ", "city": "nova andradina", "amount": "$3,744.84", "status": " L ", "unnamed": ""}, {"industry": "IT", "phone": "588-087-2597", "name": "Frank Wong", "job_title": "Engineering Manager", "unit": "  liters", "city": "  Nova Andradina", "amount": "N/A", "status": "Medium", "unnamed": ""}, {"industry": "Telco", "phone": "(501)113-2938", "name": "Ivan  Reyes", "job_title": "eng manager", "unit": " lbs ", "city": "AFIF", "amount": "$5,337.43", "status": "medium", "unnamed": ""}, {"industry": "Coms", "phone": "542.876.4165", "name": "Karl Petrov", "job_title": "Dir", "unit": "kilograms", "city": "nova andradina", "amount": "8.075,58", "status": "LOW   ", "unnamed": ""}, {"industry": "Energy", "phone": "573-136-2125", "name": " Ivan Novak", "job_title": "Admin ", "unit": " Ponds ", "city": "puerto ayacucho   ", "amount": "5.696,13", "status": "low", "unnamed": ""}, {"industry": "Software", "phone": "5682933344", "name": " Priya Brandt", "job_title": "Director", "unit": "kg", "city": "De Meern", "amount": "4.503,85", "status": "   p3", "unnamed": ""}, {"industry": "energy", "phone": "5007259930", "name": "Frank Johnson", "job_title": "prod mgr", "unit": "Lb", "city": "De Meern", "amount": "5.497,48", "status": "L", "unnamed": ""}, {"industry": "oil  & gas", "phone": "597-936-5113", "name": "Lena Moore", "job_title": "dir", "unit": "   Kg", "city": "puerto ayacucho", "amount": "$5,237.18", "status": "MED", "unnamed": ""}, {"industry": " Telecommunications", "phone": "5956032528", "name": "David Petrov", "job_title": "Admin", "unit": "L", "city": "AONLA", "amount": "$982.21", "status": "MED", "unnamed": ""}, {"industry": "Oil & Gas", "phone": "(561)958-8903", "name": " Alice Johnson", "job_title": "Eng  Manager", "unit": "  kg", "city": "NOVA ANDRADINA", "amount": "7.410,89", "status": " p2 ", "unnamed": ""}, {"industry": "tech  ", "phone": "5951004874", "name": "Alice  Diaz", "job_title": " dir ", "unit": "lb", "city": "afif", "amount": "$6,797.62", "status": " med", "unnamed": ""}, {"industry": "TELCO", "phone": "5959877902", "name": "Carol Moore", "job_title": "Dir", "unit": "liter", "city": "Puerto Ayacxcho ", "amount": "8.632,69", "status": "P3", "unnamed": ""}, {"industry": "ENERGY", "phone": "583.185.9444", "name": "Bob Diaz   ", "job_title": "Dir", "unit": "l", "city": "NOVA ANDRADINA", "amount": "$895.64", "status": "low", "unnamed": ""}, {"industry": "Energy", "phone": "514.019.0708", "name": "Priya Johnson", "job_title": "prod manager", "unit": "lb", "city": "De Meern ", "amount": "5.286,01", "status": "  Low", "unnamed": ""}, {"industry": " Software ", "phone": "(529)118-6295", "name": "Frank Diaz", "job_title": "Director", "unit": "kilograms", "city": "De Meeern", "amount": "1.184,34", "status": "  P2", "unnamed": ""}, {"industry": "comms", "phone": "(556)554-4106", "name": "David Reyes   ", "job_title": "Administrative Assistant", "unit": "kg   ", "city": "NOVA ANDRADINA", "amount": "$8,822.30", "status": "m", "unnamed": ""}, {"industry": "Technology", "phone": "(573)456-5286", "name": "Heidi Johnson ", "job_title": "prod manager", "unit": "Lb", "city": "Puerto Ayacucho", "amount": "2.250,25", "status": "L", "unnamed": ""}, {"industry": "Software", "phone": "(573)968-3388", "name": "  Omar Novak", "job_title": "Administrative Assistant", "unit": "litre", "city": "Nova Andradina", "amount": "268,36", "status": "mediuum", "unnamed": ""}, {"industry": "IT ", "phone": "553.317.8346", "name": "Sara  Lee", "job_title": "Prod Manager", "unit": "killogram", "city": "nova andradina", "amount": "636,97", "status": "LOW", "unnamed": ""}, {"industry": "telecom", "phone": "502.238.6986", "name": "Grace Lee", "job_title": "   administrative assistant", "unit": "kg", "city": "de meern", "amount": "7.210,82", "status": "p2", "unnamed": ""}, {"industry": "Tech", "phone": "5123023271", "name": "Omar  Khan", "job_title": "Admin Assistant", "unit": "lbs", "city": "Puerto Ayacucho", "amount": "862,33", "status": "   Low", "unnamed": ""}, {"industry": "Comms", "phone": "521.134.3725", "name": "Omar  Wong", "job_title": "Admin", "unit": "  L", "city": "puerto ayacucho", "amount": "$4,727.15", "status": "Med", "unnamed": ""}, {"industry": "   oil & gas", "phone": "5169283765", "name": "  Karl Ali", "job_title": "Director", "unit": "LITTRE ", "city": "de meern", "amount": "$4,380.12", "status": " LOW ", "unnamed": ""}, {"industry": "oil & gas", "phone": "591-215-9433", "name": "Bob  Brandt", "job_title": "eng mgr", "unit": "kilograms", "city": "aonla", "amount": "2.351,87", "status": "Med", "unnamed": ""}, {"industry": "telco", "phone": "509-905-3241", "name": "  Omar Johnson", "job_title": "  Administrative Assistant", "unit": "Lb", "city": "aonla", "amount": "1.094,53", "status": "low", "unnamed": ""}, {"industry": "   Tech", "phone": "531-387-5731", "name": "Heidi  Cruz", "job_title": "Admin Asst", "unit": "kg", "city": "   Afif", "amount": "$6,966.36", "status": "low", "unnamed": ""}, {"industry": "   tech", "phone": "518.776.8660", "name": "Omar Novak", "job_title": "Dir.", "unit": "kilogram", "city": "de meern", "amount": "$3,088.82", "status": "LOW", "unnamed": ""}, {"industry": "TELECOM", "phone": "523.100.5690", "name": " Sara Adams", "job_title": " DIR", "unit": "  kg", "city": "afif", "amount": "null", "status": "low", "unnamed": ""}, {"industry": "SOFTWARE", "phone": "580-284-0747", "name": "Omar Johnson ", "job_title": "administrative assistant", "unit": "Lb", "city": "Aonla", "amount": "7.731,64", "status": " m", "unnamed": ""}, {"industry": " energy", "phone": "5158223033", "name": "Ivan Lee  ", "job_title": "engineering manager", "unit": "L", "city": "nova andradina", "amount": "--", "status": "  l", "unnamed": ""}, {"industry": "Telecom", "phone": "(545)080-9626", "name": " Judy Adams", "job_title": "Director", "unit": "Lbs", "city": "Aonla", "amount": "5.242,21", "status": "Low", "unnamed": ""}, {"industry": "Enerwy", "phone": "(574)295-1955", "name": "David Novak ", "job_title": "engineering  manager", "unit": " Liter ", "city": "   NOVA ANDFADINA", "amount": "8.465,47", "status": "M", "unnamed": ""}, {"industry": "  Telecommunications", "phone": "516.072.7232", "name": "David  Khan", "job_title": "engineering manager", "unit": "kg ", "city": "  Nova Andradina", "amount": "$1,138.29", "status": "Low", "unnamed": ""}, {"industry": " Technology ", "phone": "555.179.6380", "name": " Alice Wong", "job_title": "admin asst", "unit": "lbs  ", "city": "aonla", "amount": "$6,538.02", "status": "P2", "unnamed": ""}, {"industry": "TCEH", "phone": "(516)983-7597", "name": "Omar Reyes", "job_title": "Director   ", "unit": "lbs", "city": "   Afif", "amount": "4.185,72", "status": "p2", "unnamed": ""}, {"industry": "telecommunications", "phone": "5045334577", "name": "Alice Novak", "job_title": "Admin Asst", "unit": "l", "city": "De Meern", "amount": "--", "status": "P2", "unnamed": ""}, {"industry": "Technology  ", "phone": "5266742635", "name": "Priya Novak", "job_title": "Prod Mgr", "unit": "kg", "city": "puerto ayacucho", "amount": "$2,298.09", "status": "Low", "unnamed": ""}, {"industry": " Ccmms ", "phone": "577-728-2978", "name": "Ivan Brandt", "job_title": "Engineering Manager", "unit": "Kg", "city": "nova andradina", "amount": "$2,668.68", "status": "M", "unnamed": ""}, {"industry": "Telecom", "phone": "585-222-3487", "name": "Sara  Brandt", "job_title": "Admin Assistant", "unit": "kilograams", "city": "De Meern", "amount": "6.200,62", "status": "medium", "unnamed": ""}, {"industry": "", "phone": "", "name": "", "job_title": "", "unit": "", "city": "", "amount": "", "status": "", "unnamed": ""}], "clean": [{"industry": "Technology", "phone": "(545) 047-3759", "name": "Eve Park", "job_title": "Director", "unit": "L", "city": "Nova Andradina", "amount": 4716.88, "status": "Low"}, {"industry": "Energy", "phone": "(508) 563-3579", "name": "Heidi Ali", "job_title": "Administrative Assistant", "unit": "kg", "city": "Aonla", "amount": 2072.93, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(558) 337-3978", "name": "Karl Brandt", "job_title": "Administrative Assistant", "unit": "L", "city": "De Meern", "amount": 5551.53, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(596) 871-1039", "name": "Priya Lee", "job_title": "Director", "unit": "kg", "city": "Nova Andradina", "amount": 7618.49, "status": "Medium"}, {"industry": "Technology", "phone": "(515) 660-0373", "name": "Heidi Lee", "job_title": "Engineering Manager", "unit": "L", "city": "Nova Andradina", "amount": 3744.84, "status": "Low"}, {"industry": "Technology", "phone": "(588) 087-2597", "name": "Frank Wong", "job_title": "Engineering Manager", "unit": "L", "city": "Nova Andradina", "amount": NaN, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(501) 113-2938", "name": "Ivan Reyes", "job_title": "Engineering Manager", "unit": "lb", "city": "Afif", "amount": 5337.43, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(542) 876-4165", "name": "Karl Petrov", "job_title": "Director", "unit": "kg", "city": "Nova Andradina", "amount": 8075.58, "status": "Low"}, {"industry": "Energy", "phone": "(573) 136-2125", "name": "Ivan Novak", "job_title": "Administrative Assistant", "unit": "lb", "city": "Puerto Ayacucho", "amount": 5696.13, "status": "Low"}, {"industry": "Technology", "phone": "(568) 293-3344", "name": "Priya Brandt", "job_title": "Director", "unit": "kg", "city": "De Meern", "amount": 4503.85, "status": "Low"}, {"industry": "Energy", "phone": "(500) 725-9930", "name": "Frank Johnson", "job_title": "Product Manager", "unit": "lb", "city": "De Meern", "amount": 5497.48, "status": "Low"}, {"industry": "Energy", "phone": "(597) 936-5113", "name": "Lena Moore", "job_title": "Director", "unit": "kg", "city": "Puerto Ayacucho", "amount": 5237.18, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(595) 603-2528", "name": "David Petrov", "job_title": "Administrative Assistant", "unit": "L", "city": "Aonla", "amount": 982.21, "status": "Medium"}, {"industry": "Energy", "phone": "(561) 958-8903", "name": "Alice Johnson", "job_title": "Engineering Manager", "unit": "kg", "city": "Nova Andradina", "amount": 7410.89, "status": "Medium"}, {"industry": "Technology", "phone": "(595) 100-4874", "name": "Alice Diaz", "job_title": "Director", "unit": "lb", "city": "Afif", "amount": 6797.62, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(595) 987-7902", "name": "Carol Moore", "job_title": "Director", "unit": "L", "city": "Puerto Ayacucho", "amount": 8632.69, "status": "Low"}, {"industry": "Energy", "phone": "(583) 185-9444", "name": "Bob Diaz", "job_title": "Director", "unit": "L", "city": "Nova Andradina", "amount": 895.64, "status": "Low"}, {"industry": "Energy", "phone": "(514) 019-0708", "name": "Priya Johnson", "job_title": "Product Manager", "unit": "lb", "city": "De Meern", "amount": 5286.01, "status": "Low"}, {"industry": "Technology", "phone": "(529) 118-6295", "name": "Frank Diaz", "job_title": "Director", "unit": "kg", "city": "De Meern", "amount": 1184.34, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(556) 554-4106", "name": "David Reyes", "job_title": "Administrative Assistant", "unit": "kg", "city": "Nova Andradina", "amount": 8822.3, "status": "Medium"}, {"industry": "Technology", "phone": "(573) 456-5286", "name": "Heidi Johnson", "job_title": "Product Manager", "unit": "lb", "city": "Puerto Ayacucho", "amount": 2250.25, "status": "Low"}, {"industry": "Technology", "phone": "(573) 968-3388", "name": "Omar Novak", "job_title": "Administrative Assistant", "unit": "L", "city": "Nova Andradina", "amount": 268.36, "status": "Medium"}, {"industry": "Technology", "phone": "(553) 317-8346", "name": "Sara Lee", "job_title": "Product Manager", "unit": "kg", "city": "Nova Andradina", "amount": 636.97, "status": "Low"}, {"industry": "Telecommunications", "phone": "(502) 238-6986", "name": "Grace Lee", "job_title": "Administrative Assistant", "unit": "kg", "city": "De Meern", "amount": 7210.82, "status": "Medium"}, {"industry": "Technology", "phone": "(512) 302-3271", "name": "Omar Khan", "job_title": "Administrative Assistant", "unit": "lb", "city": "Puerto Ayacucho", "amount": 862.33, "status": "Low"}, {"industry": "Telecommunications", "phone": "(521) 134-3725", "name": "Omar Wong", "job_title": "Administrative Assistant", "unit": "L", "city": "Puerto Ayacucho", "amount": 4727.15, "status": "Medium"}, {"industry": "Energy", "phone": "(516) 928-3765", "name": "Karl Ali", "job_title": "Director", "unit": "L", "city": "De Meern", "amount": 4380.12, "status": "Low"}, {"industry": "Energy", "phone": "(591) 215-9433", "name": "Bob Brandt", "job_title": "Engineering Manager", "unit": "kg", "city": "Aonla", "amount": 2351.87, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(509) 905-3241", "name": "Omar Johnson", "job_title": "Administrative Assistant", "unit": "lb", "city": "Aonla", "amount": 1094.53, "status": "Low"}, {"industry": "Technology", "phone": "(531) 387-5731", "name": "Heidi Cruz", "job_title": "Administrative Assistant", "unit": "kg", "city": "Afif", "amount": 6966.36, "status": "Low"}, {"industry": "Technology", "phone": "(518) 776-8660", "name": "Omar Novak", "job_title": "Director", "unit": "kg", "city": "De Meern", "amount": 3088.82, "status": "Low"}, {"industry": "Telecommunications", "phone": "(523) 100-5690", "name": "Sara Adams", "job_title": "Director", "unit": "kg", "city": "Afif", "amount": NaN, "status": "Low"}, {"industry": "Technology", "phone": "(580) 284-0747", "name": "Omar Johnson", "job_title": "Administrative Assistant", "unit": "lb", "city": "Aonla", "amount": 7731.64, "status": "Medium"}, {"industry": "Energy", "phone": "(515) 822-3033", "name": "Ivan Lee", "job_title": "Engineering Manager", "unit": "L", "city": "Nova Andradina", "amount": NaN, "status": "Low"}, {"industry": "Telecommunications", "phone": "(545) 080-9626", "name": "Judy Adams", "job_title": "Director", "unit": "lb", "city": "Aonla", "amount": 5242.21, "status": "Low"}, {"industry": "Energy", "phone": "(574) 295-1955", "name": "David Novak", "job_title": "Engineering Manager", "unit": "L", "city": "Nova Andradina", "amount": 8465.47, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(516) 072-7232", "name": "David Khan", "job_title": "Engineering Manager", "unit": "kg", "city": "Nova Andradina", "amount": 1138.29, "status": "Low"}, {"industry": "Technology", "phone": "(555) 179-6380", "name": "Alice Wong", "job_title": "Administrative Assistant", "unit": "lb", "city": "Aonla", "amount": 6538.02, "status": "Medium"}, {"industry": "Technology", "phone": "(516) 983-7597", "name": "Omar Reyes", "job_title": "Director", "unit": "lb", "city": "Afif", "amount": 4185.72, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(504) 533-4577", "name": "Alice Novak", "job_title": "Administrative Assistant", "unit": "L", "city": "De Meern", "amount": NaN, "status": "Medium"}, {"industry": "Technology", "phone": "(526) 674-2635", "name": "Priya Novak", "job_title": "Product Manager", "unit": "kg", "city": "Puerto Ayacucho", "amount": 2298.09, "status": "Low"}, {"industry": "Telecommunications", "phone": "(577) 728-2978", "name": "Ivan Brandt", "job_title": "Engineering Manager", "unit": "kg", "city": "Nova Andradina", "amount": 2668.68, "status": "Medium"}, {"industry": "Telecommunications", "phone": "(585) 222-3487", "name": "Sara Brandt", "job_title": "Administrative Assistant", "unit": "kg", "city": "De Meern", "amount": 6200.62, "status": "Medium"}], "dirty_cols": ["industry", "phone", "name", "job_title", "unit", "city", "amount", "status", "unnamed"], "clean_cols": ["industry", "phone", "name", "job_title", "unit", "city", "amount", "status"], "plan": {"dataset_summary": "44 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IT": "Technology", "utilitiees": "Energy", "comms": "Telecommunications", "telecom": "Telecommunications", "Telco": "Telecommunications", "Coms": "Telecommunications", "Software": "Technology", "energy": "Energy", "oil & gas": "Energy", "Oil & Gas": "Energy", "tech": "Technology", "TELCO": "Telecommunications", "ENERGY": "Energy", "Tech": "Technology", "Comms": "Telecommunications", "telco": "Telecommunications", "TELECOM": "Telecommunications", "SOFTWARE": "Technology", "Telecom": "Telecommunications", "Enerwy": "Energy", "TCEH": "Technology", "telecommunications": "Telecommunications", "Ccmms": "Telecommunications"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dir.": "Director", "Admin Assistant": "Administrative Assistant", "Engineering Mgr": "Engineering Manager", "eng manager": "Engineering Manager", "Dir": "Director", "Admin": "Administrative Assistant", "prod mgr": "Product Manager", "dir": "Director", "Eng Manager": "Engineering Manager", "prod manager": "Product Manager", "Prod Manager": "Product Manager", "administrative assistant": "Administrative Assistant", "eng mgr": "Engineering Manager", "Admin Asst": "Administrative Assistant", "DIR": "Director", "engineering manager": "Engineering Manager", "admin asst": "Administrative Assistant", "Prod Mgr": "Product Manager"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"liter": "L", "l": "L", "liters": "L", "lbs": "lb", "kilograms": "kg", "Ponds": "lb", "Lb": "lb", "Kg": "kg", "litre": "L", "killogram": "kg", "LITTRE": "L", "kilogram": "kg", "Lbs": "lb", "Liter": "L", "kilograams": "kg"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"de meern": "De Meern", "NOVA ANDRADINA": "Nova Andradina", "nova andradina": "Nova Andradina", "AFIF": "Afif", "puerto ayacucho": "Puerto Ayacucho", "AONLA": "Aonla", "afif": "Afif", "Puerto Ayacxcho": "Puerto Ayacucho", "De Meeern": "De Meern", "aonla": "Aonla", "NOVA ANDFADINA": "Nova Andradina"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P2": "Medium", "medium": "Medium", "L": "Low", "LOW": "Low", "low": "Low", "p3": "Low", "MED": "Medium", "p2": "Medium", "med": "Medium", "P3": "Low", "m": "Medium", "mediuum": "Medium", "Med": "Medium", "l": "Low", "M": "Medium"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"email": "bnfn@test.org", "company": "Initech   ", "country": "MSR", "job_title": "   Administrative Assihtant"}, {"email": "CPHG@TEST.ORG", "company": "Globex ", "country": "LIECHTENSTEIN", "job_title": "CEO"}, {"email": " JGBLM@TEST.ORG", "company": " Globex", "country": "NIUE", "job_title": "Account  Executive"}, {"email": "EPFM@TEST.ORG", "company": "Wonka Co", "country": "Mne  ", "job_title": "ae"}, {"email": "gjohg@test.org", "company": "Acme  Inc", "country": "nu", "job_title": "enj manager"}, {"email": " cdggd@test.org ", "company": "  Cyberdyne", "country": "Principality Of Liechtenstein", "job_title": "Admin Assistant"}, {"email": "GNCDPML@EXAMPLE.COM", "company": "  Acme Inc", "country": "nf", "job_title": "AE"}, {"email": "nmld@corp.io", "company": "Globex", "country": "niue", "job_title": "   engineering mgr"}, {"email": "GKANNL@TEST.ORG   ", "company": " Umbrella", "country": "   nfk", "job_title": "C.E.O.  "}, {"email": "bhach@corp.io", "company": " Wonka Co", "country": "  LIE", "job_title": "engineering maiager"}, {"email": "kdgp@example.com", "company": " Umbrella ", "country": "NFK", "job_title": "ENGINEERING MANAGER"}, {"email": "aojg@example.com", "company": "Soylent Corp", "country": "MSR", "job_title": "Eng Mgr"}, {"email": " kodh@test.org ", "company": "Wonka Co", "country": "MSR", "job_title": "  Admin Assistant"}, {"email": "odkaje@corp.io   ", "company": "Initech", "country": "Norfolk Island", "job_title": "chief executive  officer"}, {"email": "khhf@test.org", "company": "Soylent Corp", "country": "   Liechtenstein", "job_title": "Chief Executive Officer"}, {"email": "   gdpb@test.org", "company": "Umbrella ", "country": "ME", "job_title": "Engineering Managre"}, {"email": "COFAGLF@TEST.ORG", "company": "  Globex", "country": "MSR", "job_title": "acct exec"}, {"email": "GJHPJLN@EXAMPLE.COM", "company": "Globex", "country": "   ME", "job_title": "Admin Asst"}, {"email": " AOAJM@TEST.ORG ", "company": "Umbrella", "country": "MSR", "job_title": "Ceo"}], "clean": [{"email": "bnfn@test.org", "company": "Initech", "country": "Montserrat", "job_title": "Administrative Assistant"}, {"email": "cphg@test.org", "company": "Globex", "country": "Liechtenstein", "job_title": "Chief Executive Officer"}, {"email": "jgblm@test.org", "company": "Globex", "country": "Niue", "job_title": "Account Executive"}, {"email": "epfm@test.org", "company": "Wonka Co", "country": "Montenegro", "job_title": "Account Executive"}, {"email": "gjohg@test.org", "company": "Acme Inc", "country": "Niue", "job_title": "Engineering Manager"}, {"email": "cdggd@test.org", "company": "Cyberdyne", "country": "Liechtenstein", "job_title": "Administrative Assistant"}, {"email": "gncdpml@example.com", "company": "Acme Inc", "country": "Norfolk Island", "job_title": "Account Executive"}, {"email": "nmld@corp.io", "company": "Globex", "country": "Niue", "job_title": "Engineering Manager"}, {"email": "gkannl@test.org", "company": "Umbrella", "country": "Norfolk Island", "job_title": "Chief Executive Officer"}, {"email": "bhach@corp.io", "company": "Wonka Co", "country": "Liechtenstein", "job_title": "Engineering Manager"}, {"email": "kdgp@example.com", "company": "Umbrella", "country": "Norfolk Island", "job_title": "Engineering Manager"}, {"email": "aojg@example.com", "company": "Soylent Corp", "country": "Montserrat", "job_title": "Engineering Manager"}, {"email": "kodh@test.org", "company": "Wonka Co", "country": "Montserrat", "job_title": "Administrative Assistant"}, {"email": "odkaje@corp.io", "company": "Initech", "country": "Norfolk Island", "job_title": "Chief Executive Officer"}, {"email": "khhf@test.org", "company": "Soylent Corp", "country": "Liechtenstein", "job_title": "Chief Executive Officer"}, {"email": "gdpb@test.org", "company": "Umbrella", "country": "Montenegro", "job_title": "Engineering Manager"}, {"email": "cofaglf@test.org", "company": "Globex", "country": "Montserrat", "job_title": "Account Executive"}, {"email": "gjhpjln@example.com", "company": "Globex", "country": "Montenegro", "job_title": "Administrative Assistant"}, {"email": "aoajm@test.org", "company": "Umbrella", "country": "Montserrat", "job_title": "Chief Executive Officer"}], "dirty_cols": ["email", "company", "country", "job_title"], "clean_cols": ["email", "company", "country", "job_title"], "plan": {"dataset_summary": "19 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MSR": "Montserrat", "LIECHTENSTEIN": "Liechtenstein", "NIUE": "Niue", "Mne": "Montenegro", "nu": "Niue", "Principality Of Liechtenstein": "Liechtenstein", "nf": "Norfolk Island", "niue": "Niue", "nfk": "Norfolk Island", "LIE": "Liechtenstein", "NFK": "Norfolk Island", "ME": "Montenegro"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Administrative Assihtant": "Administrative Assistant", "CEO": "Chief Executive Officer", "ae": "Account Executive", "enj manager": "Engineering Manager", "Admin Assistant": "Administrative Assistant", "AE": "Account Executive", "engineering mgr": "Engineering Manager", "C.E.O.": "Chief Executive Officer", "engineering maiager": "Engineering Manager", "ENGINEERING MANAGER": "Engineering Manager", "Eng Mgr": "Engineering Manager", "chief executive officer": "Chief Executive Officer", "Engineering Managre": "Engineering Manager", "acct exec": "Account Executive", "Admin Asst": "Administrative Assistant", "Ceo": "Chief Executive Officer"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": " admin ", "email": "HIJJC@CORP.IO", "is_active": "false", "phone": "(506)665-2404", "rate": "64.7%", "city": "oral", "extra": ""}, {"job_title": "vice pres", "email": "LPFG@MAIL.COM", "is_active": "Yes", "phone": "5446238456", "rate": "93.8%", "city": "Oral  ", "extra": ""}, {"job_title": "VP ", "email": "midgo@test.org", "is_active": "false", "phone": "5958933783", "rate": "6.6%", "city": "Summerville", "extra": ""}, {"job_title": "Dir.", "email": "EMJJJDP@CORP.IO", "is_active": "TRUE", "phone": "(571)255-2507", "rate": "33.6%", "city": " ORAL ", "extra": ""}, {"job_title": "CEO", "email": "MGOAEBO@MAIL.COM", "is_active": "FALSE", "phone": "596.557.9599", "rate": "40.6%", "city": "Summerville", "extra": ""}, {"job_title": "C.E.O.", "email": "hjajdgc@corp.io   ", "is_active": "false", "phone": "5414158985", "rate": "91.2%", "city": " Summecville ", "extra": ""}, {"job_title": "Vice Pres", "email": "gmhnoj@test.org", "is_active": "Y", "phone": "(584)597-5356", "rate": "40.1%", "city": "summerville", "extra": ""}, {"job_title": "V.P.", "email": "dkmj@corp.io", "is_active": "F", "phone": "500-305-3833", "rate": "91.5%", "city": " Oral", "extra": ""}, {"job_title": " Director ", "email": "AKIO@MAIL.COM", "is_active": "T", "phone": "(571)424-9876", "rate": "22.2%", "city": "Oral", "extra": ""}, {"job_title": " administrative assistant", "email": "lkai@example.com", "is_active": "true", "phone": "5380283709", "rate": "81.6%", "city": "trendola-ducenta", "extra": ""}, {"job_title": " VP", "email": "IMLPF@CORP.IO", "is_active": "No", "phone": "505.867.4566", "rate": "22.2%", "city": "summerville", "extra": ""}, {"job_title": "Vice  Pres", "email": "LEIKN@EXAMPLE.COM", "is_active": "T", "phone": "(579)627-9959", "rate": "33.5%", "city": " Summerville ", "extra": ""}, {"job_title": "admin assistant", "email": "CFKBAC@CORP.IO", "is_active": "TRUE", "phone": "556.331.0162", "rate": "13.3%", "city": "summerville", "extra": ""}, {"job_title": "", "email": "", "is_active": "", "phone": "", "rate": "", "city": "", "extra": ""}], "clean": [{"job_title": "Administrative Assistant", "email": "hijjc@corp.io", "is_active": false, "phone": "(506) 665-2404", "rate": 0.647, "city": "Oral"}, {"job_title": "Vice President", "email": "lpfg@mail.com", "is_active": true, "phone": "(544) 623-8456", "rate": 0.938, "city": "Oral"}, {"job_title": "Vice President", "email": "midgo@test.org", "is_active": false, "phone": "(595) 893-3783", "rate": 0.066, "city": "Summerville"}, {"job_title": "Director", "email": "emjjjdp@corp.io", "is_active": true, "phone": "(571) 255-2507", "rate": 0.336, "city": "Oral"}, {"job_title": "Chief Executive Officer", "email": "mgoaebo@mail.com", "is_active": false, "phone": "(596) 557-9599", "rate": 0.406, "city": "Summerville"}, {"job_title": "Chief Executive Officer", "email": "hjajdgc@corp.io", "is_active": false, "phone": "(541) 415-8985", "rate": 0.912, "city": "Summerville"}, {"job_title": "Vice President", "email": "gmhnoj@test.org", "is_active": true, "phone": "(584) 597-5356", "rate": 0.401, "city": "Summerville"}, {"job_title": "Vice President", "email": "dkmj@corp.io", "is_active": false, "phone": "(500) 305-3833", "rate": 0.915, "city": "Oral"}, {"job_title": "Director", "email": "akio@mail.com", "is_active": true, "phone": "(571) 424-9876", "rate": 0.222, "city": "Oral"}, {"job_title": "Administrative Assistant", "email": "lkai@example.com", "is_active": true, "phone": "(538) 028-3709", "rate": 0.816, "city": "Trentola-Ducenta"}, {"job_title": "Vice President", "email": "imlpf@corp.io", "is_active": false, "phone": "(505) 867-4566", "rate": 0.222, "city": "Summerville"}, {"job_title": "Vice President", "email": "leikn@example.com", "is_active": true, "phone": "(579) 627-9959", "rate": 0.335, "city": "Summerville"}, {"job_title": "Administrative Assistant", "email": "cfkbac@corp.io", "is_active": true, "phone": "(556) 331-0162", "rate": 0.133, "city": "Summerville"}], "dirty_cols": ["job_title", "email", "is_active", "phone", "rate", "city", "extra"], "clean_cols": ["job_title", "email", "is_active", "phone", "rate", "city"], "plan": {"dataset_summary": "14 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"admin": "Administrative Assistant", "vice pres": "Vice President", "VP": "Vice President", "Dir.": "Director", "CEO": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer", "Vice Pres": "Vice President", "V.P.": "Vice President", "administrative assistant": "Administrative Assistant", "admin assistant": "Administrative Assistant"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"oral": "Oral", "ORAL": "Oral", "Summecville": "Summerville", "summerville": "Summerville", "trendola-ducenta": "Trentola-Ducenta"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "2023-07-16", "city": "kannapuram", "phone": "(572)450-0687", "is_active": "Yes", "amount": "5.929,18", "email": "   OIDM@MAIL.COM"}, {"signup_date": "26 Jun 2023", "city": "  Kannapuram", "phone": "521.230.9691", "is_active": "false", "amount": "1.443,16", "email": "gabp@example.com"}, {"signup_date": "45041", "city": "Puerto Cabello", "phone": "5637175972", "is_active": "1", "amount": "1.673,52", "email": " DIDG@CORP.IO "}, {"signup_date": "44966", "city": "PUERTO CABELLO", "phone": "(547)426-2999", "is_active": "Yes", "amount": "2.049,45", "email": "klceab@mail.com"}, {"signup_date": "2023-03-24", "city": "Balqahs ", "phone": "(573)513-0156", "is_active": "FALSE", "amount": "$2,421.40", "email": "ggpgmpc@test.org"}, {"signup_date": "4/15/2023", "city": "kannapuram", "phone": "(514)937-9212", "is_active": "0", "amount": "8.292,73", "email": "NHKG@CORP.IO"}, {"signup_date": "2023-08-22", "city": "denver", "phone": "(594)418-4279", "is_active": "F", "amount": "494,88", "email": "  kohkea@mail.com"}, {"signup_date": "4/26/2023", "city": "  Puerto Cabello", "phone": "5887258742", "is_active": "Y", "amount": "$5,908.80", "email": " ECENKNL@EXAMPLE.COM"}, {"signup_date": "45154", "city": "Balqash", "phone": "517.871.2030", "is_active": "Yes", "amount": "$3,158.67", "email": "EDEIM@CORP.IO"}, {"signup_date": "3/12/2023", "city": "kannapuram", "phone": "534-490-7218", "is_active": "N", "amount": "3.740,96", "email": "  ilpj@mail.com"}, {"signup_date": "2023-02-08", "city": "Ynacancha", "phone": "551.473.2680", "is_active": "Y", "amount": "$7,580.00", "email": "pcchg@mail.com"}, {"signup_date": "2023-05-09", "city": "Kannapuram  ", "phone": "(567)174-1176", "is_active": "T", "amount": "-100", "email": "PJLN@MAIL.COM   "}, {"signup_date": "7 May 2023", "city": "Puerto Cabelllo", "phone": "503.732.7607", "is_active": "FALSE", "amount": "3.074,81", "email": "DFEMDPP@MAIL.COM"}, {"signup_date": "2023-08-02", "city": "Kannapuram", "phone": "551-293-2546", "is_active": "N", "amount": "$1,478.87", "email": "mnkl@test.org"}, {"signup_date": "28 Apr 2023", "city": "Puerto Cabello", "phone": "543-956-7592", "is_active": "Yes", "amount": "6.380,92", "email": "ahbfnnc@test.org"}, {"signup_date": "18 Nov 2023", "city": " denver ", "phone": "5693543322", "is_active": "TRUE", "amount": "$3,105.72", "email": "BLOKHH@EXAMPLE.COM"}, {"signup_date": "22 Feb 2023", "city": "Puerto  Cabello", "phone": "(589)127-5625", "is_active": "No", "amount": "$4,273.43", "email": "OFCF@MAIL.COM"}, {"signup_date": "3/1/2023", "city": " Puerto Cabello", "phone": "5692188629", "is_active": "F", "amount": "$2,941.26", "email": "BGGHDO@EXAMPLE.COM "}, {"signup_date": "2023-03-22", "city": "  Kannapuram", "phone": "530.203.8804", "is_active": "N", "amount": "$5,451.23", "email": "nkgbikf@corp.io"}, {"signup_date": "7/5/2023", "city": "DCNVER", "phone": "528.954.6459", "is_active": "Y", "amount": "$4,761.74", "email": "NILLEO@EXAMPLE.COM"}, {"signup_date": "9/11/2023", "city": " Denvr ", "phone": "5341217333", "is_active": "1", "amount": "5.158,02", "email": "IHFKBBC@EXAMPLE.COM"}, {"signup_date": "10/9/2023", "city": "Puerto Cabello", "phone": "(551)989-5207", "is_active": "N", "amount": "None", "email": "fidmdp@test.org"}, {"signup_date": "44941", "city": "Puerto Cabello", "phone": "(567)672-4620", "is_active": "No", "amount": "$5,272.40", "email": "gfmfl@corp.io"}, {"signup_date": "45187", "city": "puerto  cabello", "phone": "568-363-8654", "is_active": "N", "amount": "$6,517.07", "email": "IPCLDM@MAIL.COM "}, {"signup_date": "45049", "city": "Kannapuram", "phone": "(519)249-0942", "is_active": "No", "amount": "$7,797.94", "email": "ODLDIG@MAIL.COM"}, {"signup_date": "1/23/2023", "city": "Kannapuram", "phone": "562.800.9526", "is_active": "F", "amount": "1.810,29", "email": "BBADGJH@MAIL.COM"}, {"signup_date": "2023-12-04", "city": "KANNAPURAM", "phone": "586.901.0568", "is_active": "true", "amount": "3.493,61", "email": " CCLCEK@EXAMPLE.COM"}, {"signup_date": "5/23/2023", "city": "Denver  ", "phone": "5152973529", "is_active": "Y", "amount": "456,62", "email": "EALMJPN@CORP.IO"}, {"signup_date": "2/4/2023", "city": "Puerto Cabelo", "phone": "561-118-8600", "is_active": "No", "amount": "5.967,63", "email": "MLJHAC@EXAMPLE.COM"}, {"signup_date": "2023-05-08", "city": "YANACANCHA", "phone": "542-899-2677", "is_active": "false", "amount": "900,98", "email": "CFDED@CORP.IO"}, {"signup_date": "7/16/2023", "city": "Denver", "phone": "512.207.9048", "is_active": "TRUE", "amount": "6.098,84", "email": "jlcmp@mail.com"}, {"signup_date": "45200", "city": " Yanacancha ", "phone": "539.205.9763", "is_active": "TRUE", "amount": "2.659,30", "email": "OMGLFH@TEST.ORG  "}, {"signup_date": "", "city": "", "phone": "", "is_active": "", "amount": "", "email": ""}], "clean": [{"signup_date": "2023-07-16", "city": "Kannapuram", "phone": "(572) 450-0687", "is_active": true, "amount": 5929.18, "email": "oidm@mail.com"}, {"signup_date": "2023-06-26", "city": "Kannapuram", "phone": "(521) 230-9691", "is_active": false, "amount": 1443.16, "email": "gabp@example.com"}, {"signup_date": "2023-04-25", "city": "Puerto Cabello", "phone": "(563) 717-5972", "is_active": true, "amount": 1673.52, "email": "didg@corp.io"}, {"signup_date": "2023-02-09", "city": "Puerto Cabello", "phone": "(547) 426-2999", "is_active": true, "amount": 2049.45, "email": "klceab@mail.com"}, {"signup_date": "2023-03-24", "city": "Balqash", "phone": "(573) 513-0156", "is_active": false, "amount": 2421.4, "email": "ggpgmpc@test.org"}, {"signup_date": "2023-04-15", "city": "Kannapuram", "phone": "(514) 937-9212", "is_active": false, "amount": 8292.73, "email": "nhkg@corp.io"}, {"signup_date": "2023-08-22", "city": "Denver", "phone": "(594) 418-4279", "is_active": false, "amount": 494.88, "email": "kohkea@mail.com"}, {"signup_date": "2023-04-26", "city": "Puerto Cabello", "phone": "(588) 725-8742", "is_active": true, "amount": 5908.8, "email": "ecenknl@example.com"}, {"signup_date": "2023-08-16", "city": "Balqash", "phone": "(517) 871-2030", "is_active": true, "amount": 3158.67, "email": "edeim@corp.io"}, {"signup_date": "2023-03-12", "city": "Kannapuram", "phone": "(534) 490-7218", "is_active": false, "amount": 3740.96, "email": "ilpj@mail.com"}, {"signup_date": "2023-02-08", "city": "Yanacancha", "phone": "(551) 473-2680", "is_active": true, "amount": 7580.0, "email": "pcchg@mail.com"}, {"signup_date": "2023-05-09", "city": "Kannapuram", "phone": "(567) 174-1176", "is_active": true, "amount": -100.0, "email": "pjln@mail.com"}, {"signup_date": "2023-05-07", "city": "Puerto Cabello", "phone": "(503) 732-7607", "is_active": false, "amount": 3074.81, "email": "dfemdpp@mail.com"}, {"signup_date": "2023-08-02", "city": "Kannapuram", "phone": "(551) 293-2546", "is_active": false, "amount": 1478.87, "email": "mnkl@test.org"}, {"signup_date": "2023-04-28", "city": "Puerto Cabello", "phone": "(543) 956-7592", "is_active": true, "amount": 6380.92, "email": "ahbfnnc@test.org"}, {"signup_date": "2023-11-18", "city": "Denver", "phone": "(569) 354-3322", "is_active": true, "amount": 3105.72, "email": "blokhh@example.com"}, {"signup_date": "2023-02-22", "city": "Puerto Cabello", "phone": "(589) 127-5625", "is_active": false, "amount": 4273.43, "email": "ofcf@mail.com"}, {"signup_date": "2023-03-01", "city": "Puerto Cabello", "phone": "(569) 218-8629", "is_active": false, "amount": 2941.26, "email": "bgghdo@example.com"}, {"signup_date": "2023-03-22", "city": "Kannapuram", "phone": "(530) 203-8804", "is_active": false, "amount": 5451.23, "email": "nkgbikf@corp.io"}, {"signup_date": "2023-07-05", "city": "Denver", "phone": "(528) 954-6459", "is_active": true, "amount": 4761.74, "email": "nilleo@example.com"}, {"signup_date": "2023-09-11", "city": "Denver", "phone": "(534) 121-7333", "is_active": true, "amount": 5158.02, "email": "ihfkbbc@example.com"}, {"signup_date": "2023-10-09", "city": "Puerto Cabello", "phone": "(551) 989-5207", "is_active": false, "amount": NaN, "email": "fidmdp@test.org"}, {"signup_date": "2023-01-15", "city": "Puerto Cabello", "phone": "(567) 672-4620", "is_active": false, "amount": 5272.4, "email": "gfmfl@corp.io"}, {"signup_date": "2023-09-18", "city": "Puerto Cabello", "phone": "(568) 363-8654", "is_active": false, "amount": 6517.07, "email": "ipcldm@mail.com"}, {"signup_date": "2023-05-03", "city": "Kannapuram", "phone": "(519) 249-0942", "is_active": false, "amount": 7797.94, "email": "odldig@mail.com"}, {"signup_date": "2023-01-23", "city": "Kannapuram", "phone": "(562) 800-9526", "is_active": false, "amount": 1810.29, "email": "bbadgjh@mail.com"}, {"signup_date": "2023-12-04", "city": "Kannapuram", "phone": "(586) 901-0568", "is_active": true, "amount": 3493.61, "email": "cclcek@example.com"}, {"signup_date": "2023-05-23", "city": "Denver", "phone": "(515) 297-3529", "is_active": true, "amount": 456.62, "email": "ealmjpn@corp.io"}, {"signup_date": "2023-02-04", "city": "Puerto Cabello", "phone": "(561) 118-8600", "is_active": false, "amount": 5967.63, "email": "mljhac@example.com"}, {"signup_date": "2023-05-08", "city": "Yanacancha", "phone": "(542) 899-2677", "is_active": false, "amount": 900.98, "email": "cfded@corp.io"}, {"signup_date": "2023-07-16", "city": "Denver", "phone": "(512) 207-9048", "is_active": true, "amount": 6098.84, "email": "jlcmp@mail.com"}, {"signup_date": "2023-10-01", "city": "Yanacancha", "phone": "(539) 205-9763", "is_active": true, "amount": 2659.3, "email": "omglfh@test.org"}], "dirty_cols": ["signup_date", "city", "phone", "is_active", "amount", "email"], "clean_cols": ["signup_date", "city", "phone", "is_active", "amount", "email"], "plan": {"dataset_summary": "33 rows × 6 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kannapuram": "Kannapuram", "PUERTO CABELLO": "Puerto Cabello", "Balqahs": "Balqash", "denver": "Denver", "Ynacancha": "Yanacancha", "Puerto Cabelllo": "Puerto Cabello", "DCNVER": "Denver", "Denvr": "Denver", "puerto cabello": "Puerto Cabello", "KANNAPURAM": "Kannapuram", "Puerto Cabelo": "Puerto Cabello", "YANACANCHA": "Yanacancha"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"unit": "mls", "is_active": "TRUE", "name": "Heidi Reyes", "country": "STATE OF ISRAEL", "amount": "5.735,59", "notes2": ""}, {"unit": "kilometer", "is_active": "F", "name": "Karl Lee", "country": "isr", "amount": "994,11", "notes2": ""}, {"unit": "   km", "is_active": "false", "name": "David  Moore", "country": "repbulic of latvia ", "amount": "3.718,52", "notes2": ""}, {"unit": "Milliliter", "is_active": "Y", "name": "Eve Johnson", "country": "tajikistan", "amount": "5.019,95", "notes2": ""}, {"unit": "KILOMETERS", "is_active": "false", "name": "Heidi Petrov", "country": "Kazakhstan", "amount": "4.570,11", "notes2": ""}, {"unit": "ml", "is_active": "No", "name": "Grace Ali", "country": "kazakhstan", "amount": "null", "notes2": ""}, {"unit": "milliliter", "is_active": "No", "name": " Judy Fischer", "country": "STATE OF ISRAEL", "amount": "$6,446.05", "notes2": ""}, {"unit": "milliliter", "is_active": "0", "name": "Alice Moore ", "country": " FI", "amount": "9999999", "notes2": ""}, {"unit": "KILOMETER", "is_active": "false", "name": "Bob Moore", "country": "Republic of Finland", "amount": "3.549,02", "notes2": ""}, {"unit": "Ml", "is_active": "F", "name": "Ivan Khan", "country": "finland", "amount": "None", "notes2": ""}, {"unit": "km ", "is_active": "T", "name": "Judy Khan", "country": "Latnia", "amount": "3.871,04", "notes2": ""}, {"unit": "ml", "is_active": "Y", "name": "Alice Diaz ", "country": " Israel ", "amount": "4.715,23", "notes2": ""}, {"unit": "   km", "is_active": "false", "name": "David  Moore", "country": "repbulic of latvia ", "amount": "3.718,52", "notes2": ""}], "clean": [{"unit": "mL", "is_active": true, "name": "Heidi Reyes", "country": "Israel", "amount": 5735.59}, {"unit": "km", "is_active": false, "name": "Karl Lee", "country": "Israel", "amount": 994.11}, {"unit": "km", "is_active": false, "name": "David Moore", "country": "Latvia", "amount": 3718.52}, {"unit": "mL", "is_active": true, "name": "Eve Johnson", "country": "Tajikistan", "amount": 5019.95}, {"unit": "km", "is_active": false, "name": "Heidi Petrov", "country": "Kazakhstan", "amount": 4570.11}, {"unit": "mL", "is_active": false, "name": "Grace Ali", "country": "Kazakhstan", "amount": NaN}, {"unit": "mL", "is_active": false, "name": "Judy Fischer", "country": "Israel", "amount": 6446.05}, {"unit": "mL", "is_active": false, "name": "Alice Moore", "country": "Finland", "amount": 9999999.0}, {"unit": "km", "is_active": false, "name": "Bob Moore", "country": "Finland", "amount": 3549.02}, {"unit": "mL", "is_active": false, "name": "Ivan Khan", "country": "Finland", "amount": NaN}, {"unit": "km", "is_active": true, "name": "Judy Khan", "country": "Latvia", "amount": 3871.04}, {"unit": "mL", "is_active": true, "name": "Alice Diaz", "country": "Israel", "amount": 4715.23}], "dirty_cols": ["unit", "is_active", "name", "country", "amount", "notes2"], "clean_cols": ["unit", "is_active", "name", "country", "amount"], "plan": {"dataset_summary": "13 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"mls": "mL", "kilometer": "km", "Milliliter": "mL", "KILOMETERS": "km", "ml": "mL", "milliliter": "mL", "KILOMETER": "km", "Ml": "mL"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"STATE OF ISRAEL": "Israel", "isr": "Israel", "repbulic of latvia": "Latvia", "tajikistan": "Tajikistan", "kazakhstan": "Kazakhstan", "FI": "Finland", "Republic of Finland": "Finland", "finland": "Finland", "Latnia": "Latvia"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value 9999999 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"currency": "AUSTRALIAN DOLLAR", "status": "Active", "email": "BFIMOO@EXAMPLE.COM", "name": "Judy Cruz  "}, {"currency": "colombian peso", "status": "  Churned", "email": "ELDLMC@CORP.IO ", "name": "Ivan Adams"}, {"currency": "australian dollar", "status": "atcive  ", "email": "ioepcmi@corp.io", "name": "  Ivan Petrov"}, {"currency": "gel", "status": "Active   ", "email": "PJKJB@CORP.IO", "name": "Bob Reyes"}, {"currency": "GEL", "status": "trial", "email": "  cehnplb@corp.io", "name": "Lena Novak"}, {"currency": "Colombian Peso", "status": "Churned", "email": "LLBLNE@EXAMPLE.COM", "name": "  Judy Novak"}, {"currency": "GEL", "status": " paused ", "email": " GFBAHBJ@CORP.IO", "name": "Heidi Brandt"}, {"currency": "GEL", "status": "actire ", "email": " JKCHOL@EXAMPLE.COM ", "name": "Sara Cruz  "}, {"currency": " aud ", "status": "ACTIVVE", "email": "mihpiel@corp.io   ", "name": "   Eve Petrov"}, {"currency": "AUD", "status": "Paused", "email": " hdmphe@mail.com", "name": "Carol Johnson"}, {"currency": "Lari", "status": "Paused", "email": "dfml@test.org", "name": "Judy Lee"}, {"currency": " Lari", "status": "Trial", "email": "ABEAMEJ@EXAMPLE.COM", "name": "Omar  Johnson"}, {"currency": "AUSTRLAIAN DOLLAR", "status": "active", "email": "MPDLBJ@CORP.IO", "name": "Frank  Khan"}, {"currency": "Colombdan Peso", "status": "on hold", "email": "jhofbab@test.org", "name": "Mona Park"}, {"currency": "aud", "status": "on  hold", "email": " ochldc@corp.io ", "name": "Carol  Fischer"}, {"currency": "GEL", "status": "trial ", "email": " GBPO@CORP.IO ", "name": "Carol Diaz   "}, {"currency": "Lari", "status": "cancellde", "email": "kjipjo@example.com   ", "name": "Lena  Novak"}, {"currency": "GEL", "status": "trial", "email": "  cehnplb@corp.io", "name": "Lena Novak"}, {"currency": "gel", "status": "Active   ", "email": "PJKJB@CORP.IO", "name": "Bob Reyes"}], "clean": [{"currency": "AUD", "status": "Active", "email": "bfimoo@example.com", "name": "Judy Cruz"}, {"currency": "COP", "status": "Churned", "email": "eldlmc@corp.io", "name": "Ivan Adams"}, {"currency": "AUD", "status": "Active", "email": "ioepcmi@corp.io", "name": "Ivan Petrov"}, {"currency": "GEL", "status": "Active", "email": "pjkjb@corp.io", "name": "Bob Reyes"}, {"currency": "GEL", "status": "Trial", "email": "cehnplb@corp.io", "name": "Lena Novak"}, {"currency": "COP", "status": "Churned", "email": "llblne@example.com", "name": "Judy Novak"}, {"currency": "GEL", "status": "Paused", "email": "gfbahbj@corp.io", "name": "Heidi Brandt"}, {"currency": "GEL", "status": "Active", "email": "jkchol@example.com", "name": "Sara Cruz"}, {"currency": "AUD", "status": "Active", "email": "mihpiel@corp.io", "name": "Eve Petrov"}, {"currency": "AUD", "status": "Paused", "email": "hdmphe@mail.com", "name": "Carol Johnson"}, {"currency": "GEL", "status": "Paused", "email": "dfml@test.org", "name": "Judy Lee"}, {"currency": "GEL", "status": "Trial", "email": "abeamej@example.com", "name": "Omar Johnson"}, {"currency": "AUD", "status": "Active", "email": "mpdlbj@corp.io", "name": "Frank Khan"}, {"currency": "COP", "status": "Paused", "email": "jhofbab@test.org", "name": "Mona Park"}, {"currency": "AUD", "status": "Paused", "email": "ochldc@corp.io", "name": "Carol Fischer"}, {"currency": "GEL", "status": "Trial", "email": "gbpo@corp.io", "name": "Carol Diaz"}, {"currency": "GEL", "status": "Churned", "email": "kjipjo@example.com", "name": "Lena Novak"}], "dirty_cols": ["currency", "status", "email", "name"], "clean_cols": ["currency", "status", "email", "name"], "plan": {"dataset_summary": "19 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"AUSTRALIAN DOLLAR": "AUD", "colombian peso": "COP", "australian dollar": "AUD", "gel": "GEL", "Colombian Peso": "COP", "aud": "AUD", "Lari": "GEL", "AUSTRLAIAN DOLLAR": "AUD", "Colombdan Peso": "COP"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"atcive": "Active", "trial": "Trial", "paused": "Paused", "actire": "Active", "ACTIVVE": "Active", "active": "Active", "on hold": "Paused", "cancellde": "Churned"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"amount": "na", "state": "Indiana", "signup_date": "44952", "unit": "inch", "col_x": ""}, {"amount": "7.439,46", "state": "Massachusetst", "signup_date": "2023-07-27", "unit": " kg ", "col_x": ""}, {"amount": "508,95", "state": "vi ", "signup_date": "45235", "unit": "kg", "col_x": ""}, {"amount": "4.537,42", "state": "Virgin Islands, U.S.", "signup_date": "2023-12-19", "unit": "Kg", "col_x": ""}, {"amount": "4.273,19", "state": "massachuusetts ", "signup_date": "2/24/2023", "unit": "kilometers", "col_x": ""}, {"amount": "$5,408.73", "state": "AL", "signup_date": "11/15/2023", "unit": "kilograms", "col_x": ""}, {"amount": "$1,392.62", "state": "Alabama", "signup_date": "3 Sep 2023", "unit": "mile", "col_x": ""}, {"amount": "3.298,91", "state": "AL   ", "signup_date": "26 Apr 2023", "unit": "mile", "col_x": ""}, {"amount": "$2,954.76", "state": "virgin islands, u.s.", "signup_date": "45247", "unit": "mi", "col_x": ""}, {"amount": "6.562,00", "state": " VI", "signup_date": "3/22/2023", "unit": "Kg ", "col_x": ""}, {"amount": "2.736,54", "state": "   Virgin Islands, U.S.", "signup_date": "11/14/2023", "unit": "miles", "col_x": ""}], "clean": [{"amount": NaN, "state": "Indiana", "signup_date": "2023-01-26", "unit": "in"}, {"amount": 7439.46, "state": "Massachusetts", "signup_date": "2023-07-27", "unit": "kg"}, {"amount": 508.95, "state": "Virgin Islands, U.S.", "signup_date": "2023-11-05", "unit": "kg"}, {"amount": 4537.42, "state": "Virgin Islands, U.S.", "signup_date": "2023-12-19", "unit": "kg"}, {"amount": 4273.19, "state": "Massachusetts", "signup_date": "2023-02-24", "unit": "km"}, {"amount": 5408.73, "state": "Alabama", "signup_date": "2023-11-15", "unit": "kg"}, {"amount": 1392.62, "state": "Alabama", "signup_date": "2023-09-03", "unit": "mi"}, {"amount": 3298.91, "state": "Alabama", "signup_date": "2023-04-26", "unit": "mi"}, {"amount": 2954.76, "state": "Virgin Islands, U.S.", "signup_date": "2023-11-17", "unit": "mi"}, {"amount": 6562.0, "state": "Virgin Islands, U.S.", "signup_date": "2023-03-22", "unit": "kg"}, {"amount": 2736.54, "state": "Virgin Islands, U.S.", "signup_date": "2023-11-14", "unit": "mi"}], "dirty_cols": ["amount", "state", "signup_date", "unit", "col_x"], "clean_cols": ["amount", "state", "signup_date", "unit"], "plan": {"dataset_summary": "11 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Massachusetst": "Massachusetts", "vi": "Virgin Islands, U.S.", "massachuusetts": "Massachusetts", "AL": "Alabama", "virgin islands, u.s.": "Virgin Islands, U.S.", "VI": "Virgin Islands, U.S."}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"inch": "in", "Kg": "kg", "kilometers": "km", "kilograms": "kg", "mile": "mi", "miles": "mi"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"unit": "Cm", "state": "SC", "rate": "48.3%", "country": "Co", "status": "Closed Lost", "department": "Marketing"}, {"unit": "Cm", "state": " GA ", "rate": "97.6%", "country": "republic of zimbabwe", "status": "   Won", "department": "human resources"}, {"unit": "cm", "state": "sc", "rate": "94.6%", "country": "PCN  ", "status": "won", "department": "Marketing"}, {"unit": "Cm", "state": "GA", "rate": "88.3%", "country": "zw", "status": "WON", "department": "FIN"}, {"unit": "centimeter", "state": "sc", "rate": "12.4%", "country": "col", "status": "lost", "department": "People Ops"}, {"unit": "L", "state": "Louisiana", "rate": "28.4%", "country": "Arab Republic of Egypt", "status": "CLOSED LOST", "department": "Mkg"}, {"unit": "  Cm", "state": "LA", "rate": "83.2%", "country": "pn", "status": "Won", "department": "People"}, {"unit": "   l", "state": "la", "rate": "35.7%", "country": " CO ", "status": "Lost", "department": "mkt"}, {"unit": "liters", "state": "GA", "rate": "80.5%", "country": "zw", "status": "new", "department": "peolpe ops"}, {"unit": "centimeters", "state": "south  carolina", "rate": "32.7%", "country": "Pn", "status": " open ", "department": " People "}, {"unit": "cm", "state": "SC", "rate": "87.0%", "country": "ZIMBABWE", "status": "   closed-won", "department": "Finance"}, {"unit": "   centimeter", "state": "La", "rate": "32.8%", "country": "zimbbabwe", "status": "closed lost", "department": "Accounting"}, {"unit": "cm", "state": "GA", "rate": "67.9%", "country": " eg", "status": "Closed Lost", "department": "  mktg"}, {"unit": "Cm", "state": " GA ", "rate": "97.6%", "country": "republic of zimbabwe", "status": "   Won", "department": "human resources"}, {"unit": "", "state": "", "rate": "", "country": "", "status": "", "department": ""}], "clean": [{"unit": "cm", "state": "South Carolina", "rate": 0.483, "country": "Colombia", "status": "Lost", "department": "Marketing"}, {"unit": "cm", "state": "Georgia", "rate": 0.976, "country": "Zimbabwe", "status": "Won", "department": "Human Resources"}, {"unit": "cm", "state": "South Carolina", "rate": 0.946, "country": "Pitcairn", "status": "Won", "department": "Marketing"}, {"unit": "cm", "state": "Georgia", "rate": 0.883, "country": "Zimbabwe", "status": "Won", "department": "Finance"}, {"unit": "cm", "state": "South Carolina", "rate": 0.124, "country": "Colombia", "status": "Lost", "department": "Human Resources"}, {"unit": "L", "state": "Louisiana", "rate": 0.284, "country": "Egypt", "status": "Lost", "department": "Marketing"}, {"unit": "cm", "state": "Louisiana", "rate": 0.8320000000000001, "country": "Pitcairn", "status": "Won", "department": "Human Resources"}, {"unit": "L", "state": "Louisiana", "rate": 0.35700000000000004, "country": "Colombia", "status": "Lost", "department": "Marketing"}, {"unit": "L", "state": "Georgia", "rate": 0.805, "country": "Zimbabwe", "status": "Open", "department": "Human Resources"}, {"unit": "cm", "state": "South Carolina", "rate": 0.327, "country": "Pitcairn", "status": "Open", "department": "Human Resources"}, {"unit": "cm", "state": "South Carolina", "rate": 0.87, "country": "Zimbabwe", "status": "Won", "department": "Finance"}, {"unit": "cm", "state": "Louisiana", "rate": 0.32799999999999996, "country": "Zimbabwe", "status": "Lost", "department": "Finance"}, {"unit": "cm", "state": "Georgia", "rate": 0.679, "country": "Egypt", "status": "Lost", "department": "Marketing"}], "dirty_cols": ["unit", "state", "rate", "country", "status", "department"], "clean_cols": ["unit", "state", "rate", "country", "status", "department"], "plan": {"dataset_summary": "15 rows × 6 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Cm": "cm", "centimeter": "cm", "l": "L", "liters": "L", "centimeters": "cm"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SC": "South Carolina", "GA": "Georgia", "sc": "South Carolina", "LA": "Louisiana", "la": "Louisiana", "south carolina": "South Carolina", "La": "Louisiana"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Co": "Colombia", "republic of zimbabwe": "Zimbabwe", "PCN": "Pitcairn", "zw": "Zimbabwe", "col": "Colombia", "Arab Republic of Egypt": "Egypt", "pn": "Pitcairn", "CO": "Colombia", "Pn": "Pitcairn", "ZIMBABWE": "Zimbabwe", "zimbbabwe": "Zimbabwe", "eg": "Egypt"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Closed Lost": "Lost", "won": "Won", "WON": "Won", "lost": "Lost", "CLOSED LOST": "Lost", "new": "Open", "open": "Open", "closed-won": "Won", "closed lost": "Lost"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"human resources": "Human Resources", "FIN": "Finance", "People Ops": "Human Resources", "Mkg": "Marketing", "People": "Human Resources", "mkt": "Marketing", "peolpe ops": "Human Resources", "Accounting": "Finance", "mktg": "Marketing"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "Convertible Mark", "signup_date": "44992", "company": "Wonka Co"}, {"currency": "  BAM", "signup_date": "2023-06-04", "company": " Hooli "}, {"currency": "Convertible Mark", "signup_date": "45249", "company": "Cyberdyne"}, {"currency": "CAD", "signup_date": "45157", "company": "   Stark Industries"}, {"currency": "cad", "signup_date": "45116", "company": "  Hooli"}, {"currency": "   Convertible Mark", "signup_date": "2023-10-05", "company": "Vehement "}, {"currency": "BAM", "signup_date": "12/15/2023", "company": "Globex"}, {"currency": "bam", "signup_date": "10/12/2023", "company": "Initech"}, {"currency": "convertible mark", "signup_date": "3/5/2023", "company": "Globex"}, {"currency": "cad", "signup_date": "44937", "company": "Globex"}, {"currency": "Canadian Dollar", "signup_date": "9/17/2023", "company": "Stark  Industries"}, {"currency": " cad ", "signup_date": "2023-07-27", "company": "Hooli"}, {"currency": "canadia dollar", "signup_date": "6/24/2023", "company": "Hooli"}, {"currency": "  Bam", "signup_date": "2023-03-09", "company": "Acme Inc"}, {"currency": "CAD", "signup_date": "10/26/2023", "company": "Hooli"}, {"currency": "Convertible Mark", "signup_date": "15 Aug 2023", "company": "Soylent Corp"}, {"currency": "canadian dollar", "signup_date": "45268", "company": "Cyberdyne"}, {"currency": "CAD", "signup_date": "12/24/2023", "company": "Stark Industries"}, {"currency": " cad", "signup_date": "2023-11-27", "company": "Cyberdyne "}, {"currency": "CAD", "signup_date": "45206", "company": " Cyberdyne "}, {"currency": "convertible mark", "signup_date": "17 Feb 2023", "company": "   Hooli"}, {"currency": " CAD ", "signup_date": "45209", "company": " Cyberdyne "}, {"currency": " cad", "signup_date": "26 Jan 2023", "company": "Initech"}, {"currency": "Canadian Dollar", "signup_date": "2/15/2023", "company": "Wonka Co"}, {"currency": "Convertible Mark", "signup_date": "45249", "company": "Cyberdyne"}, {"currency": " CAD ", "signup_date": "45209", "company": " Cyberdyne "}], "clean": [{"currency": "BAM", "signup_date": "2023-03-07", "company": "Wonka Co"}, {"currency": "BAM", "signup_date": "2023-06-04", "company": "Hooli"}, {"currency": "BAM", "signup_date": "2023-11-19", "company": "Cyberdyne"}, {"currency": "CAD", "signup_date": "2023-08-19", "company": "Stark Industries"}, {"currency": "CAD", "signup_date": "2023-07-09", "company": "Hooli"}, {"currency": "BAM", "signup_date": "2023-10-05", "company": "Vehement"}, {"currency": "BAM", "signup_date": "2023-12-15", "company": "Globex"}, {"currency": "BAM", "signup_date": "2023-10-12", "company": "Initech"}, {"currency": "BAM", "signup_date": "2023-03-05", "company": "Globex"}, {"currency": "CAD", "signup_date": "2023-01-11", "company": "Globex"}, {"currency": "CAD", "signup_date": "2023-09-17", "company": "Stark Industries"}, {"currency": "CAD", "signup_date": "2023-07-27", "company": "Hooli"}, {"currency": "CAD", "signup_date": "2023-06-24", "company": "Hooli"}, {"currency": "BAM", "signup_date": "2023-03-09", "company": "Acme Inc"}, {"currency": "CAD", "signup_date": "2023-10-26", "company": "Hooli"}, {"currency": "BAM", "signup_date": "2023-08-15", "company": "Soylent Corp"}, {"currency": "CAD", "signup_date": "2023-12-08", "company": "Cyberdyne"}, {"currency": "CAD", "signup_date": "2023-12-24", "company": "Stark Industries"}, {"currency": "CAD", "signup_date": "2023-11-27", "company": "Cyberdyne"}, {"currency": "CAD", "signup_date": "2023-10-07", "company": "Cyberdyne"}, {"currency": "BAM", "signup_date": "2023-02-17", "company": "Hooli"}, {"currency": "CAD", "signup_date": "2023-10-10", "company": "Cyberdyne"}, {"currency": "CAD", "signup_date": "2023-01-26", "company": "Initech"}, {"currency": "CAD", "signup_date": "2023-02-15", "company": "Wonka Co"}], "dirty_cols": ["currency", "signup_date", "company"], "clean_cols": ["currency", "signup_date", "company"], "plan": {"dataset_summary": "26 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Convertible Mark": "BAM", "cad": "CAD", "bam": "BAM", "convertible mark": "BAM", "Canadian Dollar": "CAD", "canadia dollar": "CAD", "Bam": "BAM", "canadian dollar": "CAD"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"signup_date": "2023-03-28", "currency": "Djibouti Franc", "rate": "65.1%", "unit": "lbs", "industry": "  Telecommunications"}, {"signup_date": "10/2/2023", "currency": "  Dibouti Franc", "rate": "61.2%", "unit": "   centimeters", "industry": "  Telecom"}, {"signup_date": "18 Jul 2023", "currency": " DJF ", "rate": "5.8%", "unit": "KM  ", "industry": "Education"}, {"signup_date": "8 Apr 2023", "currency": "   Boliviano", "rate": "17.3%", "unit": " centimeters ", "industry": "Ed "}, {"signup_date": "15 Nov 2023", "currency": "Boliviano", "rate": "2.3%", "unit": "pounds", "industry": "Comms"}, {"signup_date": "45017", "currency": "djibouti franc", "rate": "27.1%", "unit": " kilometer ", "industry": "EdTech"}, {"signup_date": "2023-03-04", "currency": "djibouti franc", "rate": "85.4%", "unit": " centimeters ", "industry": " telecom "}, {"signup_date": "45221", "currency": "djibouti franc", "rate": "39.2%", "unit": "pounnds", "industry": "Telecommunications"}, {"signup_date": "2 Jul 2023", "currency": "Bolviiano", "rate": "84.6%", "unit": " Mi ", "industry": "COOMMS"}, {"signup_date": "7 Jul 2023", "currency": "DJIBOUTI FRANC", "rate": "11.3%", "unit": " centimeter ", "industry": "Education"}, {"signup_date": "1/7/2023", "currency": "Djibouti Franc ", "rate": "24.2%", "unit": "mi", "industry": "  telco"}, {"signup_date": "2023-05-15", "currency": " Argentine Peso", "rate": "51.2%", "unit": "mile", "industry": "TELECOM"}, {"signup_date": "2023-06-24", "currency": "DJF", "rate": "9.5%", "unit": "miles", "industry": " Ed "}, {"signup_date": "2023-08-28", "currency": "Djibouti Franc", "rate": "84.4%", "unit": "Mi ", "industry": "EdTech"}, {"signup_date": "45286", "currency": "   DJF", "rate": "38.7%", "unit": "kilometer", "industry": "EDUCATIN"}, {"signup_date": "10/15/2023", "currency": "Boliviano", "rate": "4.3%", "unit": "centimeters", "industry": "TELECOM"}, {"signup_date": "2023-10-23", "currency": "BOB", "rate": "84.7%", "unit": " centimeter", "industry": "Telco  "}, {"signup_date": "2023-02-01", "currency": "BOB", "rate": "0.5%", "unit": "kiometers", "industry": "Telecommunications"}, {"signup_date": "3/10/2023", "currency": "   DJF", "rate": "80.0%", "unit": "kilometers", "industry": " ed "}, {"signup_date": "2023-09-05", "currency": "boliviano", "rate": "12.1%", "unit": " cm", "industry": "Telecommunications"}, {"signup_date": "2023-02-06", "currency": "boliviano", "rate": "15.5%", "unit": "cm", "industry": "Ed"}, {"signup_date": "44992", "currency": "BOB", "rate": "16.9%", "unit": "   KM", "industry": "comms"}, {"signup_date": "45191", "currency": "   boliviano", "rate": "2.3%", "unit": "  Centimeters", "industry": "telecom"}, {"signup_date": "2023-11-26", "currency": "bob", "rate": "74.3%", "unit": "  Cm", "industry": "Telecommunications"}, {"signup_date": "4/12/2023", "currency": "boliviano", "rate": "37.6%", "unit": "Km", "industry": "edtech"}], "clean": [{"signup_date": "2023-03-28", "currency": "DJF", "rate": 0.6509999999999999, "unit": "lb", "industry": "Telecommunications"}, {"signup_date": "2023-10-02", "currency": "DJF", "rate": 0.612, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-07-18", "currency": "DJF", "rate": 0.057999999999999996, "unit": "km", "industry": "Education"}, {"signup_date": "2023-04-08", "currency": "BOB", "rate": 0.17300000000000001, "unit": "cm", "industry": "Education"}, {"signup_date": "2023-11-15", "currency": "BOB", "rate": 0.023, "unit": "lb", "industry": "Telecommunications"}, {"signup_date": "2023-04-01", "currency": "DJF", "rate": 0.271, "unit": "km", "industry": "Education"}, {"signup_date": "2023-03-04", "currency": "DJF", "rate": 0.8540000000000001, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-10-22", "currency": "DJF", "rate": 0.392, "unit": "lb", "industry": "Telecommunications"}, {"signup_date": "2023-07-02", "currency": "BOB", "rate": 0.846, "unit": "mi", "industry": "Telecommunications"}, {"signup_date": "2023-07-07", "currency": "DJF", "rate": 0.113, "unit": "cm", "industry": "Education"}, {"signup_date": "2023-01-07", "currency": "DJF", "rate": 0.242, "unit": "mi", "industry": "Telecommunications"}, {"signup_date": "2023-05-15", "currency": "ARS", "rate": 0.512, "unit": "mi", "industry": "Telecommunications"}, {"signup_date": "2023-06-24", "currency": "DJF", "rate": 0.095, "unit": "mi", "industry": "Education"}, {"signup_date": "2023-08-28", "currency": "DJF", "rate": 0.8440000000000001, "unit": "mi", "industry": "Education"}, {"signup_date": "2023-12-26", "currency": "DJF", "rate": 0.387, "unit": "km", "industry": "Education"}, {"signup_date": "2023-10-15", "currency": "BOB", "rate": 0.043, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-10-23", "currency": "BOB", "rate": 0.847, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-02-01", "currency": "BOB", "rate": 0.005, "unit": "km", "industry": "Telecommunications"}, {"signup_date": "2023-03-10", "currency": "DJF", "rate": 0.8, "unit": "km", "industry": "Education"}, {"signup_date": "2023-09-05", "currency": "BOB", "rate": 0.121, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-02-06", "currency": "BOB", "rate": 0.155, "unit": "cm", "industry": "Education"}, {"signup_date": "2023-03-07", "currency": "BOB", "rate": 0.16899999999999998, "unit": "km", "industry": "Telecommunications"}, {"signup_date": "2023-09-22", "currency": "BOB", "rate": 0.023, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-11-26", "currency": "BOB", "rate": 0.743, "unit": "cm", "industry": "Telecommunications"}, {"signup_date": "2023-04-12", "currency": "BOB", "rate": 0.376, "unit": "km", "industry": "Education"}], "dirty_cols": ["signup_date", "currency", "rate", "unit", "industry"], "clean_cols": ["signup_date", "currency", "rate", "unit", "industry"], "plan": {"dataset_summary": "25 rows × 5 columns. 5 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Djibouti Franc": "DJF", "Dibouti Franc": "DJF", "Boliviano": "BOB", "djibouti franc": "DJF", "Bolviiano": "BOB", "DJIBOUTI FRANC": "DJF", "Argentine Peso": "ARS", "boliviano": "BOB", "bob": "BOB"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"lbs": "lb", "centimeters": "cm", "KM": "km", "pounds": "lb", "kilometer": "km", "pounnds": "lb", "Mi": "mi", "centimeter": "cm", "mile": "mi", "miles": "mi", "kiometers": "km", "kilometers": "km", "Centimeters": "cm", "Cm": "cm", "Km": "km"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Telecom": "Telecommunications", "Ed": "Education", "Comms": "Telecommunications", "EdTech": "Education", "telecom": "Telecommunications", "COOMMS": "Telecommunications", "telco": "Telecommunications", "TELECOM": "Telecommunications", "EDUCATIN": "Education", "Telco": "Telecommunications", "ed": "Education", "comms": "Telecommunications", "edtech": "Education"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "CZECH KORUNA ", "email": "GHDJMFH@CORP.IO", "company": "Wonka Co", "unit": "  miles", "col_x": ""}, {"currency": "bam", "email": "MNGF@MAIL.COM", "company": " Hooli", "unit": "Mi", "col_x": ""}, {"currency": "CZECH KORKNA", "email": "ejeob@corp.io ", "company": "Initech", "unit": "mi", "col_x": ""}, {"currency": "   Guyana Dollar", "email": "HCBBFBL@EXAMPLE.COM", "company": "   Acme Inc", "unit": "   mi", "col_x": ""}, {"currency": "Armenian Dram", "email": "IDMF@CORP.IO", "company": "   Stark Industries", "unit": "mi", "col_x": ""}, {"currency": "  Czech Koruna", "email": "EDJE@TEST.ORG", "company": "  Umbrella", "unit": "kilogram", "col_x": ""}, {"currency": "GYD", "email": "hhiaij@corp.io", "company": " Hooli ", "unit": "KILOGRAM", "col_x": ""}, {"currency": "CZK", "email": "efdaoc@test.org", "company": "   Initech", "unit": "Mile", "col_x": ""}, {"currency": "Convertible Mark ", "email": " CHGOJI@MAIL.COM", "company": "Hooli", "unit": "KG", "col_x": ""}, {"currency": "armenian dram", "email": "IDNF@CORP.IO", "company": "Wonka Co", "unit": "Mi", "col_x": ""}, {"currency": "AMD", "email": "MCIJM@TEST.ORG", "company": "Wonka Co", "unit": "miles", "col_x": ""}, {"currency": "gyd", "email": "BGHCO@CORP.IO", "company": "Vehement", "unit": "mle", "col_x": ""}, {"currency": " bam", "email": "mflkoa@example.com", "company": "Wonka Co", "unit": "mile", "col_x": ""}, {"currency": "GYD", "email": "LGEDCP@EXAMPLE.COM", "company": "   Cyberdyne", "unit": "kilograms", "col_x": ""}, {"currency": "CONVERTIBLE MARK", "email": "DCDPF@TEST.ORG", "company": "Soylent Corp", "unit": "Kg   ", "col_x": ""}, {"currency": "guyana dollar", "email": " icajm@corp.io ", "company": " Vehement", "unit": "Miles", "col_x": ""}, {"currency": "Armenian Dram", "email": "ojplhnm@example.com", "company": "Stark  Industries", "unit": " mi", "col_x": ""}, {"currency": "Armenian Dram", "email": "LLKDID@MAIL.COM", "company": "Cyberdyne", "unit": "mile  ", "col_x": ""}, {"currency": "CONVERTIBLE MARK", "email": "   HGMNG@CORP.IO", "company": "Cyberdyne", "unit": "mile", "col_x": ""}, {"currency": "   guyana dollar", "email": " BFJBF@TEST.ORG", "company": "Wonka  Co", "unit": "mile", "col_x": ""}, {"currency": "Czech Koruna", "email": "gjcc@test.org", "company": "Soylent Corp", "unit": "   kg", "col_x": ""}], "clean": [{"currency": "CZK", "email": "ghdjmfh@corp.io", "company": "Wonka Co", "unit": "mi"}, {"currency": "BAM", "email": "mngf@mail.com", "company": "Hooli", "unit": "mi"}, {"currency": "CZK", "email": "ejeob@corp.io", "company": "Initech", "unit": "mi"}, {"currency": "GYD", "email": "hcbbfbl@example.com", "company": "Acme Inc", "unit": "mi"}, {"currency": "AMD", "email": "idmf@corp.io", "company": "Stark Industries", "unit": "mi"}, {"currency": "CZK", "email": "edje@test.org", "company": "Umbrella", "unit": "kg"}, {"currency": "GYD", "email": "hhiaij@corp.io", "company": "Hooli", "unit": "kg"}, {"currency": "CZK", "email": "efdaoc@test.org", "company": "Initech", "unit": "mi"}, {"currency": "BAM", "email": "chgoji@mail.com", "company": "Hooli", "unit": "kg"}, {"currency": "AMD", "email": "idnf@corp.io", "company": "Wonka Co", "unit": "mi"}, {"currency": "AMD", "email": "mcijm@test.org", "company": "Wonka Co", "unit": "mi"}, {"currency": "GYD", "email": "bghco@corp.io", "company": "Vehement", "unit": "mi"}, {"currency": "BAM", "email": "mflkoa@example.com", "company": "Wonka Co", "unit": "mi"}, {"currency": "GYD", "email": "lgedcp@example.com", "company": "Cyberdyne", "unit": "kg"}, {"currency": "BAM", "email": "dcdpf@test.org", "company": "Soylent Corp", "unit": "kg"}, {"currency": "GYD", "email": "icajm@corp.io", "company": "Vehement", "unit": "mi"}, {"currency": "AMD", "email": "ojplhnm@example.com", "company": "Stark Industries", "unit": "mi"}, {"currency": "AMD", "email": "llkdid@mail.com", "company": "Cyberdyne", "unit": "mi"}, {"currency": "BAM", "email": "hgmng@corp.io", "company": "Cyberdyne", "unit": "mi"}, {"currency": "GYD", "email": "bfjbf@test.org", "company": "Wonka Co", "unit": "mi"}, {"currency": "CZK", "email": "gjcc@test.org", "company": "Soylent Corp", "unit": "kg"}], "dirty_cols": ["currency", "email", "company", "unit", "col_x"], "clean_cols": ["currency", "email", "company", "unit"], "plan": {"dataset_summary": "21 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CZECH KORUNA": "CZK", "bam": "BAM", "CZECH KORKNA": "CZK", "Guyana Dollar": "GYD", "Armenian Dram": "AMD", "Czech Koruna": "CZK", "Convertible Mark": "BAM", "armenian dram": "AMD", "gyd": "GYD", "CONVERTIBLE MARK": "BAM", "guyana dollar": "GYD"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miles": "mi", "Mi": "mi", "kilogram": "kg", "KILOGRAM": "kg", "Mile": "mi", "KG": "kg", "mle": "mi", "mile": "mi", "kilograms": "kg", "Kg": "kg", "Miles": "mi"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "ms", "name": "Omar  Johnson", "industry": "Telecommunications", "currency": "bdt"}, {"state": "Mississiippi", "name": "   Heidi Lee", "industry": "comms", "currency": "  Quetzal"}, {"state": "Mt", "name": "Mona Park  ", "industry": "Healthcare", "currency": "quetzal"}, {"state": "wisconsin", "name": "  Priya Johnson", "industry": "Technology", "currency": "bdt"}, {"state": "Montana", "name": "Judy Cruz", "industry": "tech", "currency": "bahraini dinar"}, {"state": "MT", "name": "Karl Johnson   ", "industry": "Telecommunications", "currency": "GTQ"}, {"state": "MT", "name": "Mona Moore   ", "industry": "ENERGY", "currency": "BND"}, {"state": "Montana", "name": "   Bob Petrov", "industry": "Health Care", "currency": "  BDT"}, {"state": "mississippi", "name": "Alice Cruz  ", "industry": "Healtth", "currency": "Quetzzal"}, {"state": "Montana", "name": "Lena Reyes  ", "industry": "Health", "currency": "Bahraini Dinar"}, {"state": "  MT", "name": "Lena  Ali", "industry": "Telecommunications", "currency": "brunei dollar"}, {"state": "ms", "name": "Lena Johnson", "industry": "energy", "currency": "Bnd"}, {"state": "MS", "name": "Frank  Reyes", "industry": "MEIDCAL", "currency": "   bnd"}, {"state": " Mississippi", "name": "Frank Fischer ", "industry": " health ", "currency": "Quetzal"}, {"state": "Vemont   ", "name": "Ivan  Smith", "industry": "energy", "currency": "BHD"}, {"state": "MT", "name": "Priya Johnson", "industry": "utilities", "currency": "Bhd"}, {"state": "Oregon", "name": "Judy Fischer  ", "industry": "tech", "currency": "taka"}, {"state": "wisconsin", "name": "Frank Park", "industry": "tech", "currency": "BDT"}, {"state": "Oregon", "name": "Heidi  Diaz", "industry": "SOFTWARE", "currency": "BHD   "}, {"state": "wisconsin", "name": "  Priya Johnson", "industry": "Technology", "currency": "bdt"}], "clean": [{"state": "Mississippi", "name": "Omar Johnson", "industry": "Telecommunications", "currency": "BDT"}, {"state": "Mississippi", "name": "Heidi Lee", "industry": "Telecommunications", "currency": "GTQ"}, {"state": "Montana", "name": "Mona Park", "industry": "Healthcare", "currency": "GTQ"}, {"state": "Wisconsin", "name": "Priya Johnson", "industry": "Technology", "currency": "BDT"}, {"state": "Montana", "name": "Judy Cruz", "industry": "Technology", "currency": "BHD"}, {"state": "Montana", "name": "Karl Johnson", "industry": "Telecommunications", "currency": "GTQ"}, {"state": "Montana", "name": "Mona Moore", "industry": "Energy", "currency": "BND"}, {"state": "Montana", "name": "Bob Petrov", "industry": "Healthcare", "currency": "BDT"}, {"state": "Mississippi", "name": "Alice Cruz", "industry": "Healthcare", "currency": "GTQ"}, {"state": "Montana", "name": "Lena Reyes", "industry": "Healthcare", "currency": "BHD"}, {"state": "Montana", "name": "Lena Ali", "industry": "Telecommunications", "currency": "BND"}, {"state": "Mississippi", "name": "Lena Johnson", "industry": "Energy", "currency": "BND"}, {"state": "Mississippi", "name": "Frank Reyes", "industry": "Healthcare", "currency": "BND"}, {"state": "Mississippi", "name": "Frank Fischer", "industry": "Healthcare", "currency": "GTQ"}, {"state": "Vermont", "name": "Ivan Smith", "industry": "Energy", "currency": "BHD"}, {"state": "Montana", "name": "Priya Johnson", "industry": "Energy", "currency": "BHD"}, {"state": "Oregon", "name": "Judy Fischer", "industry": "Technology", "currency": "BDT"}, {"state": "Wisconsin", "name": "Frank Park", "industry": "Technology", "currency": "BDT"}, {"state": "Oregon", "name": "Heidi Diaz", "industry": "Technology", "currency": "BHD"}], "dirty_cols": ["state", "name", "industry", "currency"], "clean_cols": ["state", "name", "industry", "currency"], "plan": {"dataset_summary": "20 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ms": "Mississippi", "Mississiippi": "Mississippi", "Mt": "Montana", "wisconsin": "Wisconsin", "MT": "Montana", "mississippi": "Mississippi", "MS": "Mississippi", "Vemont": "Vermont"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"comms": "Telecommunications", "tech": "Technology", "ENERGY": "Energy", "Health Care": "Healthcare", "Healtth": "Healthcare", "Health": "Healthcare", "energy": "Energy", "MEIDCAL": "Healthcare", "health": "Healthcare", "utilities": "Energy", "SOFTWARE": "Technology"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"bdt": "BDT", "Quetzal": "GTQ", "quetzal": "GTQ", "bahraini dinar": "BHD", "Quetzzal": "GTQ", "Bahraini Dinar": "BHD", "brunei dollar": "BND", "Bnd": "BND", "bnd": "BND", "Bhd": "BHD", "taka": "BDT"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"status": "due", "currency": "crc", "email": "heipkaa@corp.io", "is_active": "No"}, {"status": "PENDING", "currency": "crc", "email": " alame@mail.com ", "is_active": "N"}, {"status": "Late", "currency": "crc", "email": "mggc@mail.com ", "is_active": "TRUE"}, {"status": " pending", "currency": "Azerbaijan Manat", "email": "FMAO@EXAMPLE.COM  ", "is_active": "FALSE"}, {"status": "late", "currency": "Azerbaijan Manat", "email": "kmhlm@mail.com", "is_active": "Yes"}, {"status": "   PAID", "currency": " azn ", "email": "JCFHMI@CORP.IO", "is_active": "Yes"}, {"status": " late", "currency": "Crc", "email": " AHGKFP@EXAMPLE.COM ", "is_active": "FALSE"}, {"status": "paid", "currency": "azerbaijan manat", "email": "AADFK@EXAMPLE.COM   ", "is_active": "false"}, {"status": "overdue", "currency": "AZN", "email": "JPKOE@MAIL.COM", "is_active": "1"}, {"status": "overdue", "currency": "AZN", "email": "  ldneakn@example.com", "is_active": "0"}, {"status": "UNPAID   ", "currency": "Azerbaijan Manat", "email": " PCEENC@EXAMPLE.COM ", "is_active": "false"}, {"status": "paid", "currency": "AZN", "email": "pehmekf@corp.io", "is_active": "Yes"}, {"status": "   PENDING", "currency": "azerbaijan manat", "email": "EOLOGG@CORP.IO   ", "is_active": "true"}, {"status": " Pending ", "currency": "Azerbaijan  Manat", "email": " AILCMH@TEST.ORG ", "is_active": "0"}, {"status": " paid", "currency": "Azerbaijan Manat", "email": "NGGFO@CORP.IO", "is_active": "No"}, {"status": "due", "currency": "Azn", "email": "DOKG@CORP.IO", "is_active": "TRUE"}, {"status": "", "currency": "", "email": "", "is_active": ""}], "clean": [{"status": "Pending", "currency": "CRC", "email": "heipkaa@corp.io", "is_active": false}, {"status": "Pending", "currency": "CRC", "email": "alame@mail.com", "is_active": false}, {"status": "Overdue", "currency": "CRC", "email": "mggc@mail.com", "is_active": true}, {"status": "Pending", "currency": "AZN", "email": "fmao@example.com", "is_active": false}, {"status": "Overdue", "currency": "AZN", "email": "kmhlm@mail.com", "is_active": true}, {"status": "Paid", "currency": "AZN", "email": "jcfhmi@corp.io", "is_active": true}, {"status": "Overdue", "currency": "CRC", "email": "ahgkfp@example.com", "is_active": false}, {"status": "Paid", "currency": "AZN", "email": "aadfk@example.com", "is_active": false}, {"status": "Overdue", "currency": "AZN", "email": "jpkoe@mail.com", "is_active": true}, {"status": "Overdue", "currency": "AZN", "email": "ldneakn@example.com", "is_active": false}, {"status": "Pending", "currency": "AZN", "email": "pceenc@example.com", "is_active": false}, {"status": "Paid", "currency": "AZN", "email": "pehmekf@corp.io", "is_active": true}, {"status": "Pending", "currency": "AZN", "email": "eologg@corp.io", "is_active": true}, {"status": "Pending", "currency": "AZN", "email": "ailcmh@test.org", "is_active": false}, {"status": "Paid", "currency": "AZN", "email": "nggfo@corp.io", "is_active": false}, {"status": "Pending", "currency": "AZN", "email": "dokg@corp.io", "is_active": true}], "dirty_cols": ["status", "currency", "email", "is_active"], "clean_cols": ["status", "currency", "email", "is_active"], "plan": {"dataset_summary": "17 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"due": "Pending", "PENDING": "Pending", "Late": "Overdue", "pending": "Pending", "late": "Overdue", "PAID": "Paid", "paid": "Paid", "overdue": "Overdue", "UNPAID": "Pending"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"crc": "CRC", "Azerbaijan Manat": "AZN", "azn": "AZN", "Crc": "CRC", "azerbaijan manat": "AZN", "Azn": "AZN"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"status": "High", "unit": "in", "name": "Mona Diaz", "department": "  I..", "job_title": "   Dor."}, {"status": "high", "unit": "in", "name": "Omar Johnson", "department": "Information Technology", "job_title": " Director "}, {"status": "low", "unit": "Lb", "name": "Bob Reyes", "department": "I.T.", "job_title": "Dir   "}, {"status": " Low", "unit": "lb", "name": "  Carol Ali", "department": "i.t.", "job_title": "Dir"}, {"status": "L", "unit": "lb", "name": "Ivan Brandt", "department": " it ", "job_title": " dir. "}, {"status": "p3", "unit": "incehs", "name": "Grace Ali", "department": "IT", "job_title": " Dir "}, {"status": "P3", "unit": "IN", "name": "Ivan Novak", "department": "I.T.", "job_title": "Dir"}, {"status": "P1", "unit": "INCHES", "name": "Heidi Reyes", "department": " Prodduct", "job_title": "Dir."}, {"status": "P3   ", "unit": "lbs", "name": "Eve Reyes", "department": "PROD ", "job_title": "senir eng"}, {"status": "P3", "unit": "Lb", "name": "Grace  Brandt", "department": "product", "job_title": "DIRECTOR"}, {"status": " low ", "unit": "   inches", "name": "Carol Fischer   ", "department": "information technology", "job_title": "Senior Eng   "}, {"status": "P1", "unit": "pound  ", "name": "Ivan Johnson", "department": "Product  ", "job_title": "dir"}, {"status": "LOW", "unit": "pounds", "name": "Sara Adams ", "department": "it", "job_title": "SNR ENGINEER"}, {"status": " High", "unit": "inch ", "name": "Omar Smith", "department": "Prod", "job_title": "SR ENGINEER"}, {"status": "LOW", "unit": "\"", "name": "Frank Brandt", "department": "Prod", "job_title": "di.r"}, {"status": "Low", "unit": "\"", "name": "Carol Adams", "department": "I.T.", "job_title": "   Dir."}, {"status": "L", "unit": "Innches", "name": "   Ivan Cruz", "department": "prod", "job_title": "Sr Engineer"}, {"status": "low", "unit": "\"", "name": "Carol Novak", "department": "Information Technology", "job_title": "Sr Engineer"}, {"status": "HIGH", "unit": "\"", "name": "Bob Moore  ", "department": "Prod", "job_title": "sr. engineer"}, {"status": "P1", "unit": "in", "name": "David Diaz", "department": "PRODUCT", "job_title": "Snr Engineer"}, {"status": " h ", "unit": "POUNDS", "name": "Mona Khan", "department": "IT", "job_title": "Dir"}, {"status": "P3", "unit": "Lb", "name": "Grace  Brandt", "department": "product", "job_title": "DIRECTOR"}, {"status": " High", "unit": "inch ", "name": "Omar Smith", "department": "Prod", "job_title": "SR ENGINEER"}], "clean": [{"status": "High", "unit": "in", "name": "Mona Diaz", "department": "Information Technology", "job_title": "Director"}, {"status": "High", "unit": "in", "name": "Omar Johnson", "department": "Information Technology", "job_title": "Director"}, {"status": "Low", "unit": "lb", "name": "Bob Reyes", "department": "Information Technology", "job_title": "Director"}, {"status": "Low", "unit": "lb", "name": "Carol Ali", "department": "Information Technology", "job_title": "Director"}, {"status": "Low", "unit": "lb", "name": "Ivan Brandt", "department": "Information Technology", "job_title": "Director"}, {"status": "Low", "unit": "in", "name": "Grace Ali", "department": "Information Technology", "job_title": "Director"}, {"status": "Low", "unit": "in", "name": "Ivan Novak", "department": "Information Technology", "job_title": "Director"}, {"status": "High", "unit": "in", "name": "Heidi Reyes", "department": "Product", "job_title": "Director"}, {"status": "Low", "unit": "lb", "name": "Eve Reyes", "department": "Product", "job_title": "Senior Engineer"}, {"status": "Low", "unit": "lb", "name": "Grace Brandt", "department": "Product", "job_title": "Director"}, {"status": "Low", "unit": "in", "name": "Carol Fischer", "department": "Information Technology", "job_title": "Senior Engineer"}, {"status": "High", "unit": "lb", "name": "Ivan Johnson", "department": "Product", "job_title": "Director"}, {"status": "Low", "unit": "lb", "name": "Sara Adams", "department": "Information Technology", "job_title": "Senior Engineer"}, {"status": "High", "unit": "in", "name": "Omar Smith", "department": "Product", "job_title": "Senior Engineer"}, {"status": "Low", "unit": "in", "name": "Frank Brandt", "department": "Product", "job_title": "Director"}, {"status": "Low", "unit": "in", "name": "Carol Adams", "department": "Information Technology", "job_title": "Director"}, {"status": "Low", "unit": "in", "name": "Ivan Cruz", "department": "Product", "job_title": "Senior Engineer"}, {"status": "Low", "unit": "in", "name": "Carol Novak", "department": "Information Technology", "job_title": "Senior Engineer"}, {"status": "High", "unit": "in", "name": "Bob Moore", "department": "Product", "job_title": "Senior Engineer"}, {"status": "High", "unit": "in", "name": "David Diaz", "department": "Product", "job_title": "Senior Engineer"}, {"status": "High", "unit": "lb", "name": "Mona Khan", "department": "Information Technology", "job_title": "Director"}], "dirty_cols": ["status", "unit", "name", "department", "job_title"], "clean_cols": ["status", "unit", "name", "department", "job_title"], "plan": {"dataset_summary": "23 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"high": "High", "low": "Low", "L": "Low", "p3": "Low", "P3": "Low", "P1": "High", "LOW": "Low", "HIGH": "High", "h": "High"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Lb": "lb", "incehs": "in", "IN": "in", "INCHES": "in", "lbs": "lb", "inches": "in", "pound": "lb", "pounds": "lb", "inch": "in", "\"": "in", "Innches": "in", "POUNDS": "lb"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"I..": "Information Technology", "I.T.": "Information Technology", "i.t.": "Information Technology", "it": "Information Technology", "IT": "Information Technology", "Prodduct": "Product", "PROD": "Product", "product": "Product", "information technology": "Information Technology", "Prod": "Product", "prod": "Product", "PRODUCT": "Product"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dor.": "Director", "Dir": "Director", "dir.": "Director", "Dir.": "Director", "senir eng": "Senior Engineer", "DIRECTOR": "Director", "Senior Eng": "Senior Engineer", "dir": "Director", "SNR ENGINEER": "Senior Engineer", "SR ENGINEER": "Senior Engineer", "di.r": "Director", "Sr Engineer": "Senior Engineer", "sr. engineer": "Senior Engineer", "Snr Engineer": "Senior Engineer"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"department": " Information Techeology", "phone": "528.275.4291", "status": "trual", "company": " Hooli ", "currency": "dzd  ", "email": "pofbbf@example.com", "rate": "68.6%"}, {"department": "OPERATIONS   ", "phone": "530-368-8945", "status": "paused", "company": "Stark Industries ", "currency": "   ARMENIAN DRAM", "email": " mapgcei@corp.io ", "rate": "87.1%"}, {"department": "ENG.", "phone": "(595)835-2094", "status": "on  hlod", "company": "Soylent Corp", "currency": "DZD   ", "email": "  jcmjfo@test.org", "rate": "23.9%"}, {"department": " Ops ", "phone": "505-953-7233", "status": "trial", "company": "Vehement", "currency": "Algerian Dinar", "email": "   fokjae@mail.com", "rate": "74.4%"}, {"department": " information technology", "phone": "511.913.0543", "status": "free trial", "company": "Umbrella", "currency": "Armenian Dram", "email": " ahmfg@mail.com", "rate": "43.5%"}, {"department": "IT", "phone": "528.240.7612", "status": "paused", "company": "Wonka  Co", "currency": " AMD ", "email": "moihj@example.com", "rate": "14.6%"}, {"department": "INFORMATION TECHNOLOGY", "phone": "525.932.3654", "status": "trial", "company": "   Stark Industries", "currency": "DZD", "email": "LLGAOB@CORP.IO", "rate": "73.0%"}, {"department": "information technology", "phone": "550-280-1222", "status": "trial", "company": "Globex", "currency": "Algerian Dinar", "email": "GDHP@TEST.ORG", "rate": "86.3%"}, {"department": " Dev ", "phone": "579-374-3155", "status": "Active   ", "company": "Vehement", "currency": "Artenian Dram", "email": "LNBGPPC@CORP.IO", "rate": "74.4%"}, {"department": " eng", "phone": "(553)490-6222", "status": "ON HOLD", "company": "Wonka  Co", "currency": "   AMD", "email": "jahbffc@test.org", "rate": "53.2%"}, {"department": "ops", "phone": "545-394-3664", "status": "on hold", "company": "Vehement", "currency": " DZD", "email": "fclfo@mail.com", "rate": "26.6%"}, {"department": "i.t.", "phone": "595.725.1244", "status": "Active", "company": "Initech", "currency": " AMD ", "email": "HGBGEF@MAIL.COM", "rate": "8.7%"}, {"department": "Operations ", "phone": "596.276.1118", "status": " ACTIVE ", "company": "Umbrella", "currency": "AMD", "email": "ejdh@corp.io", "rate": "71.9%"}, {"department": "I.T.", "phone": "562-237-7757", "status": "actime", "company": "Soylent Corp", "currency": "Algerian Dnar", "email": "NDECO@EXAMPLE.COM", "rate": "66.5%"}, {"department": "infohmation technology  ", "phone": "541-559-2926", "status": "canceled", "company": "Stark Industries", "currency": "AMD", "email": "nldbj@example.com", "rate": "8.9%"}, {"department": "OPS", "phone": "533-152-8482", "status": "CHURNED", "company": "Soylent Corp", "currency": "amd", "email": "eeiopi@mail.com", "rate": "10.0%"}, {"department": "IT", "phone": "502-217-9508", "status": "  Trial", "company": "Cyberdyne ", "currency": " algeria ndinar", "email": "MPHEIC@MAIL.COM", "rate": "90.1%"}, {"department": "Operations", "phone": "(552)158-5587", "status": "churned", "company": "Acme Inc", "currency": "armenian dram", "email": " PCPM@EXAMPLE.COM", "rate": "29.9%"}, {"department": "Dev", "phone": "5010446679", "status": "Churned", "company": "Stark Industries", "currency": "Armenian Dram", "email": "JIHG@CORP.IO", "rate": "28.2%"}, {"department": "ops", "phone": "562-419-2224", "status": " Trial", "company": "Wonka Co", "currency": "AMD", "email": "EKLIGF@CORP.IO", "rate": "95.4%"}, {"department": "INFORMATION TECHNOLOGY", "phone": "567-969-9370", "status": "Churned", "company": "Umbrella", "currency": "algerian dinar ", "email": "encnkkb@mail.com", "rate": "89.2%"}, {"department": "OPS", "phone": "548-148-9841", "status": "paused", "company": "Cyberdyne  ", "currency": "   DZD", "email": " hgoddk@test.org ", "rate": "16.2%"}, {"department": "Eng", "phone": "5248864630", "status": "on hold", "company": "Hooli", "currency": "AMD", "email": " adkpbgp@example.com ", "rate": "22.3%"}, {"department": "it", "phone": "5192430150", "status": "trial", "company": " Hooli ", "currency": "Armenian Dram", "email": " plbhl@corp.io ", "rate": "76.1%"}, {"department": "ENG.", "phone": "5405200419", "status": "paused", "company": " Hooli ", "currency": "ARMENIAN DRAM", "email": "   afdgjaj@corp.io", "rate": "45.5%"}, {"department": " IT ", "phone": "(539)570-8694", "status": "paused   ", "company": " Initech ", "currency": "  DZD", "email": " kjcg@example.com ", "rate": "58.2%"}, {"department": "eng", "phone": "(536)004-3286", "status": " canceled ", "company": "Hooli", "currency": "armenian dram", "email": "FHDPBOM@MAIL.COM", "rate": "12.1%"}, {"department": "it", "phone": "5832862972", "status": "Trial", "company": "Umbrella", "currency": "AMD", "email": "bpcaeh@example.com", "rate": "21.1%"}, {"department": "Ops", "phone": "(563)149-2776", "status": " churned ", "company": "Globex", "currency": "Algerian Dinar   ", "email": "IJKFMC@TEST.ORG", "rate": "11.6%"}, {"department": "OPS", "phone": "(566)215-1158", "status": "  churned", "company": "   Globex", "currency": "algerian dinar", "email": "bnljdcb@corp.io", "rate": "49.1%"}, {"department": "Dev", "phone": "(593)185-0398", "status": "cancelled", "company": "Stark Industries", "currency": "ALGERIAN DINAR", "email": "ddmiam@mail.com", "rate": "15.5%"}, {"department": "ops", "phone": "5530810656", "status": "trial", "company": "Vehement", "currency": "DZD", "email": "   CPHOF@MAIL.COM", "rate": "47.7%"}, {"department": "  Information Technology", "phone": "(517)098-5721", "status": "TIAL", "company": "Stark  Industries", "currency": "AMD", "email": "hjnc@corp.io", "rate": "72.8%"}, {"department": "ENG", "phone": "(519)680-3547", "status": "active", "company": "Globex", "currency": "DZD", "email": "  fpbddm@test.org", "rate": "92.6%"}, {"department": "engineering", "phone": "5390761280", "status": "Churned", "company": "Hooli", "currency": "  Armenian Dram", "email": "ofnpi@test.org", "rate": "31.8%"}, {"department": "Operations  ", "phone": "5441450028", "status": "Trial", "company": "Globex", "currency": "Algerian  Dinar", "email": "HGDEO@MAIL.COM", "rate": "16.9%"}, {"department": "  eng.", "phone": "538-134-2853", "status": "cancelled", "company": " Globex ", "currency": "Algerian Dinar", "email": "niahglg@corp.io", "rate": "87.5%"}, {"department": " Ent.", "phone": "582.379.4396", "status": "FREE TRIAL", "company": "Hooli   ", "currency": "AMD", "email": "dpageap@corp.io  ", "rate": "64.0%"}, {"department": "  engineering", "phone": "549.452.7486", "status": "paused", "company": "Umbrella", "currency": "Amd", "email": "  aijhfof@example.com", "rate": "1.3%"}, {"department": "operations", "phone": "(533)618-5332", "status": "ATCIVE", "company": " Stark Industries", "currency": "algerian dinar", "email": "   hogego@test.org", "rate": "40.4%"}, {"department": "ENG.", "phone": "(595)835-2094", "status": "on  hlod", "company": "Soylent Corp", "currency": "DZD   ", "email": "  jcmjfo@test.org", "rate": "23.9%"}, {"department": "", "phone": "", "status": "", "company": "", "currency": "", "email": "", "rate": ""}], "clean": [{"department": "Information Technology", "phone": "(528) 275-4291", "status": "Trial", "company": "Hooli", "currency": "DZD", "email": "pofbbf@example.com", "rate": 0.6859999999999999}, {"department": "Operations", "phone": "(530) 368-8945", "status": "Paused", "company": "Stark Industries", "currency": "AMD", "email": "mapgcei@corp.io", "rate": 0.871}, {"department": "Engineering", "phone": "(595) 835-2094", "status": "Paused", "company": "Soylent Corp", "currency": "DZD", "email": "jcmjfo@test.org", "rate": 0.239}, {"department": "Operations", "phone": "(505) 953-7233", "status": "Trial", "company": "Vehement", "currency": "DZD", "email": "fokjae@mail.com", "rate": 0.7440000000000001}, {"department": "Information Technology", "phone": "(511) 913-0543", "status": "Trial", "company": "Umbrella", "currency": "AMD", "email": "ahmfg@mail.com", "rate": 0.435}, {"department": "Information Technology", "phone": "(528) 240-7612", "status": "Paused", "company": "Wonka Co", "currency": "AMD", "email": "moihj@example.com", "rate": 0.146}, {"department": "Information Technology", "phone": "(525) 932-3654", "status": "Trial", "company": "Stark Industries", "currency": "DZD", "email": "llgaob@corp.io", "rate": 0.73}, {"department": "Information Technology", "phone": "(550) 280-1222", "status": "Trial", "company": "Globex", "currency": "DZD", "email": "gdhp@test.org", "rate": 0.863}, {"department": "Engineering", "phone": "(579) 374-3155", "status": "Active", "company": "Vehement", "currency": "AMD", "email": "lnbgppc@corp.io", "rate": 0.7440000000000001}, {"department": "Engineering", "phone": "(553) 490-6222", "status": "Paused", "company": "Wonka Co", "currency": "AMD", "email": "jahbffc@test.org", "rate": 0.532}, {"department": "Operations", "phone": "(545) 394-3664", "status": "Paused", "company": "Vehement", "currency": "DZD", "email": "fclfo@mail.com", "rate": 0.266}, {"department": "Information Technology", "phone": "(595) 725-1244", "status": "Active", "company": "Initech", "currency": "AMD", "email": "hgbgef@mail.com", "rate": 0.087}, {"department": "Operations", "phone": "(596) 276-1118", "status": "Active", "company": "Umbrella", "currency": "AMD", "email": "ejdh@corp.io", "rate": 0.7190000000000001}, {"department": "Information Technology", "phone": "(562) 237-7757", "status": "Active", "company": "Soylent Corp", "currency": "DZD", "email": "ndeco@example.com", "rate": 0.665}, {"department": "Information Technology", "phone": "(541) 559-2926", "status": "Churned", "company": "Stark Industries", "currency": "AMD", "email": "nldbj@example.com", "rate": 0.08900000000000001}, {"department": "Operations", "phone": "(533) 152-8482", "status": "Churned", "company": "Soylent Corp", "currency": "AMD", "email": "eeiopi@mail.com", "rate": 0.1}, {"department": "Information Technology", "phone": "(502) 217-9508", "status": "Trial", "company": "Cyberdyne", "currency": "DZD", "email": "mpheic@mail.com", "rate": 0.9009999999999999}, {"department": "Operations", "phone": "(552) 158-5587", "status": "Churned", "company": "Acme Inc", "currency": "AMD", "email": "pcpm@example.com", "rate": 0.299}, {"department": "Engineering", "phone": "(501) 044-6679", "status": "Churned", "company": "Stark Industries", "currency": "AMD", "email": "jihg@corp.io", "rate": 0.282}, {"department": "Operations", "phone": "(562) 419-2224", "status": "Trial", "company": "Wonka Co", "currency": "AMD", "email": "ekligf@corp.io", "rate": 0.9540000000000001}, {"department": "Information Technology", "phone": "(567) 969-9370", "status": "Churned", "company": "Umbrella", "currency": "DZD", "email": "encnkkb@mail.com", "rate": 0.892}, {"department": "Operations", "phone": "(548) 148-9841", "status": "Paused", "company": "Cyberdyne", "currency": "DZD", "email": "hgoddk@test.org", "rate": 0.162}, {"department": "Engineering", "phone": "(524) 886-4630", "status": "Paused", "company": "Hooli", "currency": "AMD", "email": "adkpbgp@example.com", "rate": 0.223}, {"department": "Information Technology", "phone": "(519) 243-0150", "status": "Trial", "company": "Hooli", "currency": "AMD", "email": "plbhl@corp.io", "rate": 0.7609999999999999}, {"department": "Engineering", "phone": "(540) 520-0419", "status": "Paused", "company": "Hooli", "currency": "AMD", "email": "afdgjaj@corp.io", "rate": 0.455}, {"department": "Information Technology", "phone": "(539) 570-8694", "status": "Paused", "company": "Initech", "currency": "DZD", "email": "kjcg@example.com", "rate": 0.5820000000000001}, {"department": "Engineering", "phone": "(536) 004-3286", "status": "Churned", "company": "Hooli", "currency": "AMD", "email": "fhdpbom@mail.com", "rate": 0.121}, {"department": "Information Technology", "phone": "(583) 286-2972", "status": "Trial", "company": "Umbrella", "currency": "AMD", "email": "bpcaeh@example.com", "rate": 0.21100000000000002}, {"department": "Operations", "phone": "(563) 149-2776", "status": "Churned", "company": "Globex", "currency": "DZD", "email": "ijkfmc@test.org", "rate": 0.11599999999999999}, {"department": "Operations", "phone": "(566) 215-1158", "status": "Churned", "company": "Globex", "currency": "DZD", "email": "bnljdcb@corp.io", "rate": 0.491}, {"department": "Engineering", "phone": "(593) 185-0398", "status": "Churned", "company": "Stark Industries", "currency": "DZD", "email": "ddmiam@mail.com", "rate": 0.155}, {"department": "Operations", "phone": "(553) 081-0656", "status": "Trial", "company": "Vehement", "currency": "DZD", "email": "cphof@mail.com", "rate": 0.47700000000000004}, {"department": "Information Technology", "phone": "(517) 098-5721", "status": "Trial", "company": "Stark Industries", "currency": "AMD", "email": "hjnc@corp.io", "rate": 0.728}, {"department": "Engineering", "phone": "(519) 680-3547", "status": "Active", "company": "Globex", "currency": "DZD", "email": "fpbddm@test.org", "rate": 0.9259999999999999}, {"department": "Engineering", "phone": "(539) 076-1280", "status": "Churned", "company": "Hooli", "currency": "AMD", "email": "ofnpi@test.org", "rate": 0.318}, {"department": "Operations", "phone": "(544) 145-0028", "status": "Trial", "company": "Globex", "currency": "DZD", "email": "hgdeo@mail.com", "rate": 0.16899999999999998}, {"department": "Engineering", "phone": "(538) 134-2853", "status": "Churned", "company": "Globex", "currency": "DZD", "email": "niahglg@corp.io", "rate": 0.875}, {"department": "Engineering", "phone": "(582) 379-4396", "status": "Trial", "company": "Hooli", "currency": "AMD", "email": "dpageap@corp.io", "rate": 0.64}, {"department": "Engineering", "phone": "(549) 452-7486", "status": "Paused", "company": "Umbrella", "currency": "AMD", "email": "aijhfof@example.com", "rate": 0.013000000000000001}, {"department": "Operations", "phone": "(533) 618-5332", "status": "Active", "company": "Stark Industries", "currency": "DZD", "email": "hogego@test.org", "rate": 0.40399999999999997}], "dirty_cols": ["department", "phone", "status", "company", "currency", "email", "rate"], "clean_cols": ["department", "phone", "status", "company", "currency", "email", "rate"], "plan": {"dataset_summary": "42 rows × 7 columns. 7 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Information Techeology": "Information Technology", "OPERATIONS": "Operations", "ENG.": "Engineering", "Ops": "Operations", "information technology": "Information Technology", "IT": "Information Technology", "INFORMATION TECHNOLOGY": "Information Technology", "Dev": "Engineering", "eng": "Engineering", "ops": "Operations", "i.t.": "Information Technology", "I.T.": "Information Technology", "infohmation technology": "Information Technology", "OPS": "Operations", "Eng": "Engineering", "it": "Information Technology", "ENG": "Engineering", "engineering": "Engineering", "eng.": "Engineering", "Ent.": "Engineering", "operations": "Operations"}, "rationale": "Unified 21 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"trual": "Trial", "paused": "Paused", "on hlod": "Paused", "trial": "Trial", "free trial": "Trial", "ON HOLD": "Paused", "on hold": "Paused", "ACTIVE": "Active", "actime": "Active", "canceled": "Churned", "CHURNED": "Churned", "churned": "Churned", "cancelled": "Churned", "TIAL": "Trial", "active": "Active", "FREE TRIAL": "Trial", "ATCIVE": "Active"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"dzd": "DZD", "ARMENIAN DRAM": "AMD", "Algerian Dinar": "DZD", "Armenian Dram": "AMD", "Artenian Dram": "AMD", "Algerian Dnar": "DZD", "amd": "AMD", "algeria ndinar": "DZD", "armenian dram": "AMD", "algerian dinar": "DZD", "ALGERIAN DINAR": "DZD", "Amd": "AMD"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"currency": "Nakfa", "email": "fabidb@test.org", "is_active": "FALSE", "department": "   PRODUCT", "job_title": "POD MGR", "amount": "1.342,68", "status": "   TRIAL"}, {"currency": "Barbados Dollar", "email": "lhclp@mail.com", "is_active": "FALSE", "department": "Finance", "job_title": "  VP", "amount": "$4,778.43", "status": "TRIAL"}, {"currency": " Nakfa ", "email": "MJLOMGM@CORP.IO", "is_active": "T", "department": " accounting ", "job_title": "V.P.", "amount": "--", "status": "   Paused"}, {"currency": "   Nakfa", "email": "HAAOAG@CORP.IO", "is_active": "true", "department": "product", "job_title": "   VP", "amount": "$8,599.26", "status": "churned"}, {"currency": "Nakfa", "email": " eelbkfa@corp.io", "is_active": "false", "department": "Product", "job_title": "Vice Pres", "amount": "null", "status": "churned"}, {"currency": "Barbados Dollar", "email": " cgapdjb@test.org", "is_active": "Yes", "department": "IT.", "job_title": "Account  Executive", "amount": "842,41", "status": "active"}, {"currency": "Barbados Dollar", "email": "CKGN@CORP.IO", "is_active": "TRUE", "department": "FIN", "job_title": "AE", "amount": "7.905,55", "status": "paused"}, {"currency": " ERN ", "email": " ckgc@mail.com ", "is_active": "1", "department": "Biz Dev", "job_title": "Prod Manager", "amount": "$2,784.76", "status": "trial"}, {"currency": "BARBADS DOLLAR", "email": " mjebn@mail.com ", "is_active": "1", "department": "Produut", "job_title": "   vp", "amount": "null", "status": "trial"}, {"currency": "Barbados Dollar", "email": "KKAN@MAIL.COM", "is_active": "TRUE", "department": "I.T.  ", "job_title": "Prod Mgr", "amount": "$6,029.87", "status": "free trial"}, {"currency": "barbados dollar", "email": "GALOK@CORP.IO  ", "is_active": "TRUE", "department": "product", "job_title": "pm", "amount": "1.946,63", "status": "active"}, {"currency": "Babados  Dollar", "email": "ONDN@EXAMPLE.COM", "is_active": "false", "department": " Finance ", "job_title": "Vice President", "amount": "4.219,57", "status": "canceled"}, {"currency": "Nakfa", "email": "EFAG@EXAMPLE.COM", "is_active": "0", "department": "Accounting", "job_title": "Prod Mgr", "amount": "$857.47", "status": "free trial"}, {"currency": "bbd", "email": "  gnajn@mail.com", "is_active": "Yes", "department": " Pm ", "job_title": "Prod Manager", "amount": "8.048,80", "status": "active"}, {"currency": " ERN ", "email": "NHCKJ@EXAMPLE.COM", "is_active": "1", "department": "Product", "job_title": "ae", "amount": "$7,043.64", "status": " paused "}, {"currency": "   nakfa", "email": "dopan@mail.com", "is_active": "1", "department": " ACCOUNTING", "job_title": "Vice President", "amount": "$2,674.46", "status": "churrned"}, {"currency": "ERN", "email": "BFBIB@EXAMPLE.COM  ", "is_active": "F", "department": "ItT.", "job_title": "Account Executive", "amount": "4.073,53", "status": "ACTIVE"}, {"currency": "BBD", "email": "  hefmpa@example.com", "is_active": "No", "department": "BIZ DEV", "job_title": "acct exec", "amount": "8.352,72", "status": "cancelled"}, {"currency": "Barbados Dollar", "email": "fjebco@test.org", "is_active": "TRUE", "department": "Information  Technology", "job_title": "product manager", "amount": "$2,712.67", "status": "on hold"}, {"currency": "Barbados Dollar", "email": " POABAA@MAIL.COM ", "is_active": "Y", "department": "  Finance", "job_title": "vice pres", "amount": "8.343,33", "status": "Trill"}, {"currency": "nakfa", "email": " ONMJNE@TEST.ORG ", "is_active": "true", "department": "accounting", "job_title": "V.P.", "amount": "$7,982.86", "status": "   TRIAL"}, {"currency": "ERN", "email": "GFLMINM@MAIL.COM  ", "is_active": "T", "department": "Accounting", "job_title": "ae ", "amount": "$8,613.91", "status": "On Hold"}, {"currency": "ERN", "email": "kfedcg@test.org  ", "is_active": "false", "department": "biz dev   ", "job_title": "account executive", "amount": "$8,105.13", "status": "trial"}, {"currency": "Nakfa", "email": " hhml@mail.com", "is_active": "FALSE", "department": "Biz Dev ", "job_title": "Prod Manager", "amount": "4.443,50", "status": "Aczive"}, {"currency": "Nakfa", "email": " GNCNB@CORP.IO ", "is_active": "Yes", "department": " prdo", "job_title": "acct exec   ", "amount": "4.260,57", "status": " ACTIVE"}, {"currency": "barbados dollar", "email": "baecf@corp.io", "is_active": "No", "department": "accounting", "job_title": " v.p.", "amount": "1.066,38", "status": "Canceld"}, {"currency": "Ern   ", "email": "fcfbi@test.org  ", "is_active": "FALSE", "department": "salys", "job_title": "Acct Exec", "amount": "5.037,96", "status": "Active"}, {"currency": "   ern", "email": "KODE@TEST.ORG ", "is_active": "FALSE", "department": "Biz Dev", "job_title": "PM ", "amount": "7.357,90", "status": "trial  "}, {"currency": "ERN", "email": "PBCE@TEST.ORG", "is_active": "T", "department": "prooduct", "job_title": "prod mgr", "amount": "5.127,74", "status": "Paused"}, {"currency": "NAKFA", "email": "MGANLJ@TEST.ORG  ", "is_active": "true", "department": "Pm   ", "job_title": "V.P.", "amount": "$529.37", "status": "active"}, {"currency": "BARBADOS DOJLAR", "email": "GCDHDAK@TEST.ORG", "is_active": "Yes", "department": "Prod", "job_title": "Account  Executive", "amount": "2.473,71", "status": "free trial "}, {"currency": "nakfa", "email": "AIPKDB@EXAMPLE.COM", "is_active": "F", "department": "INFORMATION TECHNOLOGY", "job_title": "AE", "amount": "4.841,99", "status": "churned"}, {"currency": "Barbados Dollar", "email": "pafj@example.com", "is_active": "1", "department": "Fin", "job_title": "AE", "amount": "-", "status": "Active"}, {"currency": "   bbd", "email": "GNNDFL@MAIL.COM", "is_active": "1", "department": "Sales", "job_title": "Acct Exec", "amount": "7.407,72", "status": "on hold"}, {"currency": "barbados dollar", "email": "FHIKILJ@CORP.IO   ", "is_active": "1", "department": "I.T.", "job_title": "ae", "amount": "$8,779.09", "status": "Triwl"}, {"currency": "", "email": "", "is_active": "", "department": "", "job_title": "", "amount": "", "status": ""}], "clean": [{"currency": "ERN", "email": "fabidb@test.org", "is_active": false, "department": "Product", "job_title": "Product Manager", "amount": 1342.68, "status": "Trial"}, {"currency": "BBD", "email": "lhclp@mail.com", "is_active": false, "department": "Finance", "job_title": "Vice President", "amount": 4778.43, "status": "Trial"}, {"currency": "ERN", "email": "mjlomgm@corp.io", "is_active": true, "department": "Finance", "job_title": "Vice President", "amount": NaN, "status": "Paused"}, {"currency": "ERN", "email": "haaoag@corp.io", "is_active": true, "department": "Product", "job_title": "Vice President", "amount": 8599.26, "status": "Churned"}, {"currency": "ERN", "email": "eelbkfa@corp.io", "is_active": false, "department": "Product", "job_title": "Vice President", "amount": NaN, "status": "Churned"}, {"currency": "BBD", "email": "cgapdjb@test.org", "is_active": true, "department": "Information Technology", "job_title": "Account Executive", "amount": 842.41, "status": "Active"}, {"currency": "BBD", "email": "ckgn@corp.io", "is_active": true, "department": "Finance", "job_title": "Account Executive", "amount": 7905.55, "status": "Paused"}, {"currency": "ERN", "email": "ckgc@mail.com", "is_active": true, "department": "Sales", "job_title": "Product Manager", "amount": 2784.76, "status": "Trial"}, {"currency": "BBD", "email": "mjebn@mail.com", "is_active": true, "department": "Product", "job_title": "Vice President", "amount": NaN, "status": "Trial"}, {"currency": "BBD", "email": "kkan@mail.com", "is_active": true, "department": "Information Technology", "job_title": "Product Manager", "amount": 6029.87, "status": "Trial"}, {"currency": "BBD", "email": "galok@corp.io", "is_active": true, "department": "Product", "job_title": "Product Manager", "amount": 1946.63, "status": "Active"}, {"currency": "BBD", "email": "ondn@example.com", "is_active": false, "department": "Finance", "job_title": "Vice President", "amount": 4219.57, "status": "Churned"}, {"currency": "ERN", "email": "efag@example.com", "is_active": false, "department": "Finance", "job_title": "Product Manager", "amount": 857.47, "status": "Trial"}, {"currency": "BBD", "email": "gnajn@mail.com", "is_active": true, "department": "Product", "job_title": "Product Manager", "amount": 8048.8, "status": "Active"}, {"currency": "ERN", "email": "nhckj@example.com", "is_active": true, "department": "Product", "job_title": "Account Executive", "amount": 7043.64, "status": "Paused"}, {"currency": "ERN", "email": "dopan@mail.com", "is_active": true, "department": "Finance", "job_title": "Vice President", "amount": 2674.46, "status": "Churned"}, {"currency": "ERN", "email": "bfbib@example.com", "is_active": false, "department": "Information Technology", "job_title": "Account Executive", "amount": 4073.53, "status": "Active"}, {"currency": "BBD", "email": "hefmpa@example.com", "is_active": false, "department": "Sales", "job_title": "Account Executive", "amount": 8352.72, "status": "Churned"}, {"currency": "BBD", "email": "fjebco@test.org", "is_active": true, "department": "Information Technology", "job_title": "Product Manager", "amount": 2712.67, "status": "Paused"}, {"currency": "BBD", "email": "poabaa@mail.com", "is_active": true, "department": "Finance", "job_title": "Vice President", "amount": 8343.33, "status": "Trial"}, {"currency": "ERN", "email": "onmjne@test.org", "is_active": true, "department": "Finance", "job_title": "Vice President", "amount": 7982.86, "status": "Trial"}, {"currency": "ERN", "email": "gflminm@mail.com", "is_active": true, "department": "Finance", "job_title": "Account Executive", "amount": 8613.91, "status": "Paused"}, {"currency": "ERN", "email": "kfedcg@test.org", "is_active": false, "department": "Sales", "job_title": "Account Executive", "amount": 8105.13, "status": "Trial"}, {"currency": "ERN", "email": "hhml@mail.com", "is_active": false, "department": "Sales", "job_title": "Product Manager", "amount": 4443.5, "status": "Active"}, {"currency": "ERN", "email": "gncnb@corp.io", "is_active": true, "department": "Product", "job_title": "Account Executive", "amount": 4260.57, "status": "Active"}, {"currency": "BBD", "email": "baecf@corp.io", "is_active": false, "department": "Finance", "job_title": "Vice President", "amount": 1066.38, "status": "Churned"}, {"currency": "ERN", "email": "fcfbi@test.org", "is_active": false, "department": "Sales", "job_title": "Account Executive", "amount": 5037.96, "status": "Active"}, {"currency": "ERN", "email": "kode@test.org", "is_active": false, "department": "Sales", "job_title": "Product Manager", "amount": 7357.9, "status": "Trial"}, {"currency": "ERN", "email": "pbce@test.org", "is_active": true, "department": "Product", "job_title": "Product Manager", "amount": 5127.74, "status": "Paused"}, {"currency": "ERN", "email": "mganlj@test.org", "is_active": true, "department": "Product", "job_title": "Vice President", "amount": 529.37, "status": "Active"}, {"currency": "BBD", "email": "gcdhdak@test.org", "is_active": true, "department": "Product", "job_title": "Account Executive", "amount": 2473.71, "status": "Trial"}, {"currency": "ERN", "email": "aipkdb@example.com", "is_active": false, "department": "Information Technology", "job_title": "Account Executive", "amount": 4841.99, "status": "Churned"}, {"currency": "BBD", "email": "pafj@example.com", "is_active": true, "department": "Finance", "job_title": "Account Executive", "amount": NaN, "status": "Active"}, {"currency": "BBD", "email": "gnndfl@mail.com", "is_active": true, "department": "Sales", "job_title": "Account Executive", "amount": 7407.72, "status": "Paused"}, {"currency": "BBD", "email": "fhikilj@corp.io", "is_active": true, "department": "Information Technology", "job_title": "Account Executive", "amount": 8779.09, "status": "Trial"}], "dirty_cols": ["currency", "email", "is_active", "department", "job_title", "amount", "status"], "clean_cols": ["currency", "email", "is_active", "department", "job_title", "amount", "status"], "plan": {"dataset_summary": "36 rows × 7 columns. 7 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Nakfa": "ERN", "Barbados Dollar": "BBD", "BARBADS DOLLAR": "BBD", "barbados dollar": "BBD", "Babados Dollar": "BBD", "bbd": "BBD", "nakfa": "ERN", "Ern": "ERN", "ern": "ERN", "NAKFA": "ERN", "BARBADOS DOJLAR": "BBD"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PRODUCT": "Product", "accounting": "Finance", "product": "Product", "IT.": "Information Technology", "FIN": "Finance", "Biz Dev": "Sales", "Produut": "Product", "I.T.": "Information Technology", "Accounting": "Finance", "Pm": "Product", "ACCOUNTING": "Finance", "ItT.": "Information Technology", "BIZ DEV": "Sales", "biz dev": "Sales", "prdo": "Product", "salys": "Sales", "prooduct": "Product", "Prod": "Product", "INFORMATION TECHNOLOGY": "Information Technology", "Fin": "Finance"}, "rationale": "Unified 20 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"POD MGR": "Product Manager", "VP": "Vice President", "V.P.": "Vice President", "Vice Pres": "Vice President", "AE": "Account Executive", "Prod Manager": "Product Manager", "vp": "Vice President", "Prod Mgr": "Product Manager", "pm": "Product Manager", "ae": "Account Executive", "acct exec": "Account Executive", "product manager": "Product Manager", "vice pres": "Vice President", "account executive": "Account Executive", "v.p.": "Vice President", "Acct Exec": "Account Executive", "PM": "Product Manager", "prod mgr": "Product Manager"}, "rationale": "Unified 18 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TRIAL": "Trial", "churned": "Churned", "active": "Active", "paused": "Paused", "trial": "Trial", "free trial": "Trial", "canceled": "Churned", "churrned": "Churned", "ACTIVE": "Active", "cancelled": "Churned", "on hold": "Paused", "Trill": "Trial", "On Hold": "Paused", "Aczive": "Active", "Canceld": "Churned", "Triwl": "Trial"}, "rationale": "Unified 16 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"phone": "(577)095-8441", "email": "gjej@mail.com ", "job_title": "Ceo "}, {"phone": "560-134-8552", "email": "  KBIJJL@TEST.ORG", "job_title": "eng manager"}, {"phone": "5640943517", "email": "BLCBHGI@TEST.ORG", "job_title": "CEO"}, {"phone": "531-786-8065", "email": "   poflffk@test.org", "job_title": "dir"}, {"phone": "522-797-1847", "email": "PFJGIE@CORP.IO", "job_title": "dir."}, {"phone": "555.742.0576", "email": "OMAOO@EXAMPLE.COM", "job_title": "CHIEF EXECUTIVE OFFICER"}, {"phone": "578.136.0306", "email": "EFEG@TEST.ORG", "job_title": "Eng Mgr"}, {"phone": "535.418.1361", "email": "bmajaa@test.org", "job_title": "Dir  "}, {"phone": "545.338.4966", "email": "adockdk@mail.com", "job_title": "Eng Mgr"}, {"phone": "(547)802-6069", "email": "kdhc@mail.com", "job_title": "   DIR"}, {"phone": "589-985-7485", "email": "ioain@test.org", "job_title": "Chief Executive Officer"}, {"phone": "5281011636", "email": "DCKJKC@TEST.ORG", "job_title": "  Chief Executive Officer"}, {"phone": "5479919587", "email": "iobdk@corp.io", "job_title": "dir."}, {"phone": "(547)150-2712", "email": " iako@mail.com ", "job_title": "dic."}, {"phone": "(577)095-8441", "email": "gjej@mail.com ", "job_title": "Ceo "}], "clean": [{"phone": "(577) 095-8441", "email": "gjej@mail.com", "job_title": "Chief Executive Officer"}, {"phone": "(560) 134-8552", "email": "kbijjl@test.org", "job_title": "Engineering Manager"}, {"phone": "(564) 094-3517", "email": "blcbhgi@test.org", "job_title": "Chief Executive Officer"}, {"phone": "(531) 786-8065", "email": "poflffk@test.org", "job_title": "Director"}, {"phone": "(522) 797-1847", "email": "pfjgie@corp.io", "job_title": "Director"}, {"phone": "(555) 742-0576", "email": "omaoo@example.com", "job_title": "Chief Executive Officer"}, {"phone": "(578) 136-0306", "email": "efeg@test.org", "job_title": "Engineering Manager"}, {"phone": "(535) 418-1361", "email": "bmajaa@test.org", "job_title": "Director"}, {"phone": "(545) 338-4966", "email": "adockdk@mail.com", "job_title": "Engineering Manager"}, {"phone": "(547) 802-6069", "email": "kdhc@mail.com", "job_title": "Director"}, {"phone": "(589) 985-7485", "email": "ioain@test.org", "job_title": "Chief Executive Officer"}, {"phone": "(528) 101-1636", "email": "dckjkc@test.org", "job_title": "Chief Executive Officer"}, {"phone": "(547) 991-9587", "email": "iobdk@corp.io", "job_title": "Director"}, {"phone": "(547) 150-2712", "email": "iako@mail.com", "job_title": "Director"}], "dirty_cols": ["phone", "email", "job_title"], "clean_cols": ["phone", "email", "job_title"], "plan": {"dataset_summary": "15 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ceo": "Chief Executive Officer", "eng manager": "Engineering Manager", "CEO": "Chief Executive Officer", "dir": "Director", "dir.": "Director", "CHIEF EXECUTIVE OFFICER": "Chief Executive Officer", "Eng Mgr": "Engineering Manager", "Dir": "Director", "DIR": "Director", "dic.": "Director"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"country": "LB", "job_title": "CTO ", "rate": "13.1%", "is_active": "N", "unnamed": ""}, {"country": " TZA ", "job_title": " CTO ", "rate": "16.1%", "is_active": "T", "unnamed": ""}, {"country": " bvt ", "job_title": "Dir", "rate": "81.0%", "is_active": "0", "unnamed": ""}, {"country": "Bouvet Island", "job_title": "dir", "rate": "21.5%", "is_active": "false", "unnamed": ""}, {"country": "   lbanese republic", "job_title": "Chief Technology Officer", "rate": "68.8%", "is_active": "No", "unnamed": ""}, {"country": "Lebanese Republic", "job_title": "   cto", "rate": "92.1%", "is_active": "T", "unnamed": ""}, {"country": "LBN", "job_title": "  chief technology officer", "rate": "54.9%", "is_active": "N", "unnamed": ""}, {"country": "bouvet gsland", "job_title": "Dir.", "rate": "28.8%", "is_active": "0", "unnamed": ""}, {"country": "Tanzania, United Reublic of", "job_title": "Director", "rate": "3.1%", "is_active": "0", "unnamed": ""}, {"country": "tz", "job_title": "dir ", "rate": "13.7%", "is_active": "true", "unnamed": ""}, {"country": " bouvet island", "job_title": "cto", "rate": "41.0%", "is_active": "F", "unnamed": ""}, {"country": "BV", "job_title": "chief technology officer", "rate": "61.8%", "is_active": "F", "unnamed": ""}, {"country": "UNITED REPUBLIC OF TANZANIA", "job_title": "Chief Technology Officer", "rate": "51.8%", "is_active": "N", "unnamed": ""}], "clean": [{"country": "Lebanon", "job_title": "Chief Technology Officer", "rate": 0.131, "is_active": false}, {"country": "Tanzania, United Republic of", "job_title": "Chief Technology Officer", "rate": 0.161, "is_active": true}, {"country": "Bouvet Island", "job_title": "Director", "rate": 0.81, "is_active": false}, {"country": "Bouvet Island", "job_title": "Director", "rate": 0.215, "is_active": false}, {"country": "Lebanon", "job_title": "Chief Technology Officer", "rate": 0.688, "is_active": false}, {"country": "Lebanon", "job_title": "Chief Technology Officer", "rate": 0.9209999999999999, "is_active": true}, {"country": "Lebanon", "job_title": "Chief Technology Officer", "rate": 0.5489999999999999, "is_active": false}, {"country": "Bouvet Island", "job_title": "Director", "rate": 0.28800000000000003, "is_active": false}, {"country": "Tanzania, United Republic of", "job_title": "Director", "rate": 0.031, "is_active": false}, {"country": "Tanzania, United Republic of", "job_title": "Director", "rate": 0.13699999999999998, "is_active": true}, {"country": "Bouvet Island", "job_title": "Chief Technology Officer", "rate": 0.41, "is_active": false}, {"country": "Bouvet Island", "job_title": "Chief Technology Officer", "rate": 0.618, "is_active": false}, {"country": "Tanzania, United Republic of", "job_title": "Chief Technology Officer", "rate": 0.518, "is_active": false}], "dirty_cols": ["country", "job_title", "rate", "is_active", "unnamed"], "clean_cols": ["country", "job_title", "rate", "is_active"], "plan": {"dataset_summary": "13 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"LB": "Lebanon", "TZA": "Tanzania, United Republic of", "bvt": "Bouvet Island", "lbanese republic": "Lebanon", "Lebanese Republic": "Lebanon", "LBN": "Lebanon", "bouvet gsland": "Bouvet Island", "Tanzania, United Reublic of": "Tanzania, United Republic of", "tz": "Tanzania, United Republic of", "bouvet island": "Bouvet Island", "BV": "Bouvet Island", "UNITED REPUBLIC OF TANZANIA": "Tanzania, United Republic of"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CTO": "Chief Technology Officer", "Dir": "Director", "dir": "Director", "cto": "Chief Technology Officer", "chief technology officer": "Chief Technology Officer", "Dir.": "Director"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"phone": "(507)041-2811", "status": "in progress", "company": "Globex", "name": "Mona Diaz"}, {"phone": "5304809078", "status": "wip", "company": "Initech", "name": "Karl Adams"}, {"phone": "516-726-5098", "status": "WIP", "company": "Soylent  Corp", "name": "Omar Adams"}, {"phone": "544-360-0006", "status": "IN PROGRESS ", "company": "Acme  Inc", "name": "Frank  Cruz"}, {"phone": "535-394-2899", "status": " won ", "company": "Acme  Inc", "name": "Sara Ali"}, {"phone": "(556)790-4600", "status": "WIP  ", "company": "Acme Inc", "name": " Carol Reyes"}, {"phone": "(579)499-9266", "status": "ONGOING", "company": "Soylent  Corp", "name": "Judy Novak"}, {"phone": "(525)588-7046", "status": "WON", "company": "Wonka Co   ", "name": "Ivan Brandt"}, {"phone": "531-986-5540", "status": "won", "company": "Initech  ", "name": "Karl Petrov"}, {"phone": "573.513.4407", "status": "CLOSEDWON", "company": " Globex ", "name": "Judy Wong"}, {"phone": "568.611.8796", "status": "Closed on", "company": "Cyberdyne", "name": "Carol Johnson"}, {"phone": "506-531-1875", "status": "  won", "company": "Vehement", "name": "  Karl Brandt"}, {"phone": "596.482.8570", "status": "in progress", "company": "Acme Inc", "name": "Ivan  Moore"}, {"phone": "(542)377-9270", "status": "  in progress", "company": "Initech", "name": "Alice Petrov"}, {"phone": "522-475-3706", "status": "closed-won", "company": "   Acme Inc", "name": "  Grace Park"}, {"phone": "563-966-5103", "status": "CLOSED WON", "company": "Wonka  Co", "name": "Carol  Novak"}, {"phone": "5166099593", "status": "in progress", "company": "Globex", "name": "David Reyes"}, {"phone": "5325911119", "status": "Ongonig", "company": "Umbrella", "name": "Sara Ali"}], "clean": [{"phone": "(507) 041-2811", "status": "In Progress", "company": "Globex", "name": "Mona Diaz"}, {"phone": "(530) 480-9078", "status": "In Progress", "company": "Initech", "name": "Karl Adams"}, {"phone": "(516) 726-5098", "status": "In Progress", "company": "Soylent Corp", "name": "Omar Adams"}, {"phone": "(544) 360-0006", "status": "In Progress", "company": "Acme Inc", "name": "Frank Cruz"}, {"phone": "(535) 394-2899", "status": "Won", "company": "Acme Inc", "name": "Sara Ali"}, {"phone": "(556) 790-4600", "status": "In Progress", "company": "Acme Inc", "name": "Carol Reyes"}, {"phone": "(579) 499-9266", "status": "In Progress", "company": "Soylent Corp", "name": "Judy Novak"}, {"phone": "(525) 588-7046", "status": "Won", "company": "Wonka Co", "name": "Ivan Brandt"}, {"phone": "(531) 986-5540", "status": "Won", "company": "Initech", "name": "Karl Petrov"}, {"phone": "(573) 513-4407", "status": "Won", "company": "Globex", "name": "Judy Wong"}, {"phone": "(568) 611-8796", "status": "Won", "company": "Cyberdyne", "name": "Carol Johnson"}, {"phone": "(506) 531-1875", "status": "Won", "company": "Vehement", "name": "Karl Brandt"}, {"phone": "(596) 482-8570", "status": "In Progress", "company": "Acme Inc", "name": "Ivan Moore"}, {"phone": "(542) 377-9270", "status": "In Progress", "company": "Initech", "name": "Alice Petrov"}, {"phone": "(522) 475-3706", "status": "Won", "company": "Acme Inc", "name": "Grace Park"}, {"phone": "(563) 966-5103", "status": "Won", "company": "Wonka Co", "name": "Carol Novak"}, {"phone": "(516) 609-9593", "status": "In Progress", "company": "Globex", "name": "David Reyes"}, {"phone": "(532) 591-1119", "status": "In Progress", "company": "Umbrella", "name": "Sara Ali"}], "dirty_cols": ["phone", "status", "company", "name"], "clean_cols": ["phone", "status", "company", "name"], "plan": {"dataset_summary": "18 rows × 4 columns. 4 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"in progress": "In Progress", "wip": "In Progress", "WIP": "In Progress", "IN PROGRESS": "In Progress", "won": "Won", "ONGOING": "In Progress", "WON": "Won", "CLOSEDWON": "Won", "Closed on": "Won", "closed-won": "Won", "CLOSED WON": "Won", "Ongonig": "In Progress"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"company": " Cyberdyne ", "amount": "$1,548.46", "signup_date": "45240", "state": "SC", "city": "NOLA ", "department": "Information Technology"}, {"company": "   Stark Industries", "amount": "561,86", "signup_date": "9/11/2023", "state": "CO", "city": "Koblenz", "department": " i.t. "}, {"company": "Wonka Co", "amount": "3.364,29", "signup_date": "1 Feb 2023", "state": "SC", "city": "New Orlleans", "department": "  Eng"}, {"company": "Cyberdyne", "amount": "1.477,85", "signup_date": "1 Jan 2023", "state": "AK", "city": "Koblwnz", "department": "Information Technology"}, {"company": " Hooli ", "amount": "$770.49", "signup_date": "6 Jun 2023", "state": " Sc", "city": "Haveli  Lakha", "department": "IT"}, {"company": "Soylent  Corp", "amount": "1.888,88", "signup_date": "2023-02-04", "state": "Colorado  ", "city": "Kobelnz", "department": "IT"}, {"company": "Vehement", "amount": "$5,892.62", "signup_date": "2023-02-09", "state": "Missouri", "city": "Marsala", "department": "R&D"}, {"company": "Initech", "amount": "$5,545.60", "signup_date": "6/6/2023", "state": "AK", "city": "Marsala", "department": "r&d"}, {"company": "Soylent  Corp", "amount": "7.571,89", "signup_date": "2/12/2023", "state": "Utah", "city": "Haveli Lakha", "department": "engineering"}, {"company": "  Globex", "amount": "$5,220.56", "signup_date": "11/16/2023", "state": "Colordao", "city": "HAVELI LAKHA ", "department": "R&D"}, {"company": "Wonka Co", "amount": "6.188,42", "signup_date": "5/9/2023", "state": "sc  ", "city": "Haveli Lakha", "department": "Dev"}, {"company": "Cyberdyne", "amount": "$1,732.97", "signup_date": "3/23/2023", "state": "SONTH CAROLINA", "city": "Haveli Lakha", "department": "Eng."}, {"company": "  Soylent Corp", "amount": "8.755,86", "signup_date": "25 Sep 2023", "state": "co", "city": "gampengrejo", "department": "Engineering"}, {"company": " Acme Inc", "amount": "2.432,29", "signup_date": "5/21/2023", "state": "UTAH", "city": "new orleans", "department": "eng"}, {"company": "Wonka Co", "amount": "$5,207.27", "signup_date": "2023-12-25", "state": "alaska", "city": "new orleans", "department": "Engineernng"}, {"company": "Cyberdyne", "amount": "$8,972.21", "signup_date": "12 Jul 2023", "state": "Utxh", "city": "NEW ORLEANS", "department": "Information Technloogy  "}], "clean": [{"company": "Cyberdyne", "amount": 1548.46, "signup_date": "2023-11-10", "state": "South Carolina", "city": "New Orleans", "department": "Information Technology"}, {"company": "Stark Industries", "amount": 561.86, "signup_date": "2023-09-11", "state": "Colorado", "city": "Koblenz", "department": "Information Technology"}, {"company": "Wonka Co", "amount": 3364.29, "signup_date": "2023-02-01", "state": "South Carolina", "city": "New Orleans", "department": "Engineering"}, {"company": "Cyberdyne", "amount": 1477.85, "signup_date": "2023-01-01", "state": "Alaska", "city": "Koblenz", "department": "Information Technology"}, {"company": "Hooli", "amount": 770.49, "signup_date": "2023-06-06", "state": "South Carolina", "city": "Haveli Lakha", "department": "Information Technology"}, {"company": "Soylent Corp", "amount": 1888.88, "signup_date": "2023-02-04", "state": "Colorado", "city": "Koblenz", "department": "Information Technology"}, {"company": "Vehement", "amount": 5892.62, "signup_date": "2023-02-09", "state": "Missouri", "city": "Marsala", "department": "Engineering"}, {"company": "Initech", "amount": 5545.6, "signup_date": "2023-06-06", "state": "Alaska", "city": "Marsala", "department": "Engineering"}, {"company": "Soylent Corp", "amount": 7571.89, "signup_date": "2023-02-12", "state": "Utah", "city": "Haveli Lakha", "department": "Engineering"}, {"company": "Globex", "amount": 5220.56, "signup_date": "2023-11-16", "state": "Colorado", "city": "Haveli Lakha", "department": "Engineering"}, {"company": "Wonka Co", "amount": 6188.42, "signup_date": "2023-05-09", "state": "South Carolina", "city": "Haveli Lakha", "department": "Engineering"}, {"company": "Cyberdyne", "amount": 1732.97, "signup_date": "2023-03-23", "state": "South Carolina", "city": "Haveli Lakha", "department": "Engineering"}, {"company": "Soylent Corp", "amount": 8755.86, "signup_date": "2023-09-25", "state": "Colorado", "city": "Gampengrejo", "department": "Engineering"}, {"company": "Acme Inc", "amount": 2432.29, "signup_date": "2023-05-21", "state": "Utah", "city": "New Orleans", "department": "Engineering"}, {"company": "Wonka Co", "amount": 5207.27, "signup_date": "2023-12-25", "state": "Alaska", "city": "New Orleans", "department": "Engineering"}, {"company": "Cyberdyne", "amount": 8972.21, "signup_date": "2023-07-12", "state": "Utah", "city": "New Orleans", "department": "Information Technology"}], "dirty_cols": ["company", "amount", "signup_date", "state", "city", "department"], "clean_cols": ["company", "amount", "signup_date", "state", "city", "department"], "plan": {"dataset_summary": "16 rows × 6 columns. 6 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"SC": "South Carolina", "CO": "Colorado", "AK": "Alaska", "Sc": "South Carolina", "Colordao": "Colorado", "sc": "South Carolina", "SONTH CAROLINA": "South Carolina", "co": "Colorado", "UTAH": "Utah", "alaska": "Alaska", "Utxh": "Utah"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"NOLA": "New Orleans", "New Orlleans": "New Orleans", "Koblwnz": "Koblenz", "Kobelnz": "Koblenz", "HAVELI LAKHA": "Haveli Lakha", "gampengrejo": "Gampengrejo", "new orleans": "New Orleans", "NEW ORLEANS": "New Orleans"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"i.t.": "Information Technology", "Eng": "Engineering", "IT": "Information Technology", "R&D": "Engineering", "r&d": "Engineering", "engineering": "Engineering", "Dev": "Engineering", "Eng.": "Engineering", "eng": "Engineering", "Engineernng": "Engineering", "Information Technloogy": "Information Technology"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"state": "OH", "currency": "WIR Euro", "unit": "IN", "email": "  pkko@test.org", "country": "BAHAMAS", "status": "WON", "notes2": ""}, {"state": "OREGGON", "currency": "Covnertible Mark", "unit": "km", "email": "aibnf@corp.io", "country": "BS", "status": "won", "notes2": ""}, {"state": "OR", "currency": "Czzech Koruna", "unit": "Kilometers", "email": "hcnmfbd@mail.com", "country": "commonwealth of the bahamas", "status": "OPEN", "notes2": ""}, {"state": "Ohio", "currency": "  HTG", "unit": "  inches", "email": " IAPCIDH@MAIL.COM ", "country": "   SX", "status": "OPEN", "notes2": ""}, {"state": "oregon", "currency": "WIR Euro", "unit": "kilomteer", "email": "  ocnmm@test.org", "country": "sint maarten dutch part", "status": "OPEN", "notes2": ""}, {"state": "  NE", "currency": "Bam", "unit": "\"", "email": " NGDJ@CORP.IO", "country": "commonwealth of the bahamas", "status": "new", "notes2": ""}, {"state": "oh", "currency": "Gourde  ", "unit": "inches", "email": "hipifka@mail.com  ", "country": "Sint Maarten (Dutch part)   ", "status": "Open", "notes2": ""}, {"state": "oregon", "currency": " gourrde", "unit": "Km ", "email": " ADFN@TEST.ORG ", "country": " BS ", "status": "won", "notes2": ""}, {"state": "   NE", "currency": "gourde", "unit": "inch", "email": "MOBKCK@EXAMPLE.COM ", "country": " BHAAMAS ", "status": "Closed Won", "notes2": ""}, {"state": "or", "currency": "Convertible Mark   ", "unit": "Kilometer", "email": " cdijif@test.org", "country": "sint maarten (dutch part)", "status": "Open", "notes2": ""}, {"state": "or", "currency": "wir euro", "unit": "INCH", "email": "   FJJG@TEST.ORG", "country": "sx", "status": "Open", "notes2": ""}, {"state": "   nebraska", "currency": "CZK", "unit": "KM", "email": "  CDKKJ@MAIL.COM", "country": "sxm ", "status": "closed-won", "notes2": ""}, {"state": "Oh", "currency": "HTG", "unit": " kilometers ", "email": "iing@corp.io", "country": "BHS", "status": "Open", "notes2": ""}, {"state": "Nebraska", "currency": "   CHE", "unit": "kilometer", "email": "   eckmgc@corp.io", "country": "sx", "status": "won", "notes2": ""}, {"state": "Ohio", "currency": "CHE", "unit": "inch", "email": "EGHNJF@EXAMPLE.COM", "country": "commonwealth of the bahamas", "status": "won", "notes2": ""}, {"state": "oh", "currency": "Gourde  ", "unit": "inches", "email": "hipifka@mail.com  ", "country": "Sint Maarten (Dutch part)   ", "status": "Open", "notes2": ""}], "clean": [{"state": "Ohio", "currency": "CHE", "unit": "in", "email": "pkko@test.org", "country": "Bahamas", "status": "Won"}, {"state": "Oregon", "currency": "BAM", "unit": "km", "email": "aibnf@corp.io", "country": "Bahamas", "status": "Won"}, {"state": "Oregon", "currency": "CZK", "unit": "km", "email": "hcnmfbd@mail.com", "country": "Bahamas", "status": "Open"}, {"state": "Ohio", "currency": "HTG", "unit": "in", "email": "iapcidh@mail.com", "country": "Sint Maarten (Dutch part)", "status": "Open"}, {"state": "Oregon", "currency": "CHE", "unit": "km", "email": "ocnmm@test.org", "country": "Sint Maarten (Dutch part)", "status": "Open"}, {"state": "Nebraska", "currency": "BAM", "unit": "in", "email": "ngdj@corp.io", "country": "Bahamas", "status": "Open"}, {"state": "Ohio", "currency": "HTG", "unit": "in", "email": "hipifka@mail.com", "country": "Sint Maarten (Dutch part)", "status": "Open"}, {"state": "Oregon", "currency": "HTG", "unit": "km", "email": "adfn@test.org", "country": "Bahamas", "status": "Won"}, {"state": "Nebraska", "currency": "HTG", "unit": "in", "email": "mobkck@example.com", "country": "Bahamas", "status": "Won"}, {"state": "Oregon", "currency": "BAM", "unit": "km", "email": "cdijif@test.org", "country": "Sint Maarten (Dutch part)", "status": "Open"}, {"state": "Oregon", "currency": "CHE", "unit": "in", "email": "fjjg@test.org", "country": "Sint Maarten (Dutch part)", "status": "Open"}, {"state": "Nebraska", "currency": "CZK", "unit": "km", "email": "cdkkj@mail.com", "country": "Sint Maarten (Dutch part)", "status": "Won"}, {"state": "Ohio", "currency": "HTG", "unit": "km", "email": "iing@corp.io", "country": "Bahamas", "status": "Open"}, {"state": "Nebraska", "currency": "CHE", "unit": "km", "email": "eckmgc@corp.io", "country": "Sint Maarten (Dutch part)", "status": "Won"}, {"state": "Ohio", "currency": "CHE", "unit": "in", "email": "eghnjf@example.com", "country": "Bahamas", "status": "Won"}], "dirty_cols": ["state", "currency", "unit", "email", "country", "status", "notes2"], "clean_cols": ["state", "currency", "unit", "email", "country", "status"], "plan": {"dataset_summary": "16 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OH": "Ohio", "OREGGON": "Oregon", "OR": "Oregon", "oregon": "Oregon", "NE": "Nebraska", "oh": "Ohio", "or": "Oregon", "nebraska": "Nebraska", "Oh": "Ohio"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"WIR Euro": "CHE", "Covnertible Mark": "BAM", "Czzech Koruna": "CZK", "Bam": "BAM", "Gourde": "HTG", "gourrde": "HTG", "gourde": "HTG", "Convertible Mark": "BAM", "wir euro": "CHE"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"IN": "in", "Kilometers": "km", "inches": "in", "kilomteer": "km", "\"": "in", "Km": "km", "inch": "in", "Kilometer": "km", "INCH": "in", "KM": "km", "kilometers": "km", "kilometer": "km"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BAHAMAS": "Bahamas", "BS": "Bahamas", "commonwealth of the bahamas": "Bahamas", "SX": "Sint Maarten (Dutch part)", "sint maarten dutch part": "Sint Maarten (Dutch part)", "BHAAMAS": "Bahamas", "sint maarten (dutch part)": "Sint Maarten (Dutch part)", "sx": "Sint Maarten (Dutch part)", "sxm": "Sint Maarten (Dutch part)", "BHS": "Bahamas"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"WON": "Won", "won": "Won", "OPEN": "Open", "new": "Open", "Closed Won": "Won", "closed-won": "Won"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"signup_date": "44972", "company": "Umbrella", "department": "product", "job_title": "Dir."}, {"signup_date": "2023-05-13", "company": "  Vehement", "department": "Accounting", "job_title": "prod manaer"}, {"signup_date": "2023-11-18", "company": "Stark Industries", "department": " Accounting ", "job_title": "  Vp"}, {"signup_date": "2023-08-01", "company": "Wonka Co   ", "department": "Operations", "job_title": "DIR."}, {"signup_date": "6 Dec 2023", "company": "Vehement", "department": "Ops", "job_title": "vice pres"}, {"signup_date": "3/10/2023", "company": " Umbrella ", "department": "pm", "job_title": "PM"}, {"signup_date": "27 Feb 2023", "company": " Hooli ", "department": "Product", "job_title": "PRODUCT MANAGER"}, {"signup_date": "11 Jul 2023", "company": " Hooli ", "department": " Finance ", "job_title": "Product  Manager"}, {"signup_date": "5/12/2023", "company": "Acme Inc   ", "department": "Ops", "job_title": "VP"}, {"signup_date": "24 Jul 2023", "company": " Globex", "department": "Finance", "job_title": "Director"}, {"signup_date": "9/12/2023", "company": "Stark Industries", "department": "Finance", "job_title": " V.P. "}, {"signup_date": "1/17/2023", "company": "Umbrella  ", "department": "FINANCE", "job_title": " Prod Manager"}, {"signup_date": "44981", "company": "Soylent Corp   ", "department": "ops", "job_title": "pm"}, {"signup_date": "4/8/2023", "company": "Wonka  Co", "department": "Produtt", "job_title": "Prod Mlr"}, {"signup_date": "8/7/2023", "company": "Hooli", "department": " ACCOUNTING ", "job_title": "Vp"}, {"signup_date": "1/23/2023", "company": "Hooli", "department": "Finance  ", "job_title": "Product Manager"}, {"signup_date": "2023-01-07", "company": "Umbrella", "department": "finance", "job_title": "Dir."}, {"signup_date": "3/9/2023", "company": "Vehement", "department": " PM ", "job_title": "vice president"}, {"signup_date": "11/11/2023", "company": "Initech", "department": "Fin", "job_title": "dir."}, {"signup_date": "45275", "company": "Acme  Inc", "department": "   Finance", "job_title": "Account Executive"}, {"signup_date": "", "company": "", "department": "", "job_title": ""}], "clean": [{"signup_date": "2023-02-15", "company": "Umbrella", "department": "Product", "job_title": "Director"}, {"signup_date": "2023-05-13", "company": "Vehement", "department": "Finance", "job_title": "Product Manager"}, {"signup_date": "2023-11-18", "company": "Stark Industries", "department": "Finance", "job_title": "Vice President"}, {"signup_date": "2023-08-01", "company": "Wonka Co", "department": "Operations", "job_title": "Director"}, {"signup_date": "2023-12-06", "company": "Vehement", "department": "Operations", "job_title": "Vice President"}, {"signup_date": "2023-03-10", "company": "Umbrella", "department": "Product", "job_title": "Product Manager"}, {"signup_date": "2023-02-27", "company": "Hooli", "department": "Product", "job_title": "Product Manager"}, {"signup_date": "2023-07-11", "company": "Hooli", "department": "Finance", "job_title": "Product Manager"}, {"signup_date": "2023-05-12", "company": "Acme Inc", "department": "Operations", "job_title": "Vice President"}, {"signup_date": "2023-07-24", "company": "Globex", "department": "Finance", "job_title": "Director"}, {"signup_date": "2023-09-12", "company": "Stark Industries", "department": "Finance", "job_title": "Vice President"}, {"signup_date": "2023-01-17", "company": "Umbrella", "department": "Finance", "job_title": "Product Manager"}, {"signup_date": "2023-02-24", "company": "Soylent Corp", "department": "Operations", "job_title": "Product Manager"}, {"signup_date": "2023-04-08", "company": "Wonka Co", "department": "Product", "job_title": "Product Manager"}, {"signup_date": "2023-08-07", "company": "Hooli", "department": "Finance", "job_title": "Vice President"}, {"signup_date": "2023-01-23", "company": "Hooli", "department": "Finance", "job_title": "Product Manager"}, {"signup_date": "2023-01-07", "company": "Umbrella", "department": "Finance", "job_title": "Director"}, {"signup_date": "2023-03-09", "company": "Vehement", "department": "Product", "job_title": "Vice President"}, {"signup_date": "2023-11-11", "company": "Initech", "department": "Finance", "job_title": "Director"}, {"signup_date": "2023-12-15", "company": "Acme Inc", "department": "Finance", "job_title": "Account Executive"}], "dirty_cols": ["signup_date", "company", "department", "job_title"], "clean_cols": ["signup_date", "company", "department", "job_title"], "plan": {"dataset_summary": "21 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"product": "Product", "Accounting": "Finance", "Ops": "Operations", "pm": "Product", "FINANCE": "Finance", "ops": "Operations", "Produtt": "Product", "ACCOUNTING": "Finance", "finance": "Finance", "PM": "Product", "Fin": "Finance"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Dir.": "Director", "prod manaer": "Product Manager", "Vp": "Vice President", "DIR.": "Director", "vice pres": "Vice President", "PM": "Product Manager", "PRODUCT MANAGER": "Product Manager", "VP": "Vice President", "V.P.": "Vice President", "Prod Manager": "Product Manager", "pm": "Product Manager", "Prod Mlr": "Product Manager", "vice president": "Vice President", "dir.": "Director"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "  gel", "email": " dlbml@test.org ", "status": "P3", "name": "   Grace Fischer", "rate": "20.5%", "job_title": "CTO", "extra": ""}, {"currency": "GEL", "email": "FLAAI@EXAMPLE.COM ", "status": "low", "name": "   Sara Reyes", "rate": "50.8%", "job_title": "CTO", "extra": ""}, {"currency": "ETB", "email": "MOIIHGJ@EXAMPLE.COM", "status": "low", "name": "Judy Fischer", "rate": "91.9%", "job_title": "Vice Pres", "extra": ""}, {"currency": "Laei", "email": "OEOC@TEST.ORG", "status": "HIGH", "name": "Karl  Cruz", "rate": "53.6%", "job_title": "pm", "extra": ""}, {"currency": "Boliviano", "email": "   aiiefaj@corp.io", "status": "H", "name": "  Ivan Diaz", "rate": "72.7%", "job_title": " Cto ", "extra": ""}, {"currency": "gel", "email": "KBGILB@MAIL.COM", "status": "low", "name": "Carol Brandt", "rate": "96.6%", "job_title": "VP", "extra": ""}, {"currency": " lri ", "email": "NHBNI@CORP.IO", "status": "low", "name": "Mona Reyes ", "rate": "36.2%", "job_title": "pm ", "extra": ""}, {"currency": "etb", "email": "NFGI@TEST.ORG", "status": "low", "name": "   Alice Lee", "rate": "9.7%", "job_title": "Prod Manager", "extra": ""}, {"currency": " BOB", "email": "   NPEF@TEST.ORG", "status": " l ", "name": "Alice Lee   ", "rate": "57.1%", "job_title": "Vice President", "extra": ""}, {"currency": "ethiopian birr", "email": "   njjgnb@corp.io", "status": "HIGH", "name": "Frank Johnson  ", "rate": "92.0%", "job_title": "VICE PRES", "extra": ""}, {"currency": " etb ", "email": " LJAG@EXAMPLE.COM ", "status": "l", "name": "Priya Johnson", "rate": "67.2%", "job_title": "V.P.", "extra": ""}, {"currency": "  Ethiopian Birr", "email": "  IBJDAB@CORP.IO", "status": "P3", "name": "  Carol Petrov", "rate": "11.9%", "job_title": "prod mgr", "extra": ""}, {"currency": "  Cuban Peso", "email": " ekehi@corp.io ", "status": "Low", "name": "Heidi Diaz", "rate": "4.9%", "job_title": "   Vice Pres", "extra": ""}, {"currency": "boliviano", "email": "MMHMMK@MAIL.COM", "status": "High", "name": "Frank Lee", "rate": "36.2%", "job_title": "CTO", "extra": ""}, {"currency": "Cuban Peso", "email": "impmcbk@example.com   ", "status": "P1", "name": "Karl Fischer", "rate": "85.7%", "job_title": " Chief Technology Officer", "extra": ""}, {"currency": "Bob", "email": "AGIL@MAIL.COM", "status": "   HIGH", "name": "Karl Fischer", "rate": "51.3%", "job_title": "VICE PRESIDENT", "extra": ""}, {"currency": " ethiopian birr", "email": "ioagmjp@example.com   ", "status": " LOW ", "name": "Eve Park", "rate": "8.7%", "job_title": "Vice Pres ", "extra": ""}, {"currency": "Bohiviano", "email": "aeneei@corp.io", "status": " High ", "name": "   Frank Novak", "rate": "76.3%", "job_title": "CHIEF TECHNOLOGY OFFICER", "extra": ""}, {"currency": "boliviano", "email": " fjfa@example.com", "status": "High", "name": "Judy Brandt ", "rate": "96.0%", "job_title": "vp", "extra": ""}, {"currency": " ethiopian birr", "email": "ioagmjp@example.com   ", "status": " LOW ", "name": "Eve Park", "rate": "8.7%", "job_title": "Vice Pres ", "extra": ""}, {"currency": " lri ", "email": "NHBNI@CORP.IO", "status": "low", "name": "Mona Reyes ", "rate": "36.2%", "job_title": "pm ", "extra": ""}], "clean": [{"currency": "GEL", "email": "dlbml@test.org", "status": "Low", "name": "Grace Fischer", "rate": 0.205, "job_title": "Chief Technology Officer"}, {"currency": "GEL", "email": "flaai@example.com", "status": "Low", "name": "Sara Reyes", "rate": 0.508, "job_title": "Chief Technology Officer"}, {"currency": "ETB", "email": "moiihgj@example.com", "status": "Low", "name": "Judy Fischer", "rate": 0.919, "job_title": "Vice President"}, {"currency": "GEL", "email": "oeoc@test.org", "status": "High", "name": "Karl Cruz", "rate": 0.536, "job_title": "Product Manager"}, {"currency": "BOB", "email": "aiiefaj@corp.io", "status": "High", "name": "Ivan Diaz", "rate": 0.727, "job_title": "Chief Technology Officer"}, {"currency": "GEL", "email": "kbgilb@mail.com", "status": "Low", "name": "Carol Brandt", "rate": 0.966, "job_title": "Vice President"}, {"currency": "GEL", "email": "nhbni@corp.io", "status": "Low", "name": "Mona Reyes", "rate": 0.36200000000000004, "job_title": "Product Manager"}, {"currency": "ETB", "email": "nfgi@test.org", "status": "Low", "name": "Alice Lee", "rate": 0.09699999999999999, "job_title": "Product Manager"}, {"currency": "BOB", "email": "npef@test.org", "status": "Low", "name": "Alice Lee", "rate": 0.5710000000000001, "job_title": "Vice President"}, {"currency": "ETB", "email": "njjgnb@corp.io", "status": "High", "name": "Frank Johnson", "rate": 0.92, "job_title": "Vice President"}, {"currency": "ETB", "email": "ljag@example.com", "status": "Low", "name": "Priya Johnson", "rate": 0.672, "job_title": "Vice President"}, {"currency": "ETB", "email": "ibjdab@corp.io", "status": "Low", "name": "Carol Petrov", "rate": 0.11900000000000001, "job_title": "Product Manager"}, {"currency": "CUP", "email": "ekehi@corp.io", "status": "Low", "name": "Heidi Diaz", "rate": 0.049, "job_title": "Vice President"}, {"currency": "BOB", "email": "mmhmmk@mail.com", "status": "High", "name": "Frank Lee", "rate": 0.36200000000000004, "job_title": "Chief Technology Officer"}, {"currency": "CUP", "email": "impmcbk@example.com", "status": "High", "name": "Karl Fischer", "rate": 0.857, "job_title": "Chief Technology Officer"}, {"currency": "BOB", "email": "agil@mail.com", "status": "High", "name": "Karl Fischer", "rate": 0.513, "job_title": "Vice President"}, {"currency": "ETB", "email": "ioagmjp@example.com", "status": "Low", "name": "Eve Park", "rate": 0.087, "job_title": "Vice President"}, {"currency": "BOB", "email": "aeneei@corp.io", "status": "High", "name": "Frank Novak", "rate": 0.763, "job_title": "Chief Technology Officer"}, {"currency": "BOB", "email": "fjfa@example.com", "status": "High", "name": "Judy Brandt", "rate": 0.96, "job_title": "Vice President"}], "dirty_cols": ["currency", "email", "status", "name", "rate", "job_title", "extra"], "clean_cols": ["currency", "email", "status", "name", "rate", "job_title"], "plan": {"dataset_summary": "21 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"gel": "GEL", "Laei": "GEL", "Boliviano": "BOB", "lri": "GEL", "etb": "ETB", "ethiopian birr": "ETB", "Ethiopian Birr": "ETB", "Cuban Peso": "CUP", "boliviano": "BOB", "Bob": "BOB", "Bohiviano": "BOB"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"P3": "Low", "low": "Low", "HIGH": "High", "H": "High", "l": "Low", "P1": "High", "LOW": "Low"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CTO": "Chief Technology Officer", "Vice Pres": "Vice President", "pm": "Product Manager", "Cto": "Chief Technology Officer", "VP": "Vice President", "Prod Manager": "Product Manager", "VICE PRES": "Vice President", "V.P.": "Vice President", "prod mgr": "Product Manager", "VICE PRESIDENT": "Vice President", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "vp": "Vice President"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "Eng Manager", "city": "soyo", "company": "Globex ", "currency": "Ghbraltar Pound", "rate": "22.7%", "signup_date": "45101", "unnamed": ""}, {"job_title": "eng manager ", "city": "aonla", "company": "Cyberdyne", "currency": "HONG KONG  DOLLAR", "rate": "25.2%", "signup_date": "2023-03-06", "unnamed": ""}, {"job_title": "EngzMgr", "city": "pekin", "company": "Stark  Industries", "currency": "HKD", "rate": "15.0%", "signup_date": "19 Mar 2023", "unnamed": ""}, {"job_title": "   Account Executive", "city": "   phitsanulok", "company": "Soylent Corp", "currency": " hong kong dollar", "rate": "14.3%", "signup_date": "2023-07-25", "unnamed": ""}, {"job_title": "   engineering manager", "city": "Ptitsanulok", "company": "Soylent Corp  ", "currency": "hong kong dollar", "rate": "33.8%", "signup_date": "2/4/2023", "unnamed": ""}, {"job_title": "  Engineering Mgr", "city": "Aonla", "company": "Acme Inc", "currency": "Nakfa", "rate": "38.0%", "signup_date": "2023-08-08", "unnamed": ""}, {"job_title": "Acct Exec", "city": "phitsanulok", "company": "Umbrella", "currency": "ern", "rate": "96.1%", "signup_date": "45090", "unnamed": ""}, {"job_title": "Engieering Manager", "city": "Soyo", "company": "Stark Industries", "currency": "GIBRALTAR POUND", "rate": "62.4%", "signup_date": "2023-03-08", "unnamed": ""}, {"job_title": "Engineering Mgr", "city": " PEKIN", "company": "Acme Inc", "currency": "ern", "rate": "34.2%", "signup_date": "2023-12-18", "unnamed": ""}, {"job_title": "Account Executive   ", "city": "Phitsanulok", "company": " Vehement ", "currency": "  Hong Kong Dollar", "rate": "5.4%", "signup_date": "5/27/2023", "unnamed": ""}, {"job_title": "AE", "city": "PHITSANULOK", "company": "Wonka Co", "currency": "  Hong Kng Dollar", "rate": "45.2%", "signup_date": "12/11/2023", "unnamed": ""}, {"job_title": "Eng Manager", "city": "Phitsanulok", "company": "   Cyberdyne", "currency": "Hong  Kong Dollar", "rate": "98.4%", "signup_date": "45278", "unnamed": ""}, {"job_title": "ENG MANAGER", "city": "Sojo", "company": "Cyberdyne", "currency": "Nakfa", "rate": "97.2%", "signup_date": "4 Sep 2023", "unnamed": ""}, {"job_title": "AE", "city": "PEKIN", "company": "Wonka  Co", "currency": "nakfa", "rate": "81.8%", "signup_date": "45250", "unnamed": ""}, {"job_title": "ENGINEERING MANAGER", "city": "AONLA", "company": "Acme Inc   ", "currency": "GIP", "rate": "4.1%", "signup_date": "6/24/2023", "unnamed": ""}, {"job_title": "enginereing manager", "city": "Aonla", "company": "Soylent  Corp", "currency": "Hong Kong Dollar", "rate": "35.8%", "signup_date": "2023-06-27", "unnamed": ""}, {"job_title": "Eng Manager", "city": "Soyo", "company": "Umbrella", "currency": "HKD", "rate": "98.8%", "signup_date": "45157", "unnamed": ""}, {"job_title": "engineering mgr", "city": "Phitsanulok", "company": "Acme Inc  ", "currency": "Hong Kong Dollar   ", "rate": "54.4%", "signup_date": "2023-05-26", "unnamed": ""}, {"job_title": "Eng Manager", "city": "Aonla", "company": "   Initech", "currency": "Nakfa", "rate": "91.8%", "signup_date": "10 Dec 2023", "unnamed": ""}, {"job_title": "Account Executive", "city": "Aonla", "company": " Initech", "currency": "hkd", "rate": "29.2%", "signup_date": "7/23/2023", "unnamed": ""}, {"job_title": "Eng Manager", "city": "Soyo", "company": "Hooli", "currency": "   hong kong dollar", "rate": "17.9%", "signup_date": "11/8/2023", "unnamed": ""}, {"job_title": "ENGINEERING MGR", "city": "Phitsanulok", "company": "Vehement", "currency": "ERN", "rate": "21.1%", "signup_date": "4/20/2023", "unnamed": ""}, {"job_title": "ENGINEERING  MANAGER", "city": "Phitsanulok", "company": " Globex ", "currency": "Nakfa", "rate": "17.0%", "signup_date": "16 Mar 2023", "unnamed": ""}, {"job_title": "Eng Mgr", "city": "  Pekin", "company": "Globex", "currency": "ERN ", "rate": "29.1%", "signup_date": "1/14/2023", "unnamed": ""}, {"job_title": "Ae", "city": " Aonla ", "company": "Hooli", "currency": "GIP ", "rate": "25.4%", "signup_date": "2023-05-11", "unnamed": ""}, {"job_title": " Engineering Manager", "city": "Aonla", "company": "  Globex", "currency": "ERN", "rate": "72.8%", "signup_date": "45185", "unnamed": ""}, {"job_title": "Acount Executive", "city": "pekin", "company": " Initech ", "currency": "Hong Kong Dollar", "rate": "32.3%", "signup_date": "45144", "unnamed": ""}, {"job_title": " AE ", "city": "Soyo", "company": "Acme Inc", "currency": " ERN ", "rate": "49.4%", "signup_date": "6/13/2023", "unnamed": ""}, {"job_title": "acct exec", "city": "Pekin", "company": "   Acme Inc", "currency": "Gibraltar Pound", "rate": "87.8%", "signup_date": "7/25/2023", "unnamed": ""}, {"job_title": "ACCT EXEC", "city": "Soyo", "company": "Cyberdyne", "currency": "Hong Kong Dollar", "rate": "81.9%", "signup_date": "5/21/2023", "unnamed": ""}, {"job_title": "ACCOUNT EXECUTIVE", "city": "Soyo", "company": "Umbrella", "currency": "Nakfa", "rate": "39.8%", "signup_date": "45245", "unnamed": ""}, {"job_title": "eng mgr", "city": "  pekin", "company": "Hooli", "currency": "hkd", "rate": "32.1%", "signup_date": "9/24/2023", "unnamed": ""}, {"job_title": "ENG MGR", "city": "Aonla", "company": "Stark Industries", "currency": "Hkd", "rate": "90.5%", "signup_date": "45177", "unnamed": ""}, {"job_title": "Account Executive", "city": " Pekin ", "company": "Initech ", "currency": "HKD", "rate": "15.2%", "signup_date": "45028", "unnamed": ""}, {"job_title": "ENG MAAGER", "city": "Soyo", "company": " Wonka Co", "currency": "gibraltar pound", "rate": "47.7%", "signup_date": "7/2/2023", "unnamed": ""}, {"job_title": "   ae", "city": "Aonla", "company": "Acme Inc", "currency": "Nakfa", "rate": "71.4%", "signup_date": "9/8/2023", "unnamed": ""}, {"job_title": "ae", "city": "   Pekin", "company": "Stark Industries", "currency": "Hong  Kong Dollar", "rate": "74.7%", "signup_date": "2023-12-15", "unnamed": ""}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "  Stark Industries", "currency": "HONG KONG DOLLAR", "rate": "0.6%", "signup_date": "6/7/2023", "unnamed": ""}, {"job_title": "eng mgr", "city": "   Phitsanulok", "company": "Globex", "currency": "ERN", "rate": "18.4%", "signup_date": "23 Jul 2023", "unnamed": ""}, {"job_title": "Engineering Manager", "city": " PHITSANULOK ", "company": " Cyberdyne ", "currency": "ERN", "rate": "42.0%", "signup_date": "1 Feb 2023", "unnamed": ""}, {"job_title": "AE", "city": "Phitsanulok", "company": "Stark Industries", "currency": "ern", "rate": "99.7%", "signup_date": "17 Jun 2023", "unnamed": ""}, {"job_title": "ACCOUNT EXECUTIVE", "city": " Pbkin ", "company": "Acme Inc", "currency": "Nakfa", "rate": "68.4%", "signup_date": "4/23/2023", "unnamed": ""}, {"job_title": "   ae", "city": " Soyo ", "company": "Soylent Corp", "currency": "ERN", "rate": "77.4%", "signup_date": "27 Jan 2023", "unnamed": ""}, {"job_title": "Account Executive", "city": "phitsanulok", "company": "Initech", "currency": "gibraltar pound   ", "rate": "46.1%", "signup_date": "45153", "unnamed": ""}, {"job_title": "ENG  MGR", "city": " Aonla ", "company": "Vehement", "currency": "GIP", "rate": "97.1%", "signup_date": "4/13/2023", "unnamed": ""}, {"job_title": "Eng Mgr", "city": "Aotla", "company": "Acme Inc", "currency": "Gibraltar Pound", "rate": "81.9%", "signup_date": "2023-01-19", "unnamed": ""}, {"job_title": "Eng Manager  ", "city": "Soyo", "company": "Vehement ", "currency": "Nakfa", "rate": "93.4%", "signup_date": "20 Nov 2023", "unnamed": ""}, {"job_title": " AE", "city": "Soyo", "company": "Hooli", "currency": " nakfa ", "rate": "15.8%", "signup_date": "7 Nov 2023", "unnamed": ""}, {"job_title": "AE", "city": "Pekin", "company": "Cyberdyne   ", "currency": "Gibraltar Pound   ", "rate": "32.4%", "signup_date": "5/17/2023", "unnamed": ""}, {"job_title": "Acct Exec", "city": "   AONLA", "company": "Vehement", "currency": "hong kong dollar", "rate": "31.2%", "signup_date": "2023-09-12", "unnamed": ""}, {"job_title": "Eng Mgr", "city": "soyo", "company": "   Cyberdyne", "currency": "ERN", "rate": "87.8%", "signup_date": "11/18/2023", "unnamed": ""}, {"job_title": "Eng Mgr", "city": "Pekin", "company": "  Hooli", "currency": "GIP", "rate": "92.2%", "signup_date": "45153", "unnamed": ""}, {"job_title": "Ae", "city": "PEKIN", "company": "Umbrella", "currency": " NAKFA ", "rate": "46.4%", "signup_date": "2023-04-19", "unnamed": ""}, {"job_title": "AE", "city": "Phitsanulok", "company": " Hooli ", "currency": "GIP", "rate": "41.3%", "signup_date": "2023-04-21", "unnamed": ""}, {"job_title": "AE  ", "city": "   aonla", "company": "Hooli", "currency": "GIBRALTAR POUND", "rate": "14.9%", "signup_date": "3/8/2023", "unnamed": ""}, {"job_title": "Eng Mgr", "city": " SOYO", "company": "Umbrella", "currency": "HKD", "rate": "9.7%", "signup_date": "2023-04-26", "unnamed": ""}, {"job_title": "acct exec", "city": "Pekin", "company": "   Acme Inc", "currency": "Gibraltar Pound", "rate": "87.8%", "signup_date": "7/25/2023", "unnamed": ""}, {"job_title": "", "city": "", "company": "", "currency": "", "rate": "", "signup_date": "", "unnamed": ""}], "clean": [{"job_title": "Engineering Manager", "city": "Soyo", "company": "Globex", "currency": "GIP", "rate": 0.22699999999999998, "signup_date": "2023-06-24"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Cyberdyne", "currency": "HKD", "rate": 0.252, "signup_date": "2023-03-06"}, {"job_title": "Engineering Manager", "city": "Pekin", "company": "Stark Industries", "currency": "HKD", "rate": 0.15, "signup_date": "2023-03-19"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Soylent Corp", "currency": "HKD", "rate": 0.14300000000000002, "signup_date": "2023-07-25"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Soylent Corp", "currency": "HKD", "rate": 0.33799999999999997, "signup_date": "2023-02-04"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Acme Inc", "currency": "ERN", "rate": 0.38, "signup_date": "2023-08-08"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Umbrella", "currency": "ERN", "rate": 0.961, "signup_date": "2023-06-13"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Stark Industries", "currency": "GIP", "rate": 0.624, "signup_date": "2023-03-08"}, {"job_title": "Engineering Manager", "city": "Pekin", "company": "Acme Inc", "currency": "ERN", "rate": 0.342, "signup_date": "2023-12-18"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Vehement", "currency": "HKD", "rate": 0.054000000000000006, "signup_date": "2023-05-27"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Wonka Co", "currency": "HKD", "rate": 0.452, "signup_date": "2023-12-11"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Cyberdyne", "currency": "HKD", "rate": 0.9840000000000001, "signup_date": "2023-12-18"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Cyberdyne", "currency": "ERN", "rate": 0.972, "signup_date": "2023-09-04"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Wonka Co", "currency": "ERN", "rate": 0.818, "signup_date": "2023-11-20"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Acme Inc", "currency": "GIP", "rate": 0.040999999999999995, "signup_date": "2023-06-24"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Soylent Corp", "currency": "HKD", "rate": 0.358, "signup_date": "2023-06-27"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Umbrella", "currency": "HKD", "rate": 0.988, "signup_date": "2023-08-19"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Acme Inc", "currency": "HKD", "rate": 0.544, "signup_date": "2023-05-26"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Initech", "currency": "ERN", "rate": 0.9179999999999999, "signup_date": "2023-12-10"}, {"job_title": "Account Executive", "city": "Aonla", "company": "Initech", "currency": "HKD", "rate": 0.292, "signup_date": "2023-07-23"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Hooli", "currency": "HKD", "rate": 0.179, "signup_date": "2023-11-08"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Vehement", "currency": "ERN", "rate": 0.21100000000000002, "signup_date": "2023-04-20"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Globex", "currency": "ERN", "rate": 0.17, "signup_date": "2023-03-16"}, {"job_title": "Engineering Manager", "city": "Pekin", "company": "Globex", "currency": "ERN", "rate": 0.29100000000000004, "signup_date": "2023-01-14"}, {"job_title": "Account Executive", "city": "Aonla", "company": "Hooli", "currency": "GIP", "rate": 0.254, "signup_date": "2023-05-11"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Globex", "currency": "ERN", "rate": 0.728, "signup_date": "2023-09-16"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Initech", "currency": "HKD", "rate": 0.32299999999999995, "signup_date": "2023-08-06"}, {"job_title": "Account Executive", "city": "Soyo", "company": "Acme Inc", "currency": "ERN", "rate": 0.494, "signup_date": "2023-06-13"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Acme Inc", "currency": "GIP", "rate": 0.878, "signup_date": "2023-07-25"}, {"job_title": "Account Executive", "city": "Soyo", "company": "Cyberdyne", "currency": "HKD", "rate": 0.8190000000000001, "signup_date": "2023-05-21"}, {"job_title": "Account Executive", "city": "Soyo", "company": "Umbrella", "currency": "ERN", "rate": 0.39799999999999996, "signup_date": "2023-11-15"}, {"job_title": "Engineering Manager", "city": "Pekin", "company": "Hooli", "currency": "HKD", "rate": 0.321, "signup_date": "2023-09-24"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Stark Industries", "currency": "HKD", "rate": 0.905, "signup_date": "2023-09-08"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Initech", "currency": "HKD", "rate": 0.152, "signup_date": "2023-04-12"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Wonka Co", "currency": "GIP", "rate": 0.47700000000000004, "signup_date": "2023-07-02"}, {"job_title": "Account Executive", "city": "Aonla", "company": "Acme Inc", "currency": "ERN", "rate": 0.7140000000000001, "signup_date": "2023-09-08"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Stark Industries", "currency": "HKD", "rate": 0.747, "signup_date": "2023-12-15"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Stark Industries", "currency": "HKD", "rate": 0.006, "signup_date": "2023-06-07"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Globex", "currency": "ERN", "rate": 0.184, "signup_date": "2023-07-23"}, {"job_title": "Engineering Manager", "city": "Phitsanulok", "company": "Cyberdyne", "currency": "ERN", "rate": 0.42, "signup_date": "2023-02-01"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Stark Industries", "currency": "ERN", "rate": 0.997, "signup_date": "2023-06-17"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Acme Inc", "currency": "ERN", "rate": 0.684, "signup_date": "2023-04-23"}, {"job_title": "Account Executive", "city": "Soyo", "company": "Soylent Corp", "currency": "ERN", "rate": 0.774, "signup_date": "2023-01-27"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Initech", "currency": "GIP", "rate": 0.461, "signup_date": "2023-08-15"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Vehement", "currency": "GIP", "rate": 0.971, "signup_date": "2023-04-13"}, {"job_title": "Engineering Manager", "city": "Aonla", "company": "Acme Inc", "currency": "GIP", "rate": 0.8190000000000001, "signup_date": "2023-01-19"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Vehement", "currency": "ERN", "rate": 0.934, "signup_date": "2023-11-20"}, {"job_title": "Account Executive", "city": "Soyo", "company": "Hooli", "currency": "ERN", "rate": 0.158, "signup_date": "2023-11-07"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Cyberdyne", "currency": "GIP", "rate": 0.324, "signup_date": "2023-05-17"}, {"job_title": "Account Executive", "city": "Aonla", "company": "Vehement", "currency": "HKD", "rate": 0.312, "signup_date": "2023-09-12"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Cyberdyne", "currency": "ERN", "rate": 0.878, "signup_date": "2023-11-18"}, {"job_title": "Engineering Manager", "city": "Pekin", "company": "Hooli", "currency": "GIP", "rate": 0.922, "signup_date": "2023-08-15"}, {"job_title": "Account Executive", "city": "Pekin", "company": "Umbrella", "currency": "ERN", "rate": 0.46399999999999997, "signup_date": "2023-04-19"}, {"job_title": "Account Executive", "city": "Phitsanulok", "company": "Hooli", "currency": "GIP", "rate": 0.413, "signup_date": "2023-04-21"}, {"job_title": "Account Executive", "city": "Aonla", "company": "Hooli", "currency": "GIP", "rate": 0.149, "signup_date": "2023-03-08"}, {"job_title": "Engineering Manager", "city": "Soyo", "company": "Umbrella", "currency": "HKD", "rate": 0.09699999999999999, "signup_date": "2023-04-26"}], "dirty_cols": ["job_title", "city", "company", "currency", "rate", "signup_date", "unnamed"], "clean_cols": ["job_title", "city", "company", "currency", "rate", "signup_date"], "plan": {"dataset_summary": "58 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Eng Manager": "Engineering Manager", "eng manager": "Engineering Manager", "EngzMgr": "Engineering Manager", "engineering manager": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "Acct Exec": "Account Executive", "Engieering Manager": "Engineering Manager", "AE": "Account Executive", "ENG MANAGER": "Engineering Manager", "ENGINEERING MANAGER": "Engineering Manager", "enginereing manager": "Engineering Manager", "engineering mgr": "Engineering Manager", "ENGINEERING MGR": "Engineering Manager", "Eng Mgr": "Engineering Manager", "Ae": "Account Executive", "Acount Executive": "Account Executive", "acct exec": "Account Executive", "ACCT EXEC": "Account Executive", "ACCOUNT EXECUTIVE": "Account Executive", "eng mgr": "Engineering Manager", "ENG MGR": "Engineering Manager", "ENG MAAGER": "Engineering Manager", "ae": "Account Executive"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"soyo": "Soyo", "aonla": "Aonla", "pekin": "Pekin", "phitsanulok": "Phitsanulok", "Ptitsanulok": "Phitsanulok", "PEKIN": "Pekin", "PHITSANULOK": "Phitsanulok", "Sojo": "Soyo", "AONLA": "Aonla", "Pbkin": "Pekin", "Aotla": "Aonla", "SOYO": "Soyo"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ghbraltar Pound": "GIP", "HONG KONG DOLLAR": "HKD", "hong kong dollar": "HKD", "Nakfa": "ERN", "ern": "ERN", "GIBRALTAR POUND": "GIP", "Hong Kong Dollar": "HKD", "Hong Kng Dollar": "HKD", "nakfa": "ERN", "hkd": "HKD", "Gibraltar Pound": "GIP", "Hkd": "HKD", "gibraltar pound": "GIP", "NAKFA": "ERN"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}], "flags": []}}
{"dirty": [{"name": "Alice Adams  ", "unit": "miles   ", "is_active": "TRUE", "state": "KY", "city": "BA"}, {"name": "  Grace Diaz", "unit": "Kg ", "is_active": "0", "state": "   HI", "city": "fasano"}, {"name": " Karl Ali", "unit": " MILES ", "is_active": "F", "state": "Hawaii", "city": "Bullhead City"}, {"name": " Priya Reyes", "unit": "Kilogram", "is_active": "No", "state": "Hawaii", "city": "Esparza"}, {"name": "Omar Cruz", "unit": "ml", "is_active": "No", "state": "nj", "city": " Bitam "}, {"name": "Karl Park", "unit": "MI ", "is_active": "F", "state": "   New Jersey", "city": "Buenos  Aires"}, {"name": "Priya Khan  ", "unit": "ml", "is_active": "F", "state": "New Jersey", "city": "faasno"}, {"name": "Eve Johnson", "unit": "Centimeter ", "is_active": "N", "state": "NEVADA", "city": "espaaza"}, {"name": "Heidi Diaz", "unit": "centimeters", "is_active": "TRUE", "state": "New Jersey", "city": "FASANO"}, {"name": "   Eve Petrov", "unit": "mile", "is_active": "N", "state": "nj", "city": "buenos aires "}, {"name": "Priya Khan  ", "unit": "ml", "is_active": "F", "state": "New Jersey", "city": "faasno"}, {"name": "Omar Cruz", "unit": "ml", "is_active": "No", "state": "nj", "city": " Bitam "}], "clean": [{"name": "Alice Adams", "unit": "mi", "is_active": true, "state": "Kentucky", "city": "Buenos Aires"}, {"name": "Grace Diaz", "unit": "kg", "is_active": false, "state": "Hawaii", "city": "Fasano"}, {"name": "Karl Ali", "unit": "mi", "is_active": false, "state": "Hawaii", "city": "Bullhead City"}, {"name": "Priya Reyes", "unit": "kg", "is_active": false, "state": "Hawaii", "city": "Esparza"}, {"name": "Omar Cruz", "unit": "mL", "is_active": false, "state": "New Jersey", "city": "Bitam"}, {"name": "Karl Park", "unit": "mi", "is_active": false, "state": "New Jersey", "city": "Buenos Aires"}, {"name": "Priya Khan", "unit": "mL", "is_active": false, "state": "New Jersey", "city": "Fasano"}, {"name": "Eve Johnson", "unit": "cm", "is_active": false, "state": "Nevada", "city": "Esparza"}, {"name": "Heidi Diaz", "unit": "cm", "is_active": true, "state": "New Jersey", "city": "Fasano"}, {"name": "Eve Petrov", "unit": "mi", "is_active": false, "state": "New Jersey", "city": "Buenos Aires"}], "dirty_cols": ["name", "unit", "is_active", "state", "city"], "clean_cols": ["name", "unit", "is_active", "state", "city"], "plan": {"dataset_summary": "12 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miles": "mi", "Kg": "kg", "MILES": "mi", "Kilogram": "kg", "ml": "mL", "MI": "mi", "Centimeter": "cm", "centimeters": "cm", "mile": "mi"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KY": "Kentucky", "HI": "Hawaii", "nj": "New Jersey", "NEVADA": "Nevada"}, "rationale": "Unified 4 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BA": "Buenos Aires", "fasano": "Fasano", "faasno": "Fasano", "espaaza": "Esparza", "FASANO": "Fasano", "buenos aires": "Buenos Aires"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"currency": "djiobuti franc  ", "phone": "537-962-0809", "industry": "Health", "signup_date": "2 Nov 2023", "email": "LCHDE@CORP.IO", "col_x": ""}, {"currency": "Pula", "phone": "579.800.5781", "industry": "telco", "signup_date": "5/14/2023", "email": "   kmaaikg@corp.io", "col_x": ""}, {"currency": "BWP", "phone": "5819422310", "industry": "  Travel", "signup_date": "45179", "email": "IHHJ@EXAMPLE.COM", "col_x": ""}, {"currency": "Pula", "phone": "579-245-4380", "industry": " healthcare", "signup_date": "3/16/2023", "email": "FGCDIHH@EXAMPLE.COM  ", "col_x": ""}, {"currency": "  Egyptian Pound", "phone": "564-996-1154", "industry": " TELCO", "signup_date": "2023-02-01", "email": "FLKL@MAIL.COM", "col_x": ""}, {"currency": "Egyptian Pound", "phone": "(574)341-6591", "industry": "Halth", "signup_date": "45098", "email": "OJJBL@MAIL.COM", "col_x": ""}, {"currency": "egp", "phone": "503.713.7686", "industry": "Industrial", "signup_date": "2023-01-16", "email": "BDAIDEK@EXAMPLE.COM", "col_x": ""}, {"currency": "BWP", "phone": "579.127.0872", "industry": "Tourism", "signup_date": "28 Sep 2023", "email": "flga@test.org", "col_x": ""}, {"currency": " PULA ", "phone": "541-216-0231", "industry": "Telecommunications", "signup_date": "12/11/2023", "email": "  gofpcph@example.com", "col_x": ""}, {"currency": "BWP", "phone": "(529)211-3825", "industry": "telecommunications", "signup_date": "2023-10-26", "email": " ECMH@CORP.IO ", "col_x": ""}, {"currency": "BWP", "phone": "557.745.7445", "industry": "Hotels & Travel", "signup_date": "2023-10-23", "email": "IBCGIK@EXAMPLE.COM", "col_x": ""}, {"currency": "Djibouti Frnac", "phone": "587.335.3796", "industry": "Telecommunications", "signup_date": "2023-12-11", "email": "pmmmlmf@test.org", "col_x": ""}, {"currency": " DJF ", "phone": "530.033.4839", "industry": "mfg", "signup_date": "14 Sep 2023", "email": " gdoea@example.com", "col_x": ""}, {"currency": "Pula", "phone": "524.644.3428", "industry": "medical", "signup_date": "1/25/2023", "email": "mpjcne@test.org", "col_x": ""}, {"currency": "BWP", "phone": "535-343-2325", "industry": "Mfg", "signup_date": "4/10/2023", "email": "NLKNDE@CORP.IO", "col_x": ""}, {"currency": "DJF", "phone": "5495318961", "industry": "Healthcare", "signup_date": "5 Aug 2023", "email": "JGAA@CORP.IO", "col_x": ""}, {"currency": "BWP", "phone": "(514)147-3565", "industry": "Manufacturing  ", "signup_date": "45132", "email": "fpkemoo@test.org", "col_x": ""}, {"currency": "Egyptian Pound", "phone": "513.627.5805", "industry": "Mfg", "signup_date": "2023-08-25", "email": "MLDNAKP@CORP.IO   ", "col_x": ""}, {"currency": " BWP", "phone": "532.656.3051", "industry": "Healthcare", "signup_date": "45086", "email": "FPGAB@EXAMPLE.COM ", "col_x": ""}, {"currency": "  bwp", "phone": "(530)532-4986", "industry": "medical", "signup_date": "11/9/2023", "email": "lhmlac@test.org", "col_x": ""}, {"currency": "PULA", "phone": "582.417.4760", "industry": "telecommunications", "signup_date": "3/14/2023", "email": "CMKOF@TEST.ORG", "col_x": ""}, {"currency": "Egyptian Pound", "phone": "5397192334", "industry": "Telecommunications", "signup_date": "25 Feb 2023", "email": "  bdcc@example.com", "col_x": ""}, {"currency": "Egyptian Pound ", "phone": "513-337-8402", "industry": "Medical", "signup_date": "45236", "email": "hldfoce@example.com", "col_x": ""}, {"currency": "   EGP", "phone": "522.695.1477", "industry": "Hospitality", "signup_date": "2023-03-25", "email": " DCCLPB@MAIL.COM ", "col_x": ""}, {"currency": "Djf", "phone": "(599)673-1541", "industry": "Manufacturing", "signup_date": "21 Apr 2023", "email": "ajoc@corp.io", "col_x": ""}, {"currency": "", "phone": "", "industry": "", "signup_date": "", "email": "", "col_x": ""}], "clean": [{"currency": "DJF", "phone": "(537) 962-0809", "industry": "Healthcare", "signup_date": "2023-11-02", "email": "lchde@corp.io"}, {"currency": "BWP", "phone": "(579) 800-5781", "industry": "Telecommunications", "signup_date": "2023-05-14", "email": "kmaaikg@corp.io"}, {"currency": "BWP", "phone": "(581) 942-2310", "industry": "Hospitality", "signup_date": "2023-09-10", "email": "ihhj@example.com"}, {"currency": "BWP", "phone": "(579) 245-4380", "industry": "Healthcare", "signup_date": "2023-03-16", "email": "fgcdihh@example.com"}, {"currency": "EGP", "phone": "(564) 996-1154", "industry": "Telecommunications", "signup_date": "2023-02-01", "email": "flkl@mail.com"}, {"currency": "EGP", "phone": "(574) 341-6591", "industry": "Healthcare", "signup_date": "2023-06-21", "email": "ojjbl@mail.com"}, {"currency": "EGP", "phone": "(503) 713-7686", "industry": "Manufacturing", "signup_date": "2023-01-16", "email": "bdaidek@example.com"}, {"currency": "BWP", "phone": "(579) 127-0872", "industry": "Hospitality", "signup_date": "2023-09-28", "email": "flga@test.org"}, {"currency": "BWP", "phone": "(541) 216-0231", "industry": "Telecommunications", "signup_date": "2023-12-11", "email": "gofpcph@example.com"}, {"currency": "BWP", "phone": "(529) 211-3825", "industry": "Telecommunications", "signup_date": "2023-10-26", "email": "ecmh@corp.io"}, {"currency": "BWP", "phone": "(557) 745-7445", "industry": "Hospitality", "signup_date": "2023-10-23", "email": "ibcgik@example.com"}, {"currency": "DJF", "phone": "(587) 335-3796", "industry": "Telecommunications", "signup_date": "2023-12-11", "email": "pmmmlmf@test.org"}, {"currency": "DJF", "phone": "(530) 033-4839", "industry": "Manufacturing", "signup_date": "2023-09-14", "email": "gdoea@example.com"}, {"currency": "BWP", "phone": "(524) 644-3428", "industry": "Healthcare", "signup_date": "2023-01-25", "email": "mpjcne@test.org"}, {"currency": "BWP", "phone": "(535) 343-2325", "industry": "Manufacturing", "signup_date": "2023-04-10", "email": "nlknde@corp.io"}, {"currency": "DJF", "phone": "(549) 531-8961", "industry": "Healthcare", "signup_date": "2023-08-05", "email": "jgaa@corp.io"}, {"currency": "BWP", "phone": "(514) 147-3565", "industry": "Manufacturing", "signup_date": "2023-07-25", "email": "fpkemoo@test.org"}, {"currency": "EGP", "phone": "(513) 627-5805", "industry": "Manufacturing", "signup_date": "2023-08-25", "email": "mldnakp@corp.io"}, {"currency": "BWP", "phone": "(532) 656-3051", "industry": "Healthcare", "signup_date": "2023-06-09", "email": "fpgab@example.com"}, {"currency": "BWP", "phone": "(530) 532-4986", "industry": "Healthcare", "signup_date": "2023-11-09", "email": "lhmlac@test.org"}, {"currency": "BWP", "phone": "(582) 417-4760", "industry": "Telecommunications", "signup_date": "2023-03-14", "email": "cmkof@test.org"}, {"currency": "EGP", "phone": "(539) 719-2334", "industry": "Telecommunications", "signup_date": "2023-02-25", "email": "bdcc@example.com"}, {"currency": "EGP", "phone": "(513) 337-8402", "industry": "Healthcare", "signup_date": "2023-11-06", "email": "hldfoce@example.com"}, {"currency": "EGP", "phone": "(522) 695-1477", "industry": "Hospitality", "signup_date": "2023-03-25", "email": "dcclpb@mail.com"}, {"currency": "DJF", "phone": "(599) 673-1541", "industry": "Manufacturing", "signup_date": "2023-04-21", "email": "ajoc@corp.io"}], "dirty_cols": ["currency", "phone", "industry", "signup_date", "email", "col_x"], "clean_cols": ["currency", "phone", "industry", "signup_date", "email"], "plan": {"dataset_summary": "26 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"djiobuti franc": "DJF", "Pula": "BWP", "Egyptian Pound": "EGP", "egp": "EGP", "PULA": "BWP", "Djibouti Frnac": "DJF", "bwp": "BWP", "Djf": "DJF"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Health": "Healthcare", "telco": "Telecommunications", "Travel": "Hospitality", "healthcare": "Healthcare", "TELCO": "Telecommunications", "Halth": "Healthcare", "Industrial": "Manufacturing", "Tourism": "Hospitality", "telecommunications": "Telecommunications", "Hotels & Travel": "Hospitality", "mfg": "Manufacturing", "medical": "Healthcare", "Mfg": "Manufacturing", "Medical": "Healthcare"}, "rationale": "Unified 14 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"signup_date": "45204", "status": "m", "job_title": "Engineering Mgr", "name": " Omar Lee", "notes2": ""}, {"signup_date": "23 Oct 2023", "status": " melium ", "job_title": "CTO", "name": "Eve Reyes", "notes2": ""}, {"signup_date": "11/25/2023", "status": "HIGH", "job_title": "Director", "name": "Bob Smith ", "notes2": ""}, {"signup_date": "7 Feb 2023", "status": " med ", "job_title": "Engineering Mgr", "name": "Carol  Petrov", "notes2": ""}, {"signup_date": "45019", "status": "P2", "job_title": "dir", "name": "   Carol Adams", "notes2": ""}, {"signup_date": "45038", "status": "H", "job_title": "Eng Mgr", "name": "   Judy Wong", "notes2": ""}, {"signup_date": "25 May 2023", "status": "P2", "job_title": "Engineering Manager ", "name": " Eve Adams", "notes2": ""}, {"signup_date": "2023-02-19", "status": "M", "job_title": "Engineering Manager", "name": "Grace Adams", "notes2": ""}, {"signup_date": "9/23/2023", "status": "HIGH", "job_title": "CTO ", "name": "Heidi Johnson", "notes2": ""}, {"signup_date": "14 May 2023", "status": "high", "job_title": "eng manager", "name": "Frank  Fischer", "notes2": ""}, {"signup_date": "2023-05-01", "status": "hgh", "job_title": "C.T.O.", "name": "  Heidi Adams", "notes2": ""}, {"signup_date": "11/13/2023", "status": "P1", "job_title": "Dirctor", "name": "Heidi Reyes", "notes2": ""}, {"signup_date": "45116", "status": "high", "job_title": "Chief Technloogy Officer", "name": "Ivan Petrov", "notes2": ""}, {"signup_date": "2023-02-26", "status": "medium", "job_title": "engineering manager", "name": "  Judy Adams", "notes2": ""}, {"signup_date": "45125", "status": "M", "job_title": "Chief Technology Officer", "name": "Ivan Smith", "notes2": ""}], "clean": [{"signup_date": "2023-10-05", "status": "Medium", "job_title": "Engineering Manager", "name": "Omar Lee"}, {"signup_date": "2023-10-23", "status": "Medium", "job_title": "Chief Technology Officer", "name": "Eve Reyes"}, {"signup_date": "2023-11-25", "status": "High", "job_title": "Director", "name": "Bob Smith"}, {"signup_date": "2023-02-07", "status": "Medium", "job_title": "Engineering Manager", "name": "Carol Petrov"}, {"signup_date": "2023-04-03", "status": "Medium", "job_title": "Director", "name": "Carol Adams"}, {"signup_date": "2023-04-22", "status": "High", "job_title": "Engineering Manager", "name": "Judy Wong"}, {"signup_date": "2023-05-25", "status": "Medium", "job_title": "Engineering Manager", "name": "Eve Adams"}, {"signup_date": "2023-02-19", "status": "Medium", "job_title": "Engineering Manager", "name": "Grace Adams"}, {"signup_date": "2023-09-23", "status": "High", "job_title": "Chief Technology Officer", "name": "Heidi Johnson"}, {"signup_date": "2023-05-14", "status": "High", "job_title": "Engineering Manager", "name": "Frank Fischer"}, {"signup_date": "2023-05-01", "status": "High", "job_title": "Chief Technology Officer", "name": "Heidi Adams"}, {"signup_date": "2023-11-13", "status": "High", "job_title": "Director", "name": "Heidi Reyes"}, {"signup_date": "2023-07-09", "status": "High", "job_title": "Chief Technology Officer", "name": "Ivan Petrov"}, {"signup_date": "2023-02-26", "status": "Medium", "job_title": "Engineering Manager", "name": "Judy Adams"}, {"signup_date": "2023-07-18", "status": "Medium", "job_title": "Chief Technology Officer", "name": "Ivan Smith"}], "dirty_cols": ["signup_date", "status", "job_title", "name", "notes2"], "clean_cols": ["signup_date", "status", "job_title", "name"], "plan": {"dataset_summary": "15 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"m": "Medium", "melium": "Medium", "HIGH": "High", "med": "Medium", "P2": "Medium", "H": "High", "M": "Medium", "high": "High", "hgh": "High", "P1": "High", "medium": "Medium"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Engineering Mgr": "Engineering Manager", "CTO": "Chief Technology Officer", "dir": "Director", "Eng Mgr": "Engineering Manager", "eng manager": "Engineering Manager", "C.T.O.": "Chief Technology Officer", "Dirctor": "Director", "Chief Technloogy Officer": "Chief Technology Officer", "engineering manager": "Engineering Manager"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"amount": "$4,720.89", "city": " Bellmore", "department": "PM  ", "email": " OHAO@EXAMPLE.COM ", "country": "GS", "signup_date": "8/7/2023", "unit": "centimeters", "is_active": "FALSE", "notes2": ""}, {"amount": "483,28", "city": "N'Zeto", "department": "FIN", "email": "fmfdeo@example.com", "country": "south georgia and the south sandwich islands", "signup_date": "2023-08-27", "unit": "CENTIMETERS", "is_active": "T", "notes2": ""}, {"amount": "$1,151.49", "city": "   N'zeto", "department": "prod", "email": "   mhhcjn@corp.io", "country": "GS", "signup_date": "14 Feb 2023", "unit": "MI   ", "is_active": "true", "notes2": ""}, {"amount": "3.963,20", "city": "bellmore", "department": "PM  ", "email": "POMDG@TEST.ORG", "country": "South Georgia And The South Sanwdich Islands", "signup_date": "2023-12-01", "unit": "miles", "is_active": "F", "notes2": ""}, {"amount": "$3,222.37", "city": "N'zeto", "department": "finance", "email": "onhmb@mail.com", "country": "SGS", "signup_date": "4 Sep 2023", "unit": "miles   ", "is_active": "true", "notes2": ""}, {"amount": "$7,100.23", "city": "Danshui", "department": " pm ", "email": "aoebocl@example.com", "country": "TO", "signup_date": "22 Jan 2023", "unit": "mile", "is_active": "0", "notes2": ""}, {"amount": "7.490,18", "city": "Bellmoore", "department": "finance", "email": "mjdpdim@corp.io", "country": "PLW", "signup_date": "10/18/2023", "unit": "Cm", "is_active": "N", "notes2": ""}, {"amount": "4.306,41", "city": "  BELLMORE", "department": "   Accounting", "email": "NDLF@EXAMPLE.COM ", "country": "south georgia and the south sandwich islands", "signup_date": "2023-04-08", "unit": "milliliter  ", "is_active": "Y", "notes2": ""}, {"amount": "3.847,64", "city": "danshui", "department": "product", "email": "iggboje@test.org", "country": "SGS", "signup_date": "2023-04-20", "unit": "mL", "is_active": "1", "notes2": ""}, {"amount": "N/A", "city": "Danshui", "department": "Accounitng", "email": "djoddda@test.org", "country": "Palvu", "signup_date": "16 Jul 2023", "unit": " miles ", "is_active": "N", "notes2": ""}, {"amount": "N/A", "city": "Bellmore", "department": "Finance", "email": "ahap@test.org ", "country": "PW", "signup_date": "2023-03-09", "unit": "centimeters", "is_active": "No", "notes2": ""}, {"amount": "6.915,43", "city": "N'zjto", "department": "FINANCE", "email": "   fdapk@mail.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "11 Jun 2023", "unit": "mi", "is_active": "true", "notes2": ""}, {"amount": "6.850,81", "city": "N'ZETO", "department": "Accounting", "email": "lcefdc@test.org", "country": "fro", "signup_date": "10 May 2023", "unit": "ml", "is_active": "No", "notes2": ""}, {"amount": "716,47", "city": " N'zeto", "department": "accoulting", "email": " odfao@mail.com", "country": "Tonga", "signup_date": "6/28/2023", "unit": "centimeter", "is_active": "Y", "notes2": ""}, {"amount": "$3,186.37", "city": "bellmmore", "department": "product", "email": "MFNECPD@CORP.IO", "country": "Palau", "signup_date": "45053", "unit": "   mls", "is_active": "1", "notes2": ""}, {"amount": "$4,831.03", "city": "Nzeho", "department": "PROD", "email": "JKAIFN@EXAMPLE.COM", "country": "plw", "signup_date": "44945", "unit": "mls", "is_active": "Yes", "notes2": ""}, {"amount": "1.117,81", "city": "bellmore", "department": "fin  ", "email": "LDANC@TEST.ORG", "country": "fro", "signup_date": "45099", "unit": "mi", "is_active": "true", "notes2": ""}, {"amount": "8.359,29", "city": "danshui", "department": " fin ", "email": "   EIFB@MAIL.COM", "country": "Sgs", "signup_date": "2023-04-17", "unit": "ml   ", "is_active": "false", "notes2": ""}, {"amount": "--", "city": "  BELLMORE", "department": "PRODUT", "email": "KEJIGO@CORP.IO", "country": "REPUBLIC OF PALAU ", "signup_date": "27 Jan 2023", "unit": "   centimeters", "is_active": "N", "notes2": ""}, {"amount": "$1,533.08", "city": "  belmore", "department": "Product", "email": "  lfmfgji@corp.io", "country": "  Kingdo mof Tonga", "signup_date": "26 Aug 2023", "unit": "mi", "is_active": "N", "notes2": ""}, {"amount": "1.129,85", "city": "Bellmore", "department": " PM ", "email": " KKEMHBA@CORP.IO", "country": "South Georgia And The South Sandwich Islands", "signup_date": "2023-11-01", "unit": "centimeters", "is_active": "TRUE", "notes2": ""}, {"amount": "4.987,87", "city": "Bellxore", "department": "accounting", "email": " mihbj@example.com ", "country": "KINGDOM OF TONGA", "signup_date": "45245", "unit": "miles", "is_active": "true", "notes2": ""}, {"amount": "$4,469.51", "city": "Belmlore", "department": "PM", "email": "pkifaj@mail.com", "country": "  TO", "signup_date": "17 Nov 2023", "unit": "mile", "is_active": "N", "notes2": ""}, {"amount": "5.258,07", "city": "Nzeto", "department": "PM", "email": "ENEDDDH@CORP.IO ", "country": "Republic Of Palau", "signup_date": "21 Jan 2023", "unit": "ml ", "is_active": "N", "notes2": ""}, {"amount": "5.652,41", "city": "  Danshui", "department": "Prodduct", "email": "ELBFPJ@CORP.IO", "country": "   FRO", "signup_date": "26 Oct 2023", "unit": "ml", "is_active": "No", "notes2": ""}, {"amount": "1.117,81", "city": "bellmore", "department": "fin  ", "email": "LDANC@TEST.ORG", "country": "fro", "signup_date": "45099", "unit": "mi", "is_active": "true", "notes2": ""}, {"amount": "N/A", "city": "Danshui", "department": "Accounitng", "email": "djoddda@test.org", "country": "Palvu", "signup_date": "16 Jul 2023", "unit": " miles ", "is_active": "N", "notes2": ""}], "clean": [{"amount": 4720.89, "city": "Bellmore", "department": "Product", "email": "ohao@example.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-08-07", "unit": "cm", "is_active": false}, {"amount": 483.28, "city": "N'zeto", "department": "Finance", "email": "fmfdeo@example.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-08-27", "unit": "cm", "is_active": true}, {"amount": 1151.49, "city": "N'zeto", "department": "Product", "email": "mhhcjn@corp.io", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-02-14", "unit": "mi", "is_active": true}, {"amount": 3963.2, "city": "Bellmore", "department": "Product", "email": "pomdg@test.org", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-12-01", "unit": "mi", "is_active": false}, {"amount": 3222.37, "city": "N'zeto", "department": "Finance", "email": "onhmb@mail.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-09-04", "unit": "mi", "is_active": true}, {"amount": 7100.23, "city": "Danshui", "department": "Product", "email": "aoebocl@example.com", "country": "Tonga", "signup_date": "2023-01-22", "unit": "mi", "is_active": false}, {"amount": 7490.18, "city": "Bellmore", "department": "Finance", "email": "mjdpdim@corp.io", "country": "Palau", "signup_date": "2023-10-18", "unit": "cm", "is_active": false}, {"amount": 4306.41, "city": "Bellmore", "department": "Finance", "email": "ndlf@example.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-04-08", "unit": "mL", "is_active": true}, {"amount": 3847.64, "city": "Danshui", "department": "Product", "email": "iggboje@test.org", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-04-20", "unit": "mL", "is_active": true}, {"amount": NaN, "city": "Danshui", "department": "Finance", "email": "djoddda@test.org", "country": "Palau", "signup_date": "2023-07-16", "unit": "mi", "is_active": false}, {"amount": NaN, "city": "Bellmore", "department": "Finance", "email": "ahap@test.org", "country": "Palau", "signup_date": "2023-03-09", "unit": "cm", "is_active": false}, {"amount": 6915.43, "city": "N'zeto", "department": "Finance", "email": "fdapk@mail.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-06-11", "unit": "mi", "is_active": true}, {"amount": 6850.81, "city": "N'zeto", "department": "Finance", "email": "lcefdc@test.org", "country": "Faroe Islands", "signup_date": "2023-05-10", "unit": "mL", "is_active": false}, {"amount": 716.47, "city": "N'zeto", "department": "Finance", "email": "odfao@mail.com", "country": "Tonga", "signup_date": "2023-06-28", "unit": "cm", "is_active": true}, {"amount": 3186.37, "city": "Bellmore", "department": "Product", "email": "mfnecpd@corp.io", "country": "Palau", "signup_date": "2023-05-07", "unit": "mL", "is_active": true}, {"amount": 4831.03, "city": "N'zeto", "department": "Product", "email": "jkaifn@example.com", "country": "Palau", "signup_date": "2023-01-19", "unit": "mL", "is_active": true}, {"amount": 1117.81, "city": "Bellmore", "department": "Finance", "email": "ldanc@test.org", "country": "Faroe Islands", "signup_date": "2023-06-22", "unit": "mi", "is_active": true}, {"amount": 8359.29, "city": "Danshui", "department": "Finance", "email": "eifb@mail.com", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-04-17", "unit": "mL", "is_active": false}, {"amount": NaN, "city": "Bellmore", "department": "Product", "email": "kejigo@corp.io", "country": "Palau", "signup_date": "2023-01-27", "unit": "cm", "is_active": false}, {"amount": 1533.08, "city": "Bellmore", "department": "Product", "email": "lfmfgji@corp.io", "country": "Tonga", "signup_date": "2023-08-26", "unit": "mi", "is_active": false}, {"amount": 1129.85, "city": "Bellmore", "department": "Product", "email": "kkemhba@corp.io", "country": "South Georgia and the South Sandwich Islands", "signup_date": "2023-11-01", "unit": "cm", "is_active": true}, {"amount": 4987.87, "city": "Bellmore", "department": "Finance", "email": "mihbj@example.com", "country": "Tonga", "signup_date": "2023-11-15", "unit": "mi", "is_active": true}, {"amount": 4469.51, "city": "Bellmore", "department": "Product", "email": "pkifaj@mail.com", "country": "Tonga", "signup_date": "2023-11-17", "unit": "mi", "is_active": false}, {"amount": 5258.07, "city": "N'zeto", "department": "Product", "email": "enedddh@corp.io", "country": "Palau", "signup_date": "2023-01-21", "unit": "mL", "is_active": false}, {"amount": 5652.41, "city": "Danshui", "department": "Product", "email": "elbfpj@corp.io", "country": "Faroe Islands", "signup_date": "2023-10-26", "unit": "mL", "is_active": false}], "dirty_cols": ["amount", "city", "department", "email", "country", "signup_date", "unit", "is_active", "notes2"], "clean_cols": ["amount", "city", "department", "email", "country", "signup_date", "unit", "is_active"], "plan": {"dataset_summary": "27 rows × 9 columns. 8 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"N'Zeto": "N'zeto", "bellmore": "Bellmore", "Bellmoore": "Bellmore", "BELLMORE": "Bellmore", "danshui": "Danshui", "N'zjto": "N'zeto", "N'ZETO": "N'zeto", "bellmmore": "Bellmore", "Nzeho": "N'zeto", "belmore": "Bellmore", "Bellxore": "Bellmore", "Belmlore": "Bellmore", "Nzeto": "N'zeto"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"PM": "Product", "FIN": "Finance", "prod": "Product", "finance": "Finance", "pm": "Product", "Accounting": "Finance", "product": "Product", "Accounitng": "Finance", "FINANCE": "Finance", "accoulting": "Finance", "PROD": "Product", "fin": "Finance", "PRODUT": "Product", "accounting": "Finance", "Prodduct": "Product"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"GS": "South Georgia and the South Sandwich Islands", "south georgia and the south sandwich islands": "South Georgia and the South Sandwich Islands", "South Georgia And The South Sanwdich Islands": "South Georgia and the South Sandwich Islands", "SGS": "South Georgia and the South Sandwich Islands", "TO": "Tonga", "PLW": "Palau", "Palvu": "Palau", "PW": "Palau", "fro": "Faroe Islands", "plw": "Palau", "Sgs": "South Georgia and the South Sandwich Islands", "REPUBLIC OF PALAU": "Palau", "Kingdo mof Tonga": "Tonga", "South Georgia And The South Sandwich Islands": "South Georgia and the South Sandwich Islands", "KINGDOM OF TONGA": "Tonga", "Republic Of Palau": "Palau", "FRO": "Faroe Islands"}, "rationale": "Unified 17 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"centimeters": "cm", "CENTIMETERS": "cm", "MI": "mi", "miles": "mi", "mile": "mi", "Cm": "cm", "milliliter": "mL", "ml": "mL", "centimeter": "cm", "mls": "mL"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"company": "Soylent Corp", "amount": "2.582,72", "is_active": "T", "job_title": "c.e.o.", "state": "Ok", "city": "DUGDA", "col_x": ""}, {"company": "Wonka Co", "amount": "453,54", "is_active": "N", "job_title": "CEO", "state": " Virginia ", "city": "Yicheng", "col_x": ""}, {"company": "Wonka Co", "amount": "$2,234.33", "is_active": "true", "job_title": "prod mgr", "state": "Maryland", "city": "yicheng", "col_x": ""}, {"company": " Cyberdyne ", "amount": "1.849,14", "is_active": "FALSE", "job_title": "prod manager", "state": "   south dakota", "city": "   Yicheng", "col_x": ""}, {"company": "Acme Inc", "amount": "6.664,04", "is_active": "0", "job_title": "Vice Pres", "state": "Maryland", "city": "Dugda", "col_x": ""}, {"company": " Hooli ", "amount": "$7,424.73", "is_active": "true", "job_title": "C.E.O.", "state": "Oklahoma", "city": " yicheng ", "col_x": ""}, {"company": " Wonka Co", "amount": "$3,888.67", "is_active": "TRUE", "job_title": "Vice President", "state": "Oklahoma   ", "city": "yicheng", "col_x": ""}, {"company": "Stark Industries", "amount": "5.910,66", "is_active": "TRUE", "job_title": "Vice Pres", "state": "  South Dakota", "city": "tasbuget", "col_x": ""}, {"company": "Acme Inc", "amount": "2.134,02", "is_active": "No", "job_title": "Chief Executive Officer", "state": "VA", "city": "Ayamonte", "col_x": ""}, {"company": " Globex ", "amount": "2.185,38", "is_active": "T", "job_title": "Prod Mgr", "state": "South Dakota  ", "city": "DUGDA", "col_x": ""}, {"company": " Initech", "amount": "--", "is_active": "true", "job_title": "   Senior Engineer", "state": "   oklahoma", "city": "Yicheng", "col_x": ""}, {"company": "   Soylent Corp", "amount": "533,57", "is_active": "0", "job_title": "Chief Executive Officer", "state": "Oklahoma", "city": "Dugda", "col_x": ""}, {"company": "Initech ", "amount": "$8,356.63", "is_active": "No", "job_title": "vice  pres", "state": "oklahoma", "city": "Yicheng", "col_x": ""}, {"company": " Cyberdyne ", "amount": "353,50", "is_active": "TRUE", "job_title": "snr  ngineer", "state": "SUUTH DAKOTA", "city": "Yicheng", "col_x": ""}, {"company": "Hooli", "amount": "-100", "is_active": "false", "job_title": "Vice Pes", "state": "SD", "city": "  YICHENG", "col_x": ""}, {"company": "Soylent Corp  ", "amount": "$5,852.77", "is_active": "Y", "job_title": "Prod Manager", "state": " virginia ", "city": "Yicheng", "col_x": ""}, {"company": "Stark Industries", "amount": "1.349,69", "is_active": "Y", "job_title": "VP", "state": "South Dakota", "city": " Dugda ", "col_x": ""}, {"company": " Initech ", "amount": "6.815,70", "is_active": "false", "job_title": "Chief Executive Officer", "state": "md", "city": "Yickeng", "col_x": ""}, {"company": " Globex ", "amount": "2.185,38", "is_active": "T", "job_title": "Prod Mgr", "state": "South Dakota  ", "city": "DUGDA", "col_x": ""}, {"company": "Soylent Corp  ", "amount": "$5,852.77", "is_active": "Y", "job_title": "Prod Manager", "state": " virginia ", "city": "Yicheng", "col_x": ""}], "clean": [{"company": "Soylent Corp", "amount": 2582.72, "is_active": true, "job_title": "Chief Executive Officer", "state": "Oklahoma", "city": "Dugda"}, {"company": "Wonka Co", "amount": 453.54, "is_active": false, "job_title": "Chief Executive Officer", "state": "Virginia", "city": "Yicheng"}, {"company": "Wonka Co", "amount": 2234.33, "is_active": true, "job_title": "Product Manager", "state": "Maryland", "city": "Yicheng"}, {"company": "Cyberdyne", "amount": 1849.14, "is_active": false, "job_title": "Product Manager", "state": "South Dakota", "city": "Yicheng"}, {"company": "Acme Inc", "amount": 6664.04, "is_active": false, "job_title": "Vice President", "state": "Maryland", "city": "Dugda"}, {"company": "Hooli", "amount": 7424.73, "is_active": true, "job_title": "Chief Executive Officer", "state": "Oklahoma", "city": "Yicheng"}, {"company": "Wonka Co", "amount": 3888.67, "is_active": true, "job_title": "Vice President", "state": "Oklahoma", "city": "Yicheng"}, {"company": "Stark Industries", "amount": 5910.66, "is_active": true, "job_title": "Vice President", "state": "South Dakota", "city": "Tasbuget"}, {"company": "Acme Inc", "amount": 2134.02, "is_active": false, "job_title": "Chief Executive Officer", "state": "Virginia", "city": "Ayamonte"}, {"company": "Globex", "amount": 2185.38, "is_active": true, "job_title": "Product Manager", "state": "South Dakota", "city": "Dugda"}, {"company": "Initech", "amount": NaN, "is_active": true, "job_title": "Senior Engineer", "state": "Oklahoma", "city": "Yicheng"}, {"company": "Soylent Corp", "amount": 533.57, "is_active": false, "job_title": "Chief Executive Officer", "state": "Oklahoma", "city": "Dugda"}, {"company": "Initech", "amount": 8356.63, "is_active": false, "job_title": "Vice President", "state": "Oklahoma", "city": "Yicheng"}, {"company": "Cyberdyne", "amount": 353.5, "is_active": true, "job_title": "Senior Engineer", "state": "South Dakota", "city": "Yicheng"}, {"company": "Hooli", "amount": -100.0, "is_active": false, "job_title": "Vice President", "state": "South Dakota", "city": "Yicheng"}, {"company": "Soylent Corp", "amount": 5852.77, "is_active": true, "job_title": "Product Manager", "state": "Virginia", "city": "Yicheng"}, {"company": "Stark Industries", "amount": 1349.69, "is_active": true, "job_title": "Vice President", "state": "South Dakota", "city": "Dugda"}, {"company": "Initech", "amount": 6815.7, "is_active": false, "job_title": "Chief Executive Officer", "state": "Maryland", "city": "Yicheng"}], "dirty_cols": ["company", "amount", "is_active", "job_title", "state", "city", "col_x"], "clean_cols": ["company", "amount", "is_active", "job_title", "state", "city"], "plan": {"dataset_summary": "20 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"c.e.o.": "Chief Executive Officer", "CEO": "Chief Executive Officer", "prod mgr": "Product Manager", "prod manager": "Product Manager", "Vice Pres": "Vice President", "C.E.O.": "Chief Executive Officer", "Prod Mgr": "Product Manager", "vice pres": "Vice President", "snr ngineer": "Senior Engineer", "Vice Pes": "Vice President", "Prod Manager": "Product Manager", "VP": "Vice President"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Ok": "Oklahoma", "south dakota": "South Dakota", "VA": "Virginia", "oklahoma": "Oklahoma", "SUUTH DAKOTA": "South Dakota", "SD": "South Dakota", "virginia": "Virginia", "md": "Maryland"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"DUGDA": "Dugda", "yicheng": "Yicheng", "tasbuget": "Tasbuget", "YICHENG": "Yicheng", "Yickeng": "Yicheng"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"unit": "MLS", "currency": "BOLIVIANO", "amount": "$7,315.88", "department": "i.t.", "is_active": "F", "phone": "(514)900-1233", "extra": ""}, {"unit": "milliliter", "currency": "BOB   ", "amount": "2.062,89", "department": "I.T.", "is_active": "No", "phone": "(578)109-1545", "extra": ""}, {"unit": "mL", "currency": "UAE Dirham", "amount": "$3,646.34", "department": "information technology", "is_active": "Yes", "phone": "513-616-2079", "extra": ""}, {"unit": "   L", "currency": "AED", "amount": "$1,369.00", "department": " Operations ", "is_active": "No", "phone": "543.393.7639", "extra": ""}, {"unit": "   LB", "currency": "uae dirham", "amount": "$8,187.63", "department": " Support ", "is_active": "TRUE", "phone": "5409758094", "extra": ""}, {"unit": "POUND", "currency": "UAEDirham", "amount": "4.953,43", "department": "IT ", "is_active": "TRUE", "phone": "(580)157-2854", "extra": ""}, {"unit": "lbs", "currency": "BOB", "amount": "7.180,26", "department": "OPERATIONS", "is_active": "F", "phone": "506-946-1356", "extra": ""}, {"unit": "liters", "currency": "   AED", "amount": "8.933,70", "department": "ops", "is_active": "Yes", "phone": "5679531074", "extra": ""}, {"unit": "lb", "currency": "Uae Diham", "amount": "$1,690.44", "department": "customer support", "is_active": "T", "phone": "586-375-7822", "extra": ""}, {"unit": "lb", "currency": "AED", "amount": "6.370,72", "department": "OPS", "is_active": "Y", "phone": "(587)335-9000", "extra": ""}, {"unit": "L", "currency": "bob", "amount": "$6,310.97", "department": "Ops", "is_active": "Y", "phone": "(512)733-8814", "extra": ""}, {"unit": "Lb", "currency": "UAE Dirham", "amount": "7.573,32", "department": "IT", "is_active": "Y", "phone": "502.667.8824", "extra": ""}, {"unit": "liters", "currency": "BOB", "amount": "6.270,76", "department": "Operatiions", "is_active": "TRUE", "phone": "5185542394", "extra": ""}, {"unit": " MLS", "currency": "AED", "amount": "4.928,22", "department": "operations", "is_active": "Yes", "phone": "5113084396", "extra": ""}], "clean": [{"unit": "mL", "currency": "BOB", "amount": 7315.88, "department": "Information Technology", "is_active": false, "phone": "(514) 900-1233"}, {"unit": "mL", "currency": "BOB", "amount": 2062.89, "department": "Information Technology", "is_active": false, "phone": "(578) 109-1545"}, {"unit": "mL", "currency": "AED", "amount": 3646.34, "department": "Information Technology", "is_active": true, "phone": "(513) 616-2079"}, {"unit": "L", "currency": "AED", "amount": 1369.0, "department": "Operations", "is_active": false, "phone": "(543) 393-7639"}, {"unit": "lb", "currency": "AED", "amount": 8187.63, "department": "Customer Support", "is_active": true, "phone": "(540) 975-8094"}, {"unit": "lb", "currency": "AED", "amount": 4953.43, "department": "Information Technology", "is_active": true, "phone": "(580) 157-2854"}, {"unit": "lb", "currency": "BOB", "amount": 7180.26, "department": "Operations", "is_active": false, "phone": "(506) 946-1356"}, {"unit": "L", "currency": "AED", "amount": 8933.7, "department": "Operations", "is_active": true, "phone": "(567) 953-1074"}, {"unit": "lb", "currency": "AED", "amount": 1690.44, "department": "Customer Support", "is_active": true, "phone": "(586) 375-7822"}, {"unit": "lb", "currency": "AED", "amount": 6370.72, "department": "Operations", "is_active": true, "phone": "(587) 335-9000"}, {"unit": "L", "currency": "BOB", "amount": 6310.97, "department": "Operations", "is_active": true, "phone": "(512) 733-8814"}, {"unit": "lb", "currency": "AED", "amount": 7573.32, "department": "Information Technology", "is_active": true, "phone": "(502) 667-8824"}, {"unit": "L", "currency": "BOB", "amount": 6270.76, "department": "Operations", "is_active": true, "phone": "(518) 554-2394"}, {"unit": "mL", "currency": "AED", "amount": 4928.22, "department": "Operations", "is_active": true, "phone": "(511) 308-4396"}], "dirty_cols": ["unit", "currency", "amount", "department", "is_active", "phone", "extra"], "clean_cols": ["unit", "currency", "amount", "department", "is_active", "phone"], "plan": {"dataset_summary": "14 rows × 7 columns. 6 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MLS": "mL", "milliliter": "mL", "LB": "lb", "POUND": "lb", "lbs": "lb", "liters": "L", "Lb": "lb"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"BOLIVIANO": "BOB", "UAE Dirham": "AED", "uae dirham": "AED", "UAEDirham": "AED", "Uae Diham": "AED", "bob": "BOB"}, "rationale": "Unified 6 variant spelling(s) into canonical labels."}]}, {"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols"], "operations": [{"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"i.t.": "Information Technology", "I.T.": "Information Technology", "information technology": "Information Technology", "Support": "Customer Support", "IT": "Information Technology", "OPERATIONS": "Operations", "ops": "Operations", "customer support": "Customer Support", "OPS": "Operations", "Ops": "Operations", "Operatiions": "Operations", "operations": "Operations"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}], "flags": []}}
{"dirty": [{"status": "Closed Won", "email": "ijidao@corp.io", "is_active": "T", "country": "ua", "notes2": ""}, {"status": "IN PROGESS", "email": "miaeljo@example.com", "is_active": "Yes", "country": "MNG", "notes2": ""}, {"status": "Won", "email": " aobpl@test.org", "is_active": "1", "country": "mng", "notes2": ""}, {"status": "Won", "email": "CLOGGKF@MAIL.COM", "is_active": "1", "country": "ukr", "notes2": ""}, {"status": "won", "email": " mkecjom@example.com", "is_active": "Yes", "country": "MONACO", "notes2": ""}, {"status": "Open", "email": " DPJCLJ@MAIL.COM ", "is_active": "N", "country": "IQ", "notes2": ""}, {"status": "won", "email": "ekgijb@corp.io", "is_active": "T", "country": "iq   ", "notes2": ""}, {"status": "Won", "email": "MMPJ@EXAMPLE.COM  ", "is_active": "true", "country": "UKR", "notes2": ""}, {"status": "WON", "email": "LLOLEHP@EXAMPLE.COM", "is_active": "TRUE", "country": "Monaco ", "notes2": ""}, {"status": "in-progress", "email": "nnim@corp.io", "is_active": "T", "country": "   Ukraine", "notes2": ""}, {"status": "new", "email": "nalkd@test.org", "is_active": "N", "country": "CX", "notes2": ""}, {"status": "Open", "email": "LKAICDE@CORP.IO", "is_active": "FALSE", "country": "  UKR", "notes2": ""}, {"status": "closed-won", "email": "EGHN@TEST.ORG", "is_active": "N", "country": "Mongolia", "notes2": ""}], "clean": [{"status": "Won", "email": "ijidao@corp.io", "is_active": true, "country": "Ukraine"}, {"status": "In Progress", "email": "miaeljo@example.com", "is_active": true, "country": "Mongolia"}, {"status": "Won", "email": "aobpl@test.org", "is_active": true, "country": "Mongolia"}, {"status": "Won", "email": "cloggkf@mail.com", "is_active": true, "country": "Ukraine"}, {"status": "Won", "email": "mkecjom@example.com", "is_active": true, "country": "Monaco"}, {"status": "Open", "email": "dpjclj@mail.com", "is_active": false, "country": "Iraq"}, {"status": "Won", "email": "ekgijb@corp.io", "is_active": true, "country": "Iraq"}, {"status": "Won", "email": "mmpj@example.com", "is_active": true, "country": "Ukraine"}, {"status": "Won", "email": "llolehp@example.com", "is_active": true, "country": "Monaco"}, {"status": "In Progress", "email": "nnim@corp.io", "is_active": true, "country": "Ukraine"}, {"status": "Open", "email": "nalkd@test.org", "is_active": false, "country": "Christmas Island"}, {"status": "Open", "email": "lkaicde@corp.io", "is_active": false, "country": "Ukraine"}, {"status": "Won", "email": "eghn@test.org", "is_active": false, "country": "Mongolia"}], "dirty_cols": ["status", "email", "is_active", "country", "notes2"], "clean_cols": ["status", "email", "is_active", "country"], "plan": {"dataset_summary": "13 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"Closed Won": "Won", "IN PROGESS": "In Progress", "won": "Won", "WON": "Won", "in-progress": "In Progress", "new": "Open", "closed-won": "Won"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ua": "Ukraine", "MNG": "Mongolia", "mng": "Mongolia", "ukr": "Ukraine", "MONACO": "Monaco", "IQ": "Iraq", "iq": "Iraq", "UKR": "Ukraine", "CX": "Christmas Island"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "Y", "city": " chorley", "job_title": "Director", "signup_date": "44986", "unit": " ml ", "state": "KS", "unnamed": ""}, {"is_active": "Y", "city": " Madrid ", "job_title": "driector", "signup_date": "2023-09-16", "unit": " kilometers ", "state": " vi ", "unnamed": ""}, {"is_active": "false", "city": "Slidell", "job_title": "C.T.O.", "signup_date": "24 Apr 2023", "unit": "mls", "state": " vi", "unnamed": ""}, {"is_active": "false", "city": "nanyamba", "job_title": " Dir. ", "signup_date": "45093", "unit": "kilometer", "state": "VIRGIN ISLANDS, U.S.", "unnamed": ""}, {"is_active": "false", "city": "Madrid", "job_title": "Director   ", "signup_date": "2023-09-20", "unit": "ML  ", "state": "KANSAS  ", "unnamed": ""}, {"is_active": "Y", "city": "nnyamba", "job_title": "DIREECTOR", "signup_date": "2023-09-27", "unit": "kilometers", "state": "VI", "unnamed": ""}, {"is_active": "Y", "city": "chorley", "job_title": "CTO", "signup_date": "11 Dec 2023", "unit": "KILOMETER", "state": "VI", "unnamed": ""}, {"is_active": "TRUE", "city": "Nanyamba", "job_title": "director", "signup_date": "9/7/2023", "unit": "kilometer", "state": "ks", "unnamed": ""}, {"is_active": "TRUE", "city": "Nanyamba", "job_title": "  cto", "signup_date": "8 May 2023", "unit": "kilometers", "state": "Virgin Islands, U.S.", "unnamed": ""}, {"is_active": "FALSE", "city": "Nanyama   ", "job_title": "Dir.", "signup_date": "2023-10-12", "unit": "mL", "state": "Kansas", "unnamed": ""}, {"is_active": "false", "city": "NANYAMBA", "job_title": "  Dir.", "signup_date": "1 Feb 2023", "unit": "KILOMETERS ", "state": "Virgin Islands US", "unnamed": ""}, {"is_active": "T", "city": " Chorley", "job_title": "Dir.", "signup_date": "11 Dec 2023", "unit": " kilometers ", "state": "VI", "unnamed": ""}, {"is_active": "Y", "city": "madrid", "job_title": "CHIEF TECHNOLOGY OFFICER", "signup_date": "2023-02-25", "unit": " kilometer ", "state": "vi", "unnamed": ""}, {"is_active": "N", "city": " madrid ", "job_title": "c.t.o.", "signup_date": "45236", "unit": "KILOMETES", "state": "Kanssas", "unnamed": ""}, {"is_active": "Y", "city": "MADRID", "job_title": "Dir", "signup_date": "9/12/2023", "unit": " mL ", "state": " vi ", "unnamed": ""}, {"is_active": "N", "city": "chorley ", "job_title": "director", "signup_date": "12 Oct 2023", "unit": "KILOMETER", "state": "kansas", "unnamed": ""}, {"is_active": "T", "city": "Nanyamqa", "job_title": "dir", "signup_date": "45088", "unit": "mls", "state": "Kansas", "unnamed": ""}, {"is_active": "true", "city": "Slidel", "job_title": "   Director", "signup_date": "28 Nov 2023", "unit": "Milliliter", "state": "VI   ", "unnamed": ""}, {"is_active": "Y", "city": "Nanyamba ", "job_title": " chief technology officer", "signup_date": "2023-04-08", "unit": "kilometer", "state": "KS  ", "unnamed": ""}, {"is_active": "true", "city": "Chorley", "job_title": "c.t.o.   ", "signup_date": "45165", "unit": "km", "state": "KS", "unnamed": ""}, {"is_active": "true", "city": "  Chroley", "job_title": "CHIIEF TECHNOLOGY OFFICER", "signup_date": "5/28/2023", "unit": "KM", "state": "  VI", "unnamed": ""}, {"is_active": "Yes", "city": "CHHORLEY", "job_title": "chief technology  officer", "signup_date": "2023-08-23", "unit": "kilometer", "state": "Kansas", "unnamed": ""}, {"is_active": "No", "city": " NANYAMBA", "job_title": "DIR", "signup_date": "23 Aug 2023", "unit": "milliliter", "state": "VI", "unnamed": ""}, {"is_active": "0", "city": "   Chorley", "job_title": "Dir.", "signup_date": "2023-10-07", "unit": "KM", "state": "VI", "unnamed": ""}, {"is_active": "true", "city": " Slidell ", "job_title": " CTO ", "signup_date": "45124", "unit": "km", "state": " Kanas ", "unnamed": ""}, {"is_active": "F", "city": "Nanyamba", "job_title": "Dir", "signup_date": "4/19/2023", "unit": "km", "state": "ks", "unnamed": ""}, {"is_active": "false", "city": "Madrid", "job_title": "Dir", "signup_date": "45073", "unit": "KILOMETERS", "state": " ks ", "unnamed": ""}, {"is_active": "Y", "city": "Chorley", "job_title": "Dir.", "signup_date": "26 Dec 2023", "unit": "mL", "state": "KS", "unnamed": ""}, {"is_active": "F", "city": "Slidell", "job_title": "cto", "signup_date": "24 Jan 2023", "unit": " kilometers ", "state": "Kansas", "unnamed": ""}, {"is_active": "false", "city": "   madrid", "job_title": "dir", "signup_date": "13 Nov 2023", "unit": "KM", "state": "KS", "unnamed": ""}, {"is_active": "false", "city": "madrid ", "job_title": "  di.r", "signup_date": "5/4/2023", "unit": "km", "state": " kansas ", "unnamed": ""}, {"is_active": "1", "city": "Slidell", "job_title": "Dir", "signup_date": "2/16/2023", "unit": "ml", "state": "kansas", "unnamed": ""}, {"is_active": "No", "city": "Nanyamba", "job_title": "cto", "signup_date": "2023-03-02", "unit": "kilometers", "state": " ks", "unnamed": ""}, {"is_active": "T", "city": "   Nanyamba", "job_title": "dir.", "signup_date": "45254", "unit": "  milliliter", "state": "KANSAS", "unnamed": ""}, {"is_active": "F", "city": "   mdrid", "job_title": "Diir.", "signup_date": "2023-10-16", "unit": "MLS", "state": "Kansas ", "unnamed": ""}, {"is_active": "No", "city": "Slidell", "job_title": "Dir", "signup_date": "28 Aug 2023", "unit": "Kilometers", "state": "vi", "unnamed": ""}, {"is_active": "Y", "city": " chorley ", "job_title": "Dir  ", "signup_date": "28 Dec 2023", "unit": "km", "state": "Ks", "unnamed": ""}, {"is_active": "TRUE", "city": "NANYAMBA", "job_title": "Dir.", "signup_date": "18 Aug 2023", "unit": "km", "state": "  ks", "unnamed": ""}, {"is_active": "Y", "city": "Slidell", "job_title": "cto", "signup_date": "4/24/2023", "unit": "MLS", "state": "Vi", "unnamed": ""}, {"is_active": "1", "city": "Madrid", "job_title": "Dir", "signup_date": "45174", "unit": " KILOMETER ", "state": "Kansas", "unnamed": ""}, {"is_active": "Y", "city": "Chorley", "job_title": "C.T.O.", "signup_date": "14 Apr 2023", "unit": "ml", "state": "virgin islands, u.s.", "unnamed": ""}, {"is_active": "T", "city": "nanyamba", "job_title": "dir", "signup_date": "45206", "unit": "ml", "state": "VirginyIslands, U.S.", "unnamed": ""}, {"is_active": "F", "city": "  Nanyamba", "job_title": " CTO ", "signup_date": "12/16/2023", "unit": "milliliter", "state": "vi", "unnamed": ""}, {"is_active": "F", "city": "slidell", "job_title": "Dir.   ", "signup_date": "9/2/2023", "unit": "kilometer", "state": "VI", "unnamed": ""}, {"is_active": "F", "city": "Maddrid", "job_title": " CTO ", "signup_date": "45074", "unit": "Ml", "state": "VI", "unnamed": ""}, {"is_active": "false", "city": "CHORLEY", "job_title": "  Chief Technology Officer", "signup_date": "2023-01-01", "unit": "ml   ", "state": "vi ", "unnamed": ""}, {"is_active": "N", "city": "Slidell", "job_title": "Direcctor ", "signup_date": "8/4/2023", "unit": " kilometers ", "state": "KS  ", "unnamed": ""}, {"is_active": "TRUE", "city": "Nanyamba", "job_title": "cto", "signup_date": "2023-09-10", "unit": "ML", "state": "ks", "unnamed": ""}, {"is_active": "FALSE", "city": "Chorley", "job_title": "cto", "signup_date": "3/28/2023", "unit": "  kilometer", "state": "KS", "unnamed": ""}, {"is_active": "No", "city": "Slidell", "job_title": "C.T.O. ", "signup_date": "2023-01-13", "unit": "milliliter   ", "state": "  Virgin Islands, U.S.", "unnamed": ""}, {"is_active": "FALSE", "city": "Madrid", "job_title": "Chief Technology Officer", "signup_date": "9/2/2023", "unit": "kiloometer", "state": "Virgin Isllands, U.S.", "unnamed": ""}, {"is_active": "1", "city": "  Nanyamba", "job_title": "CTO   ", "signup_date": "45105", "unit": "Kilometers", "state": "virgin islands, u.s.", "unnamed": ""}, {"is_active": "true", "city": "chorley", "job_title": "diretor", "signup_date": "2023-06-27", "unit": "milliliter", "state": "VI", "unnamed": ""}, {"is_active": "TRUE", "city": "nanyamba", "job_title": "chief technology officer  ", "signup_date": "2023-08-02", "unit": "mls", "state": "VI", "unnamed": ""}, {"is_active": "1", "city": "Madrid", "job_title": "Dir", "signup_date": "45174", "unit": " KILOMETER ", "state": "Kansas", "unnamed": ""}, {"is_active": "TRUE", "city": "Nanyamba", "job_title": "  cto", "signup_date": "8 May 2023", "unit": "kilometers", "state": "Virgin Islands, U.S.", "unnamed": ""}, {"is_active": "", "city": "", "job_title": "", "signup_date": "", "unit": "", "state": "", "unnamed": ""}], "clean": [{"is_active": true, "city": "Chorley", "job_title": "Director", "signup_date": "2023-03-01", "unit": "mL", "state": "Kansas"}, {"is_active": true, "city": "Madrid", "job_title": "Director", "signup_date": "2023-09-16", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Slidell", "job_title": "Chief Technology Officer", "signup_date": "2023-04-24", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-06-16", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Madrid", "job_title": "Director", "signup_date": "2023-09-20", "unit": "mL", "state": "Kansas"}, {"is_active": true, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-09-27", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-12-11", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-09-07", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-05-08", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-10-12", "unit": "mL", "state": "Kansas"}, {"is_active": false, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-02-01", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Chorley", "job_title": "Director", "signup_date": "2023-12-11", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Madrid", "job_title": "Chief Technology Officer", "signup_date": "2023-02-25", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Madrid", "job_title": "Chief Technology Officer", "signup_date": "2023-11-06", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Madrid", "job_title": "Director", "signup_date": "2023-09-12", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Chorley", "job_title": "Director", "signup_date": "2023-10-12", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-06-11", "unit": "mL", "state": "Kansas"}, {"is_active": true, "city": "Slidell", "job_title": "Director", "signup_date": "2023-11-28", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-04-08", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-08-27", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-05-28", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-08-23", "unit": "km", "state": "Kansas"}, {"is_active": false, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-08-23", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Chorley", "job_title": "Director", "signup_date": "2023-10-07", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Slidell", "job_title": "Chief Technology Officer", "signup_date": "2023-07-17", "unit": "km", "state": "Kansas"}, {"is_active": false, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-04-19", "unit": "km", "state": "Kansas"}, {"is_active": false, "city": "Madrid", "job_title": "Director", "signup_date": "2023-05-27", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Chorley", "job_title": "Director", "signup_date": "2023-12-26", "unit": "mL", "state": "Kansas"}, {"is_active": false, "city": "Slidell", "job_title": "Chief Technology Officer", "signup_date": "2023-01-24", "unit": "km", "state": "Kansas"}, {"is_active": false, "city": "Madrid", "job_title": "Director", "signup_date": "2023-11-13", "unit": "km", "state": "Kansas"}, {"is_active": false, "city": "Madrid", "job_title": "Director", "signup_date": "2023-05-04", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Slidell", "job_title": "Director", "signup_date": "2023-02-16", "unit": "mL", "state": "Kansas"}, {"is_active": false, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-03-02", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-11-24", "unit": "mL", "state": "Kansas"}, {"is_active": false, "city": "Madrid", "job_title": "Director", "signup_date": "2023-10-16", "unit": "mL", "state": "Kansas"}, {"is_active": false, "city": "Slidell", "job_title": "Director", "signup_date": "2023-08-28", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Chorley", "job_title": "Director", "signup_date": "2023-12-28", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-08-18", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Slidell", "job_title": "Chief Technology Officer", "signup_date": "2023-04-24", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Madrid", "job_title": "Director", "signup_date": "2023-09-05", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-04-14", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Nanyamba", "job_title": "Director", "signup_date": "2023-10-07", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-12-16", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Slidell", "job_title": "Director", "signup_date": "2023-09-02", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Madrid", "job_title": "Chief Technology Officer", "signup_date": "2023-05-28", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-01-01", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Slidell", "job_title": "Director", "signup_date": "2023-08-04", "unit": "km", "state": "Kansas"}, {"is_active": true, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-09-10", "unit": "mL", "state": "Kansas"}, {"is_active": false, "city": "Chorley", "job_title": "Chief Technology Officer", "signup_date": "2023-03-28", "unit": "km", "state": "Kansas"}, {"is_active": false, "city": "Slidell", "job_title": "Chief Technology Officer", "signup_date": "2023-01-13", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": false, "city": "Madrid", "job_title": "Chief Technology Officer", "signup_date": "2023-09-02", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-06-28", "unit": "km", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Chorley", "job_title": "Director", "signup_date": "2023-06-27", "unit": "mL", "state": "Virgin Islands, U.S."}, {"is_active": true, "city": "Nanyamba", "job_title": "Chief Technology Officer", "signup_date": "2023-08-02", "unit": "mL", "state": "Virgin Islands, U.S."}], "dirty_cols": ["is_active", "city", "job_title", "signup_date", "unit", "state", "unnamed"], "clean_cols": ["is_active", "city", "job_title", "signup_date", "unit", "state"], "plan": {"dataset_summary": "57 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"chorley": "Chorley", "nanyamba": "Nanyamba", "nnyamba": "Nanyamba", "Nanyama": "Nanyamba", "NANYAMBA": "Nanyamba", "madrid": "Madrid", "MADRID": "Madrid", "Nanyamqa": "Nanyamba", "Slidel": "Slidell", "Chroley": "Chorley", "CHHORLEY": "Chorley", "mdrid": "Madrid", "slidell": "Slidell", "Maddrid": "Madrid", "CHORLEY": "Chorley"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"driector": "Director", "C.T.O.": "Chief Technology Officer", "Dir.": "Director", "DIREECTOR": "Director", "CTO": "Chief Technology Officer", "director": "Director", "cto": "Chief Technology Officer", "CHIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "c.t.o.": "Chief Technology Officer", "Dir": "Director", "dir": "Director", "chief technology officer": "Chief Technology Officer", "CHIIEF TECHNOLOGY OFFICER": "Chief Technology Officer", "DIR": "Director", "di.r": "Director", "dir.": "Director", "Diir.": "Director", "Direcctor": "Director", "diretor": "Director"}, "rationale": "Unified 19 variant spelling(s) into canonical labels."}]}, {"name": "signup_date", "detected_semantic_type": "date", "issues": ["mixed_date_formats"], "operations": [{"op": "parse_date", "rationale": "Unified mixed date formats to ISO YYYY-MM-DD."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ml": "mL", "kilometers": "km", "mls": "mL", "kilometer": "km", "ML": "mL", "KILOMETER": "km", "KILOMETERS": "km", "KILOMETES": "km", "Milliliter": "mL", "KM": "km", "milliliter": "mL", "MLS": "mL", "Kilometers": "km", "Ml": "mL", "kiloometer": "km"}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}, {"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KS": "Kansas", "vi": "Virgin Islands, U.S.", "VIRGIN ISLANDS, U.S.": "Virgin Islands, U.S.", "KANSAS": "Kansas", "VI": "Virgin Islands, U.S.", "ks": "Kansas", "Virgin Islands US": "Virgin Islands, U.S.", "Kanssas": "Kansas", "kansas": "Kansas", "Kanas": "Kansas", "Ks": "Kansas", "Vi": "Virgin Islands, U.S.", "virgin islands, u.s.": "Virgin Islands, U.S.", "VirginyIslands, U.S.": "Virgin Islands, U.S.", "Virgin Isllands, U.S.": "Virgin Islands, U.S."}, "rationale": "Unified 15 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "$748.20", "is_active": "T", "currency": "all", "country": "  QAT", "unit": "kilograms", "col_x": ""}, {"amount": "$7,041.49", "is_active": "F", "currency": "ALL", "country": "sint maarten (dtch part)", "unit": "L", "col_x": ""}, {"amount": "5.168,78", "is_active": "FALSE", "currency": " Taka", "country": "SXM", "unit": "  KG", "col_x": ""}, {"amount": "92,42", "is_active": "Y", "currency": "Lek", "country": "SJ", "unit": "kilometers", "col_x": ""}, {"amount": "2.843,25", "is_active": "false", "currency": "Taaka", "country": "QA", "unit": "   kilogram", "col_x": ""}, {"amount": "8.033,67", "is_active": "false", "currency": "Lek", "country": "QA", "unit": "kilometer", "col_x": ""}, {"amount": "8.270,16", "is_active": "true", "currency": " ALL", "country": "   SJM", "unit": "L", "col_x": ""}, {"amount": "63,79", "is_active": "Yes", "currency": "  bdt", "country": "SINT MAARTEN (DUTCH PART)", "unit": " L", "col_x": ""}, {"amount": "$3,091.75", "is_active": "true", "currency": "Lek", "country": "QAT", "unit": " liter ", "col_x": ""}, {"amount": "5.113,10", "is_active": "1", "currency": "Cuban Peso", "country": "SX", "unit": "kg", "col_x": ""}, {"amount": "3.129,52", "is_active": "No", "currency": "Lek", "country": "sjm", "unit": "kilometeer", "col_x": ""}, {"amount": "$4,544.72", "is_active": "0", "currency": "taka", "country": "QAT", "unit": "liters", "col_x": ""}, {"amount": "1.318,06", "is_active": "F", "currency": "Lek", "country": "Svalbard and Jan Mayen", "unit": "liter", "col_x": ""}, {"amount": "$8,382.51", "is_active": "T", "currency": "Bdt", "country": "qat", "unit": "km", "col_x": ""}, {"amount": "8.166,69", "is_active": "Yes", "currency": "taka", "country": " SX ", "unit": "l", "col_x": ""}, {"amount": "-", "is_active": "false", "currency": "all", "country": "qaear", "unit": "l", "col_x": ""}, {"amount": "6.810,69", "is_active": "No", "currency": "Tika", "country": "SJM", "unit": "kilometer", "col_x": ""}, {"amount": "$2,735.61", "is_active": "false", "currency": "TAKA", "country": "Qatar", "unit": "LITER", "col_x": ""}, {"amount": "5.422,86", "is_active": "1", "currency": " cuban peso", "country": "Svalbard and Jan Mayen", "unit": "   LITRE", "col_x": ""}, {"amount": "3.129,52", "is_active": "No", "currency": "Lek", "country": "sjm", "unit": "kilometeer", "col_x": ""}], "clean": [{"amount": 748.2, "is_active": true, "currency": "ALL", "country": "Qatar", "unit": "kg"}, {"amount": 7041.49, "is_active": false, "currency": "ALL", "country": "Sint Maarten (Dutch part)", "unit": "L"}, {"amount": 5168.78, "is_active": false, "currency": "BDT", "country": "Sint Maarten (Dutch part)", "unit": "kg"}, {"amount": 92.42, "is_active": true, "currency": "ALL", "country": "Svalbard and Jan Mayen", "unit": "km"}, {"amount": 2843.25, "is_active": false, "currency": "BDT", "country": "Qatar", "unit": "kg"}, {"amount": 8033.67, "is_active": false, "currency": "ALL", "country": "Qatar", "unit": "km"}, {"amount": 8270.16, "is_active": true, "currency": "ALL", "country": "Svalbard and Jan Mayen", "unit": "L"}, {"amount": 63.79, "is_active": true, "currency": "BDT", "country": "Sint Maarten (Dutch part)", "unit": "L"}, {"amount": 3091.75, "is_active": true, "currency": "ALL", "country": "Qatar", "unit": "L"}, {"amount": 5113.1, "is_active": true, "currency": "CUP", "country": "Sint Maarten (Dutch part)", "unit": "kg"}, {"amount": 3129.52, "is_active": false, "currency": "ALL", "country": "Svalbard and Jan Mayen", "unit": "km"}, {"amount": 4544.72, "is_active": false, "currency": "BDT", "country": "Qatar", "unit": "L"}, {"amount": 1318.06, "is_active": false, "currency": "ALL", "country": "Svalbard and Jan Mayen", "unit": "L"}, {"amount": 8382.51, "is_active": true, "currency": "BDT", "country": "Qatar", "unit": "km"}, {"amount": 8166.69, "is_active": true, "currency": "BDT", "country": "Sint Maarten (Dutch part)", "unit": "L"}, {"amount": NaN, "is_active": false, "currency": "ALL", "country": "Qatar", "unit": "L"}, {"amount": 6810.69, "is_active": false, "currency": "BDT", "country": "Svalbard and Jan Mayen", "unit": "km"}, {"amount": 2735.61, "is_active": false, "currency": "BDT", "country": "Qatar", "unit": "L"}, {"amount": 5422.86, "is_active": true, "currency": "CUP", "country": "Svalbard and Jan Mayen", "unit": "L"}], "dirty_cols": ["amount", "is_active", "currency", "country", "unit", "col_x"], "clean_cols": ["amount", "is_active", "currency", "country", "unit"], "plan": {"dataset_summary": "20 rows × 6 columns. 5 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["col_x"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"all": "ALL", "Taka": "BDT", "Lek": "ALL", "Taaka": "BDT", "bdt": "BDT", "Cuban Peso": "CUP", "taka": "BDT", "Bdt": "BDT", "Tika": "BDT", "TAKA": "BDT", "cuban peso": "CUP"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"QAT": "Qatar", "sint maarten (dtch part)": "Sint Maarten (Dutch part)", "SXM": "Sint Maarten (Dutch part)", "SJ": "Svalbard and Jan Mayen", "QA": "Qatar", "SJM": "Svalbard and Jan Mayen", "SINT MAARTEN (DUTCH PART)": "Sint Maarten (Dutch part)", "SX": "Sint Maarten (Dutch part)", "sjm": "Svalbard and Jan Mayen", "qat": "Qatar", "qaear": "Qatar"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kilograms": "kg", "KG": "kg", "kilometers": "km", "kilogram": "kg", "kilometer": "km", "liter": "L", "kilometeer": "km", "liters": "L", "l": "L", "LITER": "L", "LITRE": "L"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"unit": "KG", "name": "Grace  Khan", "is_active": "1", "currency": " eruo ", "extra": ""}, {"unit": "kg", "name": "Alice Park  ", "is_active": "F", "currency": "FKP", "extra": ""}, {"unit": "KG", "name": "Carol  Novak", "is_active": "T", "currency": "Falkland Islands Pound", "extra": ""}, {"unit": "Kilogram", "name": "Grace  Park", "is_active": "N", "currency": "EUR", "extra": ""}, {"unit": "kg", "name": "  Eve Moore", "is_active": "TRUE", "currency": "euro", "extra": ""}, {"unit": "Lbs", "name": "Karl  Cruz", "is_active": "N", "currency": "Euro", "extra": ""}, {"unit": "lb", "name": "Karl Brandt", "is_active": "true", "currency": "euro", "extra": ""}, {"unit": " kg ", "name": "Eve Fischer", "is_active": "Yes", "currency": "falkland islands pound", "extra": ""}, {"unit": "Lb", "name": "Grace  Reyes", "is_active": "FALSE", "currency": "€", "extra": ""}, {"unit": "lbs  ", "name": "Alice  Moore", "is_active": "N", "currency": "Fkp ", "extra": ""}, {"unit": "Lb", "name": "Grace Johnson", "is_active": "Yes", "currency": "FKP", "extra": ""}], "clean": [{"unit": "kg", "name": "Grace Khan", "is_active": true, "currency": "EUR"}, {"unit": "kg", "name": "Alice Park", "is_active": false, "currency": "FKP"}, {"unit": "kg", "name": "Carol Novak", "is_active": true, "currency": "FKP"}, {"unit": "kg", "name": "Grace Park", "is_active": false, "currency": "EUR"}, {"unit": "kg", "name": "Eve Moore", "is_active": true, "currency": "EUR"}, {"unit": "lb", "name": "Karl Cruz", "is_active": false, "currency": "EUR"}, {"unit": "lb", "name": "Karl Brandt", "is_active": true, "currency": "EUR"}, {"unit": "kg", "name": "Eve Fischer", "is_active": true, "currency": "FKP"}, {"unit": "lb", "name": "Grace Reyes", "is_active": false, "currency": "EUR"}, {"unit": "lb", "name": "Alice Moore", "is_active": false, "currency": "FKP"}, {"unit": "lb", "name": "Grace Johnson", "is_active": true, "currency": "FKP"}], "dirty_cols": ["unit", "name", "is_active", "currency", "extra"], "clean_cols": ["unit", "name", "is_active", "currency"], "plan": {"dataset_summary": "11 rows × 5 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["extra"], "rationale": "Dropped column(s) with no data."}], "columns": [{"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"KG": "kg", "Kilogram": "kg", "Lbs": "lb", "Lb": "lb", "lbs": "lb"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"eruo": "EUR", "Falkland Islands Pound": "FKP", "euro": "EUR", "Euro": "EUR", "falkland islands pound": "FKP", "€": "EUR", "Fkp": "FKP"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"job_title": "dir", "city": "shreveport", "rate": "32.5%"}, {"job_title": "Director", "city": "   maitland", "rate": "88.1%"}, {"job_title": "Dir.", "city": "Shreveport", "rate": "52.8%"}, {"job_title": "ENGINEERING MGR", "city": "Maitland", "rate": "2.6%"}, {"job_title": "Director", "city": " SHREVEPORT ", "rate": "99.5%"}, {"job_title": "dir", "city": "pekin", "rate": "62.3%"}, {"job_title": "ENG MANAGER", "city": "Shreveport", "rate": "56.1%"}, {"job_title": "  Engineering Mgr", "city": " Pekin ", "rate": "0.5%"}, {"job_title": " dir ", "city": "maitland  ", "rate": "68.5%"}, {"job_title": "Dir", "city": "Shrevpeort", "rate": "20.6%"}, {"job_title": "Dir", "city": "Pekin", "rate": "54.5%"}, {"job_title": "Dir.", "city": "shreveport", "rate": "5.8%"}, {"job_title": "Eng Mgr", "city": "Shreveport", "rate": "88.8%"}, {"job_title": "", "city": "", "rate": ""}], "clean": [{"job_title": "Director", "city": "Shreveport", "rate": 0.325}, {"job_title": "Director", "city": "Maitland", "rate": 0.8809999999999999}, {"job_title": "Director", "city": "Shreveport", "rate": 0.528}, {"job_title": "Engineering Manager", "city": "Maitland", "rate": 0.026000000000000002}, {"job_title": "Director", "city": "Shreveport", "rate": 0.995}, {"job_title": "Director", "city": "Pekin", "rate": 0.623}, {"job_title": "Engineering Manager", "city": "Shreveport", "rate": 0.561}, {"job_title": "Engineering Manager", "city": "Pekin", "rate": 0.005}, {"job_title": "Director", "city": "Maitland", "rate": 0.685}, {"job_title": "Director", "city": "Shreveport", "rate": 0.20600000000000002}, {"job_title": "Director", "city": "Pekin", "rate": 0.545}, {"job_title": "Director", "city": "Shreveport", "rate": 0.057999999999999996}, {"job_title": "Engineering Manager", "city": "Shreveport", "rate": 0.888}], "dirty_cols": ["job_title", "city", "rate"], "clean_cols": ["job_title", "city", "rate"], "plan": {"dataset_summary": "14 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"dir": "Director", "Dir.": "Director", "ENGINEERING MGR": "Engineering Manager", "ENG MANAGER": "Engineering Manager", "Engineering Mgr": "Engineering Manager", "Dir": "Director", "Eng Mgr": "Engineering Manager"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"shreveport": "Shreveport", "maitland": "Maitland", "SHREVEPORT": "Shreveport", "pekin": "Pekin", "Shrevpeort": "Shreveport"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"company": "Initech", "email": "BNDL@TEST.ORG", "rate": "10.1%", "status": "in-progress", "job_title": "ceo"}, {"company": "  Umbrella", "email": " MGCIDH@CORP.IO ", "rate": "34.8%", "status": "lost", "job_title": "director"}, {"company": "Globex", "email": "GFNJJB@CORP.IO", "rate": "11.4%", "status": "Clsed Lost", "job_title": "   director"}, {"company": "Soylent Corp", "email": " BGKIAPP@TEST.ORG ", "rate": "66.2%", "status": "Closde Lost", "job_title": "C.T.O."}, {"company": "Soylent Corp", "email": "fgaieh@mail.com", "rate": "78.4%", "status": "Lst", "job_title": "CEO"}, {"company": "Globex", "email": "nefl@corp.io", "rate": "81.0%", "status": "ClosedzWon", "job_title": "CTO"}, {"company": "Hooli", "email": " emoffeg@mail.com ", "rate": "75.6%", "status": "Open", "job_title": "CTO"}, {"company": "Globex", "email": "neamg@mail.com", "rate": "0.1%", "status": "closed  lost", "job_title": "C.EgO."}, {"company": "Stark Industries", "email": "bpdgi@corp.io", "rate": "28.2%", "status": "wip", "job_title": "Ceo"}, {"company": "Vehement", "email": "ONJJAH@TEST.ORG ", "rate": "17.5%", "status": "Lost", "job_title": "   CTO"}, {"company": "Cyberdyne", "email": "pblhmc@corp.io   ", "rate": "18.1%", "status": "won", "job_title": "C.E.O. "}, {"company": "Globex", "email": "neamg@mail.com", "rate": "0.1%", "status": "closed  lost", "job_title": "C.EgO."}, {"company": "Cyberdyne", "email": "pblhmc@corp.io   ", "rate": "18.1%", "status": "won", "job_title": "C.E.O. "}], "clean": [{"company": "Initech", "email": "bndl@test.org", "rate": 0.10099999999999999, "status": "In Progress", "job_title": "Chief Executive Officer"}, {"company": "Umbrella", "email": "mgcidh@corp.io", "rate": 0.348, "status": "Lost", "job_title": "Director"}, {"company": "Globex", "email": "gfnjjb@corp.io", "rate": 0.114, "status": "Lost", "job_title": "Director"}, {"company": "Soylent Corp", "email": "bgkiapp@test.org", "rate": 0.662, "status": "Lost", "job_title": "Chief Technology Officer"}, {"company": "Soylent Corp", "email": "fgaieh@mail.com", "rate": 0.784, "status": "Lost", "job_title": "Chief Executive Officer"}, {"company": "Globex", "email": "nefl@corp.io", "rate": 0.81, "status": "Won", "job_title": "Chief Technology Officer"}, {"company": "Hooli", "email": "emoffeg@mail.com", "rate": 0.7559999999999999, "status": "Open", "job_title": "Chief Technology Officer"}, {"company": "Globex", "email": "neamg@mail.com", "rate": 0.001, "status": "Lost", "job_title": "Chief Executive Officer"}, {"company": "Stark Industries", "email": "bpdgi@corp.io", "rate": 0.282, "status": "In Progress", "job_title": "Chief Executive Officer"}, {"company": "Vehement", "email": "onjjah@test.org", "rate": 0.175, "status": "Lost", "job_title": "Chief Technology Officer"}, {"company": "Cyberdyne", "email": "pblhmc@corp.io", "rate": 0.18100000000000002, "status": "Won", "job_title": "Chief Executive Officer"}], "dirty_cols": ["company", "email", "rate", "status", "job_title"], "clean_cols": ["company", "email", "rate", "status", "job_title"], "plan": {"dataset_summary": "13 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"in-progress": "In Progress", "lost": "Lost", "Clsed Lost": "Lost", "Closde Lost": "Lost", "Lst": "Lost", "ClosedzWon": "Won", "closed  lost": "Lost", "wip": "In Progress", "won": "Won"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ceo": "Chief Executive Officer", "director": "Director", "C.T.O.": "Chief Technology Officer", "CEO": "Chief Executive Officer", "CTO": "Chief Technology Officer", "C.EgO.": "Chief Executive Officer", "Ceo": "Chief Executive Officer", "C.E.O.": "Chief Executive Officer"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "false", "unit": "  kg", "rate": "73.9%", "name": "David  Wong", "currency": "yuan  renminbi", "email": "BCBBLG@CORP.IO   ", "notes2": ""}, {"is_active": "false", "unit": " Km", "rate": "50.2%", "name": "Alice Wong", "currency": "FJD   ", "email": "PEMPIB@TEST.ORG", "notes2": ""}, {"is_active": "false", "unit": "kilograms", "rate": "78.3%", "name": "Eve  Park", "currency": "fiji  dollar", "email": "CBIMJ@EXAMPLE.COM", "notes2": ""}, {"is_active": "Y", "unit": "Kg", "rate": "36.4%", "name": "Eve Smith", "currency": "yuan  renminbi", "email": "kcdjc@corp.io", "notes2": ""}, {"is_active": "FALSE", "unit": "Mi ", "rate": "85.0%", "name": "  Alice Moore", "currency": " fiji dollar", "email": "mchbjhj@example.com", "notes2": ""}, {"is_active": "T", "unit": "in   ", "rate": "0.5%", "name": "Frank Fischer   ", "currency": "FJD", "email": "OOBF@TEST.ORG", "notes2": ""}, {"is_active": "FALSE", "unit": "Mi ", "rate": "10.9%", "name": " Ivan Diaz", "currency": "  Yuan Renminbi", "email": "pdka@mail.com  ", "notes2": ""}, {"is_active": "TRUE", "unit": "mi", "rate": "97.6%", "name": "Omar Diaz", "currency": "BURUNDI FRANC", "email": "  fcimlaj@example.com", "notes2": ""}, {"is_active": "0", "unit": "\"", "rate": "52.4%", "name": "Lena Khan", "currency": "Yuan  Rneminbi", "email": "DNJNO@EXAMPLE.COM", "notes2": ""}, {"is_active": "F", "unit": "km", "rate": "27.8%", "name": "Karl  Petrov", "currency": "   Yuan enminbi", "email": "iokj@test.org", "notes2": ""}, {"is_active": "0", "unit": "  MILE", "rate": "26.0%", "name": "Lena Moore", "currency": "Yuan Renminbi ", "email": " MMPDJ@EXAMPLE.COM", "notes2": ""}, {"is_active": "F", "unit": "  Kg", "rate": "3.4%", "name": "Ivan Park", "currency": "yuan renminbi", "email": "AEEKIDM@MAIL.COM", "notes2": ""}, {"is_active": "TRUE", "unit": "   kilometer", "rate": "65.0%", "name": "Eve  Cruz", "currency": "FIJI DOLLAR", "email": "kpkbloe@example.com ", "notes2": ""}, {"is_active": "0", "unit": " kilograms", "rate": "64.1%", "name": "Mona Moore", "currency": " CNY ", "email": "DJILCE@MAIL.COM   ", "notes2": ""}, {"is_active": "", "unit": "", "rate": "", "name": "", "currency": "", "email": "", "notes2": ""}], "clean": [{"is_active": false, "unit": "kg", "rate": 0.7390000000000001, "name": "David Wong", "currency": "CNY", "email": "bcbblg@corp.io"}, {"is_active": false, "unit": "km", "rate": 0.502, "name": "Alice Wong", "currency": "FJD", "email": "pempib@test.org"}, {"is_active": false, "unit": "kg", "rate": 0.7829999999999999, "name": "Eve Park", "currency": "FJD", "email": "cbimj@example.com"}, {"is_active": true, "unit": "kg", "rate": 0.364, "name": "Eve Smith", "currency": "CNY", "email": "kcdjc@corp.io"}, {"is_active": false, "unit": "mi", "rate": 0.85, "name": "Alice Moore", "currency": "FJD", "email": "mchbjhj@example.com"}, {"is_active": true, "unit": "in", "rate": 0.005, "name": "Frank Fischer", "currency": "FJD", "email": "oobf@test.org"}, {"is_active": false, "unit": "mi", "rate": 0.109, "name": "Ivan Diaz", "currency": "CNY", "email": "pdka@mail.com"}, {"is_active": true, "unit": "mi", "rate": 0.976, "name": "Omar Diaz", "currency": "BIF", "email": "fcimlaj@example.com"}, {"is_active": false, "unit": "in", "rate": 0.524, "name": "Lena Khan", "currency": "CNY", "email": "dnjno@example.com"}, {"is_active": false, "unit": "km", "rate": 0.278, "name": "Karl Petrov", "currency": "CNY", "email": "iokj@test.org"}, {"is_active": false, "unit": "mi", "rate": 0.26, "name": "Lena Moore", "currency": "CNY", "email": "mmpdj@example.com"}, {"is_active": false, "unit": "kg", "rate": 0.034, "name": "Ivan Park", "currency": "CNY", "email": "aeekidm@mail.com"}, {"is_active": true, "unit": "km", "rate": 0.65, "name": "Eve Cruz", "currency": "FJD", "email": "kpkbloe@example.com"}, {"is_active": false, "unit": "kg", "rate": 0.6409999999999999, "name": "Mona Moore", "currency": "CNY", "email": "djilce@mail.com"}], "dirty_cols": ["is_active", "unit", "rate", "name", "currency", "email", "notes2"], "clean_cols": ["is_active", "unit", "rate", "name", "currency", "email"], "plan": {"dataset_summary": "15 rows × 7 columns. 6 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Km": "km", "kilograms": "kg", "Kg": "kg", "Mi": "mi", "\"": "in", "MILE": "mi", "kilometer": "km"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"yuan renminbi": "CNY", "fiji dollar": "FJD", "Yuan Renminbi": "CNY", "BURUNDI FRANC": "BIF", "Yuan Rneminbi": "CNY", "Yuan enminbi": "CNY", "FIJI DOLLAR": "FJD"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"email": "NEGG@MAIL.COM", "status": "OVERDUE", "industry": "Mfg", "department": "MKTG", "is_active": "TRUE"}, {"email": "CNDFNL@CORP.IO", "status": "overdue", "industry": "tech", "department": " I.T. ", "is_active": "N"}, {"email": "  IKIEM@CORP.IO", "status": "PAID", "industry": "TECHNOLOGY", "department": "DEV", "is_active": "No"}, {"email": "  IMBMLL@CORP.IO", "status": "late", "industry": "mfg", "department": "IT", "is_active": "T"}, {"email": "DBLING@TEST.ORG", "status": "OVEDRUE", "industry": " Manufacturing ", "department": "Mktg", "is_active": "Yes"}, {"email": "DMJC@TEST.ORG   ", "status": " late", "industry": "Software", "department": "marketing", "is_active": "Yes"}, {"email": "oklo@corp.io", "status": " Refunded ", "industry": "manufacturing", "department": "  Growth", "is_active": "TRUE"}, {"email": "AEGKKP@TEST.ORG", "status": "pending", "industry": " Mfg ", "department": "it", "is_active": "N"}, {"email": "iegg@test.org", "status": "   pending", "industry": "Industrial", "department": "   DEV", "is_active": "true"}, {"email": "MBPPJJP@CORP.IO", "status": "overdue", "industry": "Mfg", "department": "mktg", "is_active": "TRUE"}, {"email": "ECNEAHJ@TEST.ORG", "status": " REFUNDED", "industry": "Industrial", "department": "I.T.", "is_active": "0"}, {"email": "PJGEAC@MAIL.COM", "status": " late ", "industry": " tech ", "department": " it ", "is_active": "1"}, {"email": "PJGEAC@MAIL.COM", "status": " late ", "industry": " tech ", "department": " it ", "is_active": "1"}, {"email": "  IKIEM@CORP.IO", "status": "PAID", "industry": "TECHNOLOGY", "department": "DEV", "is_active": "No"}], "clean": [{"email": "negg@mail.com", "status": "Overdue", "industry": "Manufacturing", "department": "Marketing", "is_active": true}, {"email": "cndfnl@corp.io", "status": "Overdue", "industry": "Technology", "department": "Information Technology", "is_active": false}, {"email": "ikiem@corp.io", "status": "Paid", "industry": "Technology", "department": "Engineering", "is_active": false}, {"email": "imbmll@corp.io", "status": "Overdue", "industry": "Manufacturing", "department": "Information Technology", "is_active": true}, {"email": "dbling@test.org", "status": "Overdue", "industry": "Manufacturing", "department": "Marketing", "is_active": true}, {"email": "dmjc@test.org", "status": "Overdue", "industry": "Technology", "department": "Marketing", "is_active": true}, {"email": "oklo@corp.io", "status": "Refunded", "industry": "Manufacturing", "department": "Marketing", "is_active": true}, {"email": "aegkkp@test.org", "status": "Pending", "industry": "Manufacturing", "department": "Information Technology", "is_active": false}, {"email": "iegg@test.org", "status": "Pending", "industry": "Manufacturing", "department": "Engineering", "is_active": true}, {"email": "mbppjjp@corp.io", "status": "Overdue", "industry": "Manufacturing", "department": "Marketing", "is_active": true}, {"email": "ecneahj@test.org", "status": "Refunded", "industry": "Manufacturing", "department": "Information Technology", "is_active": false}, {"email": "pjgeac@mail.com", "status": "Overdue", "industry": "Technology", "department": "Information Technology", "is_active": true}], "dirty_cols": ["email", "status", "industry", "department", "is_active"], "clean_cols": ["email", "status", "industry", "department", "is_active"], "plan": {"dataset_summary": "14 rows × 5 columns. 5 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"OVERDUE": "Overdue", "overdue": "Overdue", "PAID": "Paid", "late": "Overdue", "OVEDRUE": "Overdue", "pending": "Pending", "REFUNDED": "Refunded"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "industry", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Mfg": "Manufacturing", "tech": "Technology", "TECHNOLOGY": "Technology", "mfg": "Manufacturing", "Software": "Technology", "manufacturing": "Manufacturing", "Industrial": "Manufacturing"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}, {"name": "department", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"MKTG": "Marketing", "I.T.": "Information Technology", "DEV": "Engineering", "IT": "Information Technology", "Mktg": "Marketing", "marketing": "Marketing", "Growth": "Marketing", "it": "Information Technology", "mktg": "Marketing"}, "rationale": "Unified 9 variant spelling(s) into canonical labels."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}], "flags": []}}
{"dirty": [{"state": "DC", "unit": "l", "email": "fmmimo@corp.io", "phone": "547-720-6317", "company": "Wonka  Co", "currency": "GHS", "notes2": ""}, {"state": "NEW YORK", "unit": "Mi", "email": "olopahm@example.com", "phone": "512-962-5811", "company": "Wonka  Co", "currency": "GHS", "notes2": ""}, {"state": "DC", "unit": "mi", "email": "DBHJL@CORP.IO", "phone": "558-773-9225", "company": "Initech", "currency": "  CONVERTIBLE MARK", "notes2": ""}, {"state": "Michigan", "unit": "Lb", "email": "BALCMFJ@CORP.IO", "phone": "(512)963-6877", "company": "  Initech", "currency": "Bam", "notes2": ""}, {"state": "   Michigan", "unit": "L", "email": "iejhk@mail.com", "phone": "504-269-9108", "company": "Wonka Co", "currency": "Ghs", "notes2": ""}, {"state": "michigan", "unit": "centimetesr", "email": "fmoe@mail.com", "phone": "552.925.4261", "company": "Stark Industries", "currency": "Ghana Cedi", "notes2": ""}, {"state": "  mi", "unit": "lbs", "email": "oeofk@example.com ", "phone": "(531)948-8883", "company": "Soylent Corp", "currency": "ghana cei", "notes2": ""}, {"state": "mi", "unit": "lbs", "email": "ecnk@test.org", "phone": "5848992878", "company": "Vehement", "currency": "GHS", "notes2": ""}, {"state": "   DC", "unit": "l", "email": " ljhdbca@mail.com ", "phone": "5985196500", "company": "Wonka Co", "currency": "Ghana Cedi", "notes2": ""}, {"state": "DISTRICT OF COLUMBIA", "unit": "MI", "email": "BAAIKF@EXAMPLE.COM", "phone": "581.055.0594", "company": "Acme Inc", "currency": "Convertible Mark", "notes2": ""}, {"state": "Michigan", "unit": "l", "email": " COIEM@TEST.ORG", "phone": "584-327-2835", "company": "Globex   ", "currency": "Convertible  Mark", "notes2": ""}, {"state": "Michigan", "unit": "mile", "email": "cpkh@test.org", "phone": "(554)066-8498", "company": "Hooli", "currency": " GHS ", "notes2": ""}, {"state": "DC", "unit": "Lb", "email": "jcia@example.com", "phone": "589-488-5061", "company": "Globex", "currency": "Convertible Mark", "notes2": ""}, {"state": " MI ", "unit": "cm", "email": "IOMKO@CORP.IO", "phone": "5035020924", "company": "Umbrella", "currency": "Convertible Mark", "notes2": ""}, {"state": "mi", "unit": "mi", "email": "ffibkh@mail.com", "phone": "5974944041", "company": "Umbrella  ", "currency": "Ghana Cedi", "notes2": ""}, {"state": "NY", "unit": "Pound", "email": "IIMJGK@CORP.IO", "phone": "5631759547", "company": "Acme Inc", "currency": "GHS", "notes2": ""}, {"state": "dc ", "unit": "l", "email": " EGGALE@CORP.IO ", "phone": "554.435.9134", "company": "Cyberdyne   ", "currency": " Convertible Mark", "notes2": ""}, {"state": "Michiagn", "unit": "lb", "email": "kgjbjjp@mail.com", "phone": "5635526784", "company": "Stark Industries", "currency": "Conveatible Mark", "notes2": ""}, {"state": "MICHIGAN", "unit": "Cm", "email": "llgjoeo@test.org", "phone": "5167830167", "company": "   Initech", "currency": "BAM", "notes2": ""}, {"state": "Michigan", "unit": "MILES", "email": "lbbjlcm@example.com", "phone": "5571517030", "company": "  Initech", "currency": "GHS", "notes2": ""}, {"state": " MI ", "unit": "L", "email": "EPIBP@TEST.ORG ", "phone": "(576)376-2404", "company": "  Hooli", "currency": "Ghana  Cedi", "notes2": ""}, {"state": "michigan", "unit": "Litre", "email": "eofnap@test.org", "phone": "(528)870-5174", "company": "Globex", "currency": "ghs", "notes2": ""}, {"state": "DC", "unit": "Lb", "email": "jcia@example.com", "phone": "589-488-5061", "company": "Globex", "currency": "Convertible Mark", "notes2": ""}, {"state": "", "unit": "", "email": "", "phone": "", "company": "", "currency": "", "notes2": ""}], "clean": [{"state": "District of Columbia", "unit": "L", "email": "fmmimo@corp.io", "phone": "(547) 720-6317", "company": "Wonka Co", "currency": "GHS"}, {"state": "New York", "unit": "mi", "email": "olopahm@example.com", "phone": "(512) 962-5811", "company": "Wonka Co", "currency": "GHS"}, {"state": "District of Columbia", "unit": "mi", "email": "dbhjl@corp.io", "phone": "(558) 773-9225", "company": "Initech", "currency": "BAM"}, {"state": "Michigan", "unit": "lb", "email": "balcmfj@corp.io", "phone": "(512) 963-6877", "company": "Initech", "currency": "BAM"}, {"state": "Michigan", "unit": "L", "email": "iejhk@mail.com", "phone": "(504) 269-9108", "company": "Wonka Co", "currency": "GHS"}, {"state": "Michigan", "unit": "cm", "email": "fmoe@mail.com", "phone": "(552) 925-4261", "company": "Stark Industries", "currency": "GHS"}, {"state": "Michigan", "unit": "lb", "email": "oeofk@example.com", "phone": "(531) 948-8883", "company": "Soylent Corp", "currency": "GHS"}, {"state": "Michigan", "unit": "lb", "email": "ecnk@test.org", "phone": "(584) 899-2878", "company": "Vehement", "currency": "GHS"}, {"state": "District of Columbia", "unit": "L", "email": "ljhdbca@mail.com", "phone": "(598) 519-6500", "company": "Wonka Co", "currency": "GHS"}, {"state": "District of Columbia", "unit": "mi", "email": "baaikf@example.com", "phone": "(581) 055-0594", "company": "Acme Inc", "currency": "BAM"}, {"state": "Michigan", "unit": "L", "email": "coiem@test.org", "phone": "(584) 327-2835", "company": "Globex", "currency": "BAM"}, {"state": "Michigan", "unit": "mi", "email": "cpkh@test.org", "phone": "(554) 066-8498", "company": "Hooli", "currency": "GHS"}, {"state": "District of Columbia", "unit": "lb", "email": "jcia@example.com", "phone": "(589) 488-5061", "company": "Globex", "currency": "BAM"}, {"state": "Michigan", "unit": "cm", "email": "iomko@corp.io", "phone": "(503) 502-0924", "company": "Umbrella", "currency": "BAM"}, {"state": "Michigan", "unit": "mi", "email": "ffibkh@mail.com", "phone": "(597) 494-4041", "company": "Umbrella", "currency": "GHS"}, {"state": "New York", "unit": "lb", "email": "iimjgk@corp.io", "phone": "(563) 175-9547", "company": "Acme Inc", "currency": "GHS"}, {"state": "District of Columbia", "unit": "L", "email": "eggale@corp.io", "phone": "(554) 435-9134", "company": "Cyberdyne", "currency": "BAM"}, {"state": "Michigan", "unit": "lb", "email": "kgjbjjp@mail.com", "phone": "(563) 552-6784", "company": "Stark Industries", "currency": "BAM"}, {"state": "Michigan", "unit": "cm", "email": "llgjoeo@test.org", "phone": "(516) 783-0167", "company": "Initech", "currency": "BAM"}, {"state": "Michigan", "unit": "mi", "email": "lbbjlcm@example.com", "phone": "(557) 151-7030", "company": "Initech", "currency": "GHS"}, {"state": "Michigan", "unit": "L", "email": "epibp@test.org", "phone": "(576) 376-2404", "company": "Hooli", "currency": "GHS"}, {"state": "Michigan", "unit": "L", "email": "eofnap@test.org", "phone": "(528) 870-5174", "company": "Globex", "currency": "GHS"}], "dirty_cols": ["state", "unit", "email", "phone", "company", "currency", "notes2"], "clean_cols": ["state", "unit", "email", "phone", "company", "currency"], "plan": {"dataset_summary": "24 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "state", "detected_semantic_type": "state", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"DC": "District of Columbia", "NEW YORK": "New York", "michigan": "Michigan", "mi": "Michigan", "DISTRICT OF COLUMBIA": "District of Columbia", "MI": "Michigan", "NY": "New York", "dc": "District of Columbia", "Michiagn": "Michigan", "MICHIGAN": "Michigan"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"l": "L", "Mi": "mi", "Lb": "lb", "centimetesr": "cm", "lbs": "lb", "MI": "mi", "mile": "mi", "Pound": "lb", "Cm": "cm", "MILES": "mi", "Litre": "L"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"CONVERTIBLE MARK": "BAM", "Bam": "BAM", "Ghs": "GHS", "Ghana Cedi": "GHS", "ghana cei": "GHS", "Convertible Mark": "BAM", "Conveatible Mark": "BAM", "ghs": "GHS"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"rate": "11.0%", "unit": "inch", "phone": "5022712745", "company": "   Vehement", "unnamed": ""}, {"rate": "6.6%", "unit": "INCHES  ", "phone": "523.025.2616", "company": "Cyberdyne", "unnamed": ""}, {"rate": "93.1%", "unit": "centiimeters", "phone": "(524)391-9659", "company": "Wonka Co", "unnamed": ""}, {"rate": "23.8%", "unit": "centimeter", "phone": "552-099-9702", "company": "Soylent  Corp", "unnamed": ""}, {"rate": "45.0%", "unit": " L ", "phone": "(587)932-8008", "company": "Wonka Co", "unnamed": ""}, {"rate": "58.9%", "unit": "   inch", "phone": "(557)201-4921", "company": "Wonka Co ", "unnamed": ""}, {"rate": "17.6%", "unit": "IN", "phone": "5364568159", "company": " Hooli", "unnamed": ""}, {"rate": "10.4%", "unit": "l", "phone": "(577)171-0530", "company": "Hooli", "unnamed": ""}, {"rate": "47.7%", "unit": "cm", "phone": "555.210.9684", "company": "Stark Industries", "unnamed": ""}, {"rate": "47.8%", "unit": "liters", "phone": "556.507.6294", "company": "Umbrella", "unnamed": ""}, {"rate": "96.2%", "unit": "LITRE", "phone": "5356599504", "company": " Vehement", "unnamed": ""}, {"rate": "82.6%", "unit": "\"", "phone": "530.055.6319", "company": "Vehement", "unnamed": ""}, {"rate": "30.1%", "unit": "CM", "phone": "5997741240", "company": " Cyberdyne ", "unnamed": ""}, {"rate": "1.3%", "unit": "litre", "phone": "564-487-8389", "company": "Wonka Co", "unnamed": ""}, {"rate": "90.0%", "unit": "cm", "phone": "534-780-4249", "company": "Wonka Co ", "unnamed": ""}, {"rate": "3.9%", "unit": "   litre", "phone": "(555)849-8280", "company": "Vehement", "unnamed": ""}, {"rate": "13.3%", "unit": "  liters", "phone": "5315153527", "company": "Vehement", "unnamed": ""}, {"rate": "8.9%", "unit": "cm", "phone": "548-737-3254", "company": "Soylent Corp", "unnamed": ""}, {"rate": "30.1%", "unit": "L", "phone": "5483240221", "company": "Acme Inc", "unnamed": ""}, {"rate": "82.6%", "unit": "\"", "phone": "(577)176-5355", "company": " Vehement ", "unnamed": ""}, {"rate": "74.8%", "unit": "in", "phone": "5711791486", "company": "   Cyberdyne", "unnamed": ""}, {"rate": "88.9%", "unit": "L", "phone": "(543)231-6905", "company": "Globex   ", "unnamed": ""}, {"rate": "76.5%", "unit": "liter", "phone": "520-902-6036", "company": "Hooli   ", "unnamed": ""}, {"rate": "25.4%", "unit": "Liter", "phone": "503-155-1513", "company": "Umbrella", "unnamed": ""}, {"rate": "90.0%", "unit": "cm", "phone": "534-780-4249", "company": "Wonka Co ", "unnamed": ""}], "clean": [{"rate": 0.11, "unit": "in", "phone": "(502) 271-2745", "company": "Vehement"}, {"rate": 0.066, "unit": "in", "phone": "(523) 025-2616", "company": "Cyberdyne"}, {"rate": 0.9309999999999999, "unit": "cm", "phone": "(524) 391-9659", "company": "Wonka Co"}, {"rate": 0.23800000000000002, "unit": "cm", "phone": "(552) 099-9702", "company": "Soylent Corp"}, {"rate": 0.45, "unit": "L", "phone": "(587) 932-8008", "company": "Wonka Co"}, {"rate": 0.589, "unit": "in", "phone": "(557) 201-4921", "company": "Wonka Co"}, {"rate": 0.17600000000000002, "unit": "in", "phone": "(536) 456-8159", "company": "Hooli"}, {"rate": 0.10400000000000001, "unit": "L", "phone": "(577) 171-0530", "company": "Hooli"}, {"rate": 0.47700000000000004, "unit": "cm", "phone": "(555) 210-9684", "company": "Stark Industries"}, {"rate": 0.478, "unit": "L", "phone": "(556) 507-6294", "company": "Umbrella"}, {"rate": 0.9620000000000001, "unit": "L", "phone": "(535) 659-9504", "company": "Vehement"}, {"rate": 0.826, "unit": "in", "phone": "(530) 055-6319", "company": "Vehement"}, {"rate": 0.301, "unit": "cm", "phone": "(599) 774-1240", "company": "Cyberdyne"}, {"rate": 0.013000000000000001, "unit": "L", "phone": "(564) 487-8389", "company": "Wonka Co"}, {"rate": 0.9, "unit": "cm", "phone": "(534) 780-4249", "company": "Wonka Co"}, {"rate": 0.039, "unit": "L", "phone": "(555) 849-8280", "company": "Vehement"}, {"rate": 0.133, "unit": "L", "phone": "(531) 515-3527", "company": "Vehement"}, {"rate": 0.08900000000000001, "unit": "cm", "phone": "(548) 737-3254", "company": "Soylent Corp"}, {"rate": 0.301, "unit": "L", "phone": "(548) 324-0221", "company": "Acme Inc"}, {"rate": 0.826, "unit": "in", "phone": "(577) 176-5355", "company": "Vehement"}, {"rate": 0.748, "unit": "in", "phone": "(571) 179-1486", "company": "Cyberdyne"}, {"rate": 0.889, "unit": "L", "phone": "(543) 231-6905", "company": "Globex"}, {"rate": 0.765, "unit": "L", "phone": "(520) 902-6036", "company": "Hooli"}, {"rate": 0.254, "unit": "L", "phone": "(503) 155-1513", "company": "Umbrella"}], "dirty_cols": ["rate", "unit", "phone", "company", "unnamed"], "clean_cols": ["rate", "unit", "phone", "company"], "plan": {"dataset_summary": "25 rows × 5 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"inch": "in", "INCHES": "in", "centiimeters": "cm", "centimeter": "cm", "IN": "in", "l": "L", "liters": "L", "LITRE": "L", "\"": "in", "CM": "cm", "litre": "L", "liter": "L", "Liter": "L"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}, {"name": "phone", "detected_semantic_type": "phone", "issues": ["inconsistent_formats"], "operations": [{"op": "standardize_phone", "rationale": "Standardized phone formatting."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"job_title": "  Engineering Mgr", "rate": "63.0%", "company": "Acme Inc", "email": "lkobb@corp.io", "notes2": ""}, {"job_title": "VP ", "rate": "23.2%", "company": "Globex", "email": "cppjdeg@test.org", "notes2": ""}, {"job_title": "Vice Prrs", "rate": "14.2%", "company": "Acme Inc", "email": "  klmod@mail.com", "notes2": ""}, {"job_title": " VP ", "rate": "8.4%", "company": "Hooli", "email": "ABMID@MAIL.COM", "notes2": ""}, {"job_title": "Eng Mgr ", "rate": "61.7%", "company": "Stark Industries", "email": " BLMKJMG@TEST.ORG ", "notes2": ""}, {"job_title": "Eng Manager", "rate": "18.7%", "company": " Hooli", "email": "   LDBHP@CORP.IO", "notes2": ""}, {"job_title": "CEO", "rate": "86.0%", "company": "   Hooli", "email": "INMNK@TEST.ORG  ", "notes2": ""}, {"job_title": "ceo   ", "rate": "28.4%", "company": "Soylent Corp", "email": "JHCHJ@CORP.IO", "notes2": ""}, {"job_title": "CEO", "rate": "33.6%", "company": " Globex ", "email": " bkcafb@corp.io", "notes2": ""}, {"job_title": "vic epres", "rate": "81.4%", "company": "Cyberdyne", "email": "FPIBC@EXAMPLE.COM", "notes2": ""}, {"job_title": "Vice President", "rate": "85.8%", "company": "Umbrella", "email": "fbckf@mail.com   ", "notes2": ""}, {"job_title": "Chief Executive Officer", "rate": "31.0%", "company": "Acme Inc", "email": "   ONCKN@EXAMPLE.COM", "notes2": ""}, {"job_title": "chief executive officer", "rate": "42.3%", "company": "Vehement", "email": "bnfglae@test.org", "notes2": ""}, {"job_title": "V.P.", "rate": "0.3%", "company": "Cyberdyne", "email": "onkhd@mail.com", "notes2": ""}, {"job_title": "  vice president", "rate": "96.8%", "company": "Cyberdyne", "email": "JBIPP@EXAMPLE.COM", "notes2": ""}, {"job_title": "C.E.O.", "rate": "15.5%", "company": "Soylent Corp", "email": "PBECMKI@EXAMPLE.COM", "notes2": ""}, {"job_title": "Eng Mgr", "rate": "12.4%", "company": "Cyberdyne   ", "email": "JDEP@EXAMPLE.COM", "notes2": ""}, {"job_title": "Eng Mgr", "rate": "89.5%", "company": "   Vehement", "email": "kgnkn@test.org", "notes2": ""}, {"job_title": "CEO", "rate": "86.0%", "company": "   Hooli", "email": "INMNK@TEST.ORG  ", "notes2": ""}], "clean": [{"job_title": "Engineering Manager", "rate": 0.63, "company": "Acme Inc", "email": "lkobb@corp.io"}, {"job_title": "Vice President", "rate": 0.23199999999999998, "company": "Globex", "email": "cppjdeg@test.org"}, {"job_title": "Vice President", "rate": 0.142, "company": "Acme Inc", "email": "klmod@mail.com"}, {"job_title": "Vice President", "rate": 0.084, "company": "Hooli", "email": "abmid@mail.com"}, {"job_title": "Engineering Manager", "rate": 0.617, "company": "Stark Industries", "email": "blmkjmg@test.org"}, {"job_title": "Engineering Manager", "rate": 0.187, "company": "Hooli", "email": "ldbhp@corp.io"}, {"job_title": "Chief Executive Officer", "rate": 0.86, "company": "Hooli", "email": "inmnk@test.org"}, {"job_title": "Chief Executive Officer", "rate": 0.284, "company": "Soylent Corp", "email": "jhchj@corp.io"}, {"job_title": "Chief Executive Officer", "rate": 0.336, "company": "Globex", "email": "bkcafb@corp.io"}, {"job_title": "Vice President", "rate": 0.8140000000000001, "company": "Cyberdyne", "email": "fpibc@example.com"}, {"job_title": "Vice President", "rate": 0.858, "company": "Umbrella", "email": "fbckf@mail.com"}, {"job_title": "Chief Executive Officer", "rate": 0.31, "company": "Acme Inc", "email": "onckn@example.com"}, {"job_title": "Chief Executive Officer", "rate": 0.423, "company": "Vehement", "email": "bnfglae@test.org"}, {"job_title": "Vice President", "rate": 0.003, "company": "Cyberdyne", "email": "onkhd@mail.com"}, {"job_title": "Vice President", "rate": 0.968, "company": "Cyberdyne", "email": "jbipp@example.com"}, {"job_title": "Chief Executive Officer", "rate": 0.155, "company": "Soylent Corp", "email": "pbecmki@example.com"}, {"job_title": "Engineering Manager", "rate": 0.124, "company": "Cyberdyne", "email": "jdep@example.com"}, {"job_title": "Engineering Manager", "rate": 0.895, "company": "Vehement", "email": "kgnkn@test.org"}], "dirty_cols": ["job_title", "rate", "company", "email", "notes2"], "clean_cols": ["job_title", "rate", "company", "email"], "plan": {"dataset_summary": "19 rows × 5 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["notes2"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "job_title", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Engineering Mgr": "Engineering Manager", "VP": "Vice President", "Vice Prrs": "Vice President", "Eng Mgr": "Engineering Manager", "Eng Manager": "Engineering Manager", "CEO": "Chief Executive Officer", "ceo": "Chief Executive Officer", "vic epres": "Vice President", "chief executive officer": "Chief Executive Officer", "V.P.": "Vice President", "vice president": "Vice President", "C.E.O.": "Chief Executive Officer"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}], "flags": []}}
{"dirty": [{"company": "Acme Inc   ", "rate": "50.9%", "name": "Sara Khan", "unit": "miles", "country": "Saint Kitts and Nevis", "currency": "CNY", "unnamed": ""}, {"company": "Globex", "rate": "97.5%", "name": "Carol Reyes", "unit": "mls", "country": "kn", "currency": "Yuan Renminbi", "unnamed": ""}, {"company": "Stark Industries ", "rate": "36.0%", "name": "  Alice Reyes", "unit": "kilograsm", "country": " fji", "currency": "HKD", "unnamed": ""}, {"company": "Soylent  Corp", "rate": "11.0%", "name": "Omar  Park", "unit": "   MI", "country": "FJI", "currency": "GBP", "unnamed": ""}, {"company": "Wonka Co", "rate": "23.0%", "name": "Sara Cruz", "unit": " Mi ", "country": "BB", "currency": "CNY", "unnamed": ""}, {"company": "  Acme Inc", "rate": "26.1%", "name": "  Ivan Wong", "unit": "Kg", "country": "Sweden", "currency": "£", "unnamed": ""}, {"company": "Initech", "rate": "92.8%", "name": "Lena Diaz", "unit": "kilgram", "country": "  republic of fiji", "currency": "CAD", "unnamed": ""}, {"company": " Umbrella", "rate": "65.3%", "name": "Mona  Cruz", "unit": " ml ", "country": "KN", "currency": "Canadian Dollar", "unnamed": ""}, {"company": "Vehement", "rate": "47.2%", "name": "Grace Reyes", "unit": "KG", "country": "Kingdom Of Sweden", "currency": "Pound Sterling", "unnamed": ""}, {"company": "Soylent  Corp", "rate": "66.5%", "name": "Mona Johnson", "unit": "kilograms", "country": "brb", "currency": "cny", "unnamed": ""}, {"company": "Cyberdyne   ", "rate": "24.9%", "name": "David Cruz ", "unit": "ml", "country": "Saint Kitts and Nevis", "currency": "Yuan Renminbi", "unnamed": ""}, {"company": "Soylent  Corp", "rate": "11.0%", "name": "Omar  Park", "unit": "   MI", "country": "FJI", "currency": "GBP", "unnamed": ""}, {"company": "Soylent  Corp", "rate": "66.5%", "name": "Mona Johnson", "unit": "kilograms", "country": "brb", "currency": "cny", "unnamed": ""}, {"company": "", "rate": "", "name": "", "unit": "", "country": "", "currency": "", "unnamed": ""}], "clean": [{"company": "Acme Inc", "rate": 0.509, "name": "Sara Khan", "unit": "mi", "country": "Saint Kitts and Nevis", "currency": "CNY"}, {"company": "Globex", "rate": 0.975, "name": "Carol Reyes", "unit": "mL", "country": "Saint Kitts and Nevis", "currency": "CNY"}, {"company": "Stark Industries", "rate": 0.36, "name": "Alice Reyes", "unit": "kg", "country": "Fiji", "currency": "HKD"}, {"company": "Soylent Corp", "rate": 0.11, "name": "Omar Park", "unit": "mi", "country": "Fiji", "currency": "GBP"}, {"company": "Wonka Co", "rate": 0.23, "name": "Sara Cruz", "unit": "mi", "country": "Barbados", "currency": "CNY"}, {"company": "Acme Inc", "rate": 0.261, "name": "Ivan Wong", "unit": "kg", "country": "Sweden", "currency": "GBP"}, {"company": "Initech", "rate": 0.9279999999999999, "name": "Lena Diaz", "unit": "kg", "country": "Fiji", "currency": "CAD"}, {"company": "Umbrella", "rate": 0.653, "name": "Mona Cruz", "unit": "mL", "country": "Saint Kitts and Nevis", "currency": "CAD"}, {"company": "Vehement", "rate": 0.47200000000000003, "name": "Grace Reyes", "unit": "kg", "country": "Sweden", "currency": "GBP"}, {"company": "Soylent Corp", "rate": 0.665, "name": "Mona Johnson", "unit": "kg", "country": "Barbados", "currency": "CNY"}, {"company": "Cyberdyne", "rate": 0.249, "name": "David Cruz", "unit": "mL", "country": "Saint Kitts and Nevis", "currency": "CNY"}], "dirty_cols": ["company", "rate", "name", "unit", "country", "currency", "unnamed"], "clean_cols": ["company", "rate", "name", "unit", "country", "currency"], "plan": {"dataset_summary": "14 rows × 7 columns. 6 column(s) need cleanup, 3 table-level fix(es).", "table_operations": [{"op": "drop_empty_columns", "columns": ["unnamed"], "rationale": "Dropped column(s) with no data."}, {"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"miles": "mi", "mls": "mL", "kilograsm": "kg", "MI": "mi", "Mi": "mi", "Kg": "kg", "kilgram": "kg", "ml": "mL", "KG": "kg", "kilograms": "kg"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"kn": "Saint Kitts and Nevis", "fji": "Fiji", "FJI": "Fiji", "BB": "Barbados", "republic of fiji": "Fiji", "KN": "Saint Kitts and Nevis", "Kingdom Of Sweden": "Sweden", "brb": "Barbados"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing"], "operations": [{"op": "canonicalize_categories", "mapping": {"Yuan Renminbi": "CNY", "£": "GBP", "Canadian Dollar": "CAD", "Pound Sterling": "GBP", "cny": "CNY"}, "rationale": "Unified 5 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"amount": "2.136,64", "company": "Soylent  Corp", "name": "Judy Fischer", "country": "srb", "city": "Nanyamba", "status": "trial", "currency": "ALL "}, {"amount": "2.843,79", "company": " Hooli ", "name": "Judy Khan ", "country": "RS", "city": "paris", "status": "churned", "currency": "ALL"}, {"amount": "$6,123.12", "company": "  Hooli", "name": "Bob Smith", "country": "BY", "city": "parsi", "status": "churned", "currency": "Lek"}, {"amount": "$4,012.71", "company": "  Stark Industries", "name": "Frank Moore", "country": "BLR", "city": "paris", "status": "Chrned", "currency": "lek"}, {"amount": "7.088,75", "company": "Soylent Corp   ", "name": "Heidi Park  ", "country": "as ", "city": "   Paris", "status": "Churned", "currency": "  Fiji Dollar"}, {"amount": "$3,241.87", "company": " Umbrella", "name": "Karl Wong", "country": "  Belarus", "city": "paris", "status": "   Trial", "currency": "FJD"}, {"amount": "4.737,19", "company": "Soylent Corp", "name": "Sara Diaz", "country": " Belarus ", "city": " nanyamba ", "status": "FREE TRIAL", "currency": "Lek"}, {"amount": "$8,039.09", "company": " Initech", "name": "  Frank Lee", "country": "AS", "city": " prais", "status": " canceled ", "currency": "FJD"}, {"amount": "8.837,84", "company": "Stark Industries ", "name": " Sara Park", "country": "BY", "city": "  nanyamba", "status": "Trial", "currency": "UAE DIRHAM"}, {"amount": "-100", "company": "Soylent Corp", "name": "David Novak", "country": " ASM ", "city": "paris", "status": "trial", "currency": "ALL"}, {"amount": "8.291,40", "company": " Initech ", "name": "Ivan Reyes", "country": "SRB", "city": "NANYAMBA", "status": "   trial", "currency": " Fiji Dollar"}, {"amount": "--", "company": "Umbrella ", "name": "Priya Moore", "country": "AMERICAN SAMOA", "city": "paris   ", "status": "TRIAL", "currency": "all"}, {"amount": "$8,150.21", "company": " Hooli ", "name": "Carol Smith", "country": "RS   ", "city": "  paris", "status": " Trial", "currency": " UAE Dirham"}, {"amount": "4.023,03", "company": "Soylent Corp", "name": "Lena Khan", "country": "republic of serbia", "city": "nanyamba", "status": "CHURNED", "currency": "Fiji Dollar"}, {"amount": "$5,063.65", "company": "Initech ", "name": " Eve Khan", "country": "Republic of Serbia", "city": "Paris", "status": "trial", "currency": "AED"}, {"amount": "2.297,89", "company": "Initech", "name": "Bob  Novak", "country": "  AS", "city": "Nanyamba ", "status": "canceled ", "currency": "uae diraam"}, {"amount": "$1,463.22", "company": "Umbrella", "name": "Grace Adams ", "country": "Sebia", "city": "paris ", "status": "   TRIAL", "currency": "all"}, {"amount": "$5,656.05", "company": "Umbrella   ", "name": "Sara Petrov", "country": "republic of belarus", "city": "paris", "status": "cancelled", "currency": "Lek"}, {"amount": "355,82", "company": "Soylent Corp", "name": "  Alice Smith", "country": "American Samoa", "city": "NANYAMBA", "status": "TRIAL", "currency": "ALL"}, {"amount": "4.390,54", "company": "   Stark Industries", "name": "Lena Johnson ", "country": "Belarus", "city": "Nanyamba   ", "status": "TRIAL", "currency": "Lek"}, {"amount": "$1,153.97", "company": "Initech", "name": "Mona Park", "country": "AMERICAN SMOA", "city": "Nanyamba", "status": " Trial ", "currency": "AED"}, {"amount": "3.027,95", "company": "Hooli", "name": "Carol Adams", "country": "SRB ", "city": " Nanyamba", "status": "CHURNED", "currency": "UAE Dirham"}, {"amount": "$3,354.34", "company": "Soylent  Corp", "name": "Ivan Reyes", "country": "   Republic of Belarus", "city": "nanyamba ", "status": "free trial", "currency": "Uae Dirham"}, {"amount": "2.540,06", "company": "Soylent  Corp", "name": "Grace  Brandt", "country": "  Belarus", "city": "paxis", "status": "trial ", "currency": "all"}, {"amount": "$6,428.79", "company": "Umbrella", "name": "Sara Reyes", "country": "Belarus", "city": "Nanyamba", "status": " churned", "currency": "AED  "}, {"amount": "na", "company": " Cyberdyne ", "name": "Judy Diaz", "country": "Asm", "city": "Nanyamba", "status": "churned", "currency": "Lek"}, {"amount": "$1,263.61", "company": "Soylent Corp", "name": "Ivan Adams", "country": "AS", "city": "Paris", "status": "canceled", "currency": "UAE Dirham"}, {"amount": "4.986,34", "company": "Cyberdyne", "name": "Mona Reyes", "country": "SRB", "city": "Nanyamba", "status": "trial", "currency": "ALL "}, {"amount": "2.481,65", "company": "Globex", "name": "Bob  Lee", "country": "Serbia", "city": "Nanyamba", "status": "  FREE TRRIAL", "currency": " all "}, {"amount": "4.616,63", "company": "Umbrella", "name": "Lena Adams", "country": "BLR   ", "city": "NANYAMBA", "status": "TRIAL", "currency": "ALL"}, {"amount": "875,11", "company": "Acme Inc", "name": "Ivan Novak", "country": "BLR", "city": "   Paris", "status": "Trial", "currency": "AED"}, {"amount": "--", "company": " Globex ", "name": "   Alice Smith", "country": "SRB", "city": "nanyamba", "status": "TRIAL", "currency": "Lek"}, {"amount": "TBD", "company": "Vehement", "name": "Carol Smith", "country": "American Samoa", "city": "Nanyamba", "status": "free trial", "currency": "UAE Dirham"}, {"amount": "3.576,17", "company": "Soylent Corp", "name": "Grace Fischer", "country": "Belaus", "city": "paris", "status": "TRIAL", "currency": "fiji dollar"}, {"amount": "TBD", "company": "Soylent Corp  ", "name": "Karl Diaz", "country": "Belarus", "city": " paris ", "status": "FREE TRIAL", "currency": " aed "}, {"amount": "$1,629.83", "company": "Initech", "name": "Judy Johnson", "country": "Serbia", "city": " Paris ", "status": " Trial ", "currency": "AED"}, {"amount": "1.780,34", "company": "   Wonka Co", "name": "Frank  Cruz", "country": "AS   ", "city": "nanyamba ", "status": "canceeld", "currency": "UAE Dirham"}, {"amount": "375,32", "company": "   Vehement", "name": "Ivan Novak", "country": " serbia", "city": "PARIS", "status": "canceled", "currency": "Lek"}, {"amount": "null", "company": "Soylent Corp", "name": "Karl Park   ", "country": "American Samoa", "city": "nanyamba", "status": " TRIAL ", "currency": "lek"}, {"amount": "$4,790.43", "company": " Cyberdyne ", "name": "Lena Johnson   ", "country": "belarus", "city": " NANYMABA ", "status": " trial ", "currency": "UAE Driham "}, {"amount": "$2,841.16", "company": "Stark Industries", "name": "Priya  Fischer", "country": "BY", "city": "NANYAMBA", "status": "churned", "currency": " ALL "}, {"amount": "3.385,31", "company": "Vehement", "name": "David Fischer  ", "country": "AS", "city": "PARIS", "status": "TRIAL", "currency": "UAE Dirham"}, {"amount": "$590.68", "company": "Acme Inc   ", "name": "Grace Johnson", "country": "Asm ", "city": "Nanyamba", "status": "cancelled  ", "currency": "aed"}, {"amount": "?", "company": "Acme  Inc", "name": "Judy Park", "country": " ASM ", "city": "   Nanyamba", "status": "   trial", "currency": "UAE DIRHAM"}, {"amount": "$8,269.69", "company": "Wonka Co", "name": "Alice Wong", "country": "Republic o Serbia", "city": "Paris", "status": "free trial", "currency": "lek"}, {"amount": "$6,487.76", "company": "Stark Industries", "name": "Carol Khan", "country": "republic of serbia", "city": "Nanyamba", "status": "churned", "currency": "all   "}, {"amount": "6.425,47", "company": "Stark Industries   ", "name": "Priya Wong", "country": " srb ", "city": "Nanyamba", "status": "trial", "currency": "AED"}, {"amount": "6.019,20", "company": "   Stark Industries", "name": "Bob Lee", "country": "AS", "city": "PARIIS", "status": "trial", "currency": "ALL"}, {"amount": "3.516,95", "company": "Globex", "name": "Karl Adams", "country": "Republic of Belarus", "city": "paris", "status": "Trial", "currency": "LEK"}, {"amount": "$2,198.89", "company": "  Globex", "name": "Eve Petrov ", "country": "SRB", "city": "Paris", "status": "trial", "currency": "Fiji Dollar"}, {"amount": "TBD", "company": "Soylent Corp  ", "name": " Omar Wong", "country": "republic of  serbia", "city": "Nanyamba", "status": "FREE TRIAL", "currency": "   Fiji Dollar"}, {"amount": "5.281,75", "company": "  Vehement", "name": "Frank Park ", "country": "ASM", "city": " nanyamba ", "status": "trial", "currency": "all"}, {"amount": "#N/A", "company": "Globex", "name": "  Bob Wong", "country": "BLR", "city": "Paris", "status": "Trial ", "currency": "FJD"}, {"amount": "--", "company": " Hooli ", "name": "   Bob Moore", "country": "Seerbia", "city": "nanyamba", "status": "CANCEPED", "currency": "UAE DIRHAM"}, {"amount": "$138.89", "company": "Cyberdyne   ", "name": " Bob Khan", "country": "   by", "city": "Paris", "status": "CHURNED", "currency": "fjd"}, {"amount": "8.953,45", "company": "Acme Inc", "name": "Mona Novak   ", "country": "as", "city": "Paris", "status": "canceled", "currency": "Lek"}, {"amount": "$3,163.99", "company": "Stark Industries", "name": "Alice  Brandt", "country": "american samoa", "city": "Paaris", "status": " trial ", "currency": "Lek"}, {"amount": "8.765,17", "company": "Wonka  Co", "name": "  Carol Wong", "country": "republic of serbia", "city": " Nanyamba", "status": "trial", "currency": "FJD"}, {"amount": "$7,243.21", "company": " Hooli ", "name": "Omar  Adams", "country": "  american samoa", "city": "Paris", "status": "canceled", "currency": "   lek"}], "clean": [{"amount": 2136.64, "company": "Soylent Corp", "name": "Judy Fischer", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 2843.79, "company": "Hooli", "name": "Judy Khan", "country": "Serbia", "city": "Paris", "status": "Churned", "currency": "ALL"}, {"amount": 6123.12, "company": "Hooli", "name": "Bob Smith", "country": "Belarus", "city": "Paris", "status": "Churned", "currency": "ALL"}, {"amount": 4012.71, "company": "Stark Industries", "name": "Frank Moore", "country": "Belarus", "city": "Paris", "status": "Churned", "currency": "ALL"}, {"amount": 7088.75, "company": "Soylent Corp", "name": "Heidi Park", "country": "American Samoa", "city": "Paris", "status": "Churned", "currency": "FJD"}, {"amount": 3241.87, "company": "Umbrella", "name": "Karl Wong", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "FJD"}, {"amount": 4737.19, "company": "Soylent Corp", "name": "Sara Diaz", "country": "Belarus", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 8039.09, "company": "Initech", "name": "Frank Lee", "country": "American Samoa", "city": "Paris", "status": "Churned", "currency": "FJD"}, {"amount": 8837.84, "company": "Stark Industries", "name": "Sara Park", "country": "Belarus", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": -100.0, "company": "Soylent Corp", "name": "David Novak", "country": "American Samoa", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 8291.4, "company": "Initech", "name": "Ivan Reyes", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "FJD"}, {"amount": NaN, "company": "Umbrella", "name": "Priya Moore", "country": "American Samoa", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 8150.21, "company": "Hooli", "name": "Carol Smith", "country": "Serbia", "city": "Paris", "status": "Trial", "currency": "AED"}, {"amount": 4023.03, "company": "Soylent Corp", "name": "Lena Khan", "country": "Serbia", "city": "Nanyamba", "status": "Churned", "currency": "FJD"}, {"amount": 5063.65, "company": "Initech", "name": "Eve Khan", "country": "Serbia", "city": "Paris", "status": "Trial", "currency": "AED"}, {"amount": 2297.89, "company": "Initech", "name": "Bob Novak", "country": "American Samoa", "city": "Nanyamba", "status": "Churned", "currency": "AED"}, {"amount": 1463.22, "company": "Umbrella", "name": "Grace Adams", "country": "Serbia", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 5656.05, "company": "Umbrella", "name": "Sara Petrov", "country": "Belarus", "city": "Paris", "status": "Churned", "currency": "ALL"}, {"amount": 355.82, "company": "Soylent Corp", "name": "Alice Smith", "country": "American Samoa", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 4390.54, "company": "Stark Industries", "name": "Lena Johnson", "country": "Belarus", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 1153.97, "company": "Initech", "name": "Mona Park", "country": "American Samoa", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": 3027.95, "company": "Hooli", "name": "Carol Adams", "country": "Serbia", "city": "Nanyamba", "status": "Churned", "currency": "AED"}, {"amount": 3354.34, "company": "Soylent Corp", "name": "Ivan Reyes", "country": "Belarus", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": 2540.06, "company": "Soylent Corp", "name": "Grace Brandt", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 6428.79, "company": "Umbrella", "name": "Sara Reyes", "country": "Belarus", "city": "Nanyamba", "status": "Churned", "currency": "AED"}, {"amount": NaN, "company": "Cyberdyne", "name": "Judy Diaz", "country": "American Samoa", "city": "Nanyamba", "status": "Churned", "currency": "ALL"}, {"amount": 1263.61, "company": "Soylent Corp", "name": "Ivan Adams", "country": "American Samoa", "city": "Paris", "status": "Churned", "currency": "AED"}, {"amount": 4986.34, "company": "Cyberdyne", "name": "Mona Reyes", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 2481.65, "company": "Globex", "name": "Bob Lee", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 4616.63, "company": "Umbrella", "name": "Lena Adams", "country": "Belarus", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 875.11, "company": "Acme Inc", "name": "Ivan Novak", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "AED"}, {"amount": NaN, "company": "Globex", "name": "Alice Smith", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": NaN, "company": "Vehement", "name": "Carol Smith", "country": "American Samoa", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": 3576.17, "company": "Soylent Corp", "name": "Grace Fischer", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "FJD"}, {"amount": NaN, "company": "Soylent Corp", "name": "Karl Diaz", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "AED"}, {"amount": 1629.83, "company": "Initech", "name": "Judy Johnson", "country": "Serbia", "city": "Paris", "status": "Trial", "currency": "AED"}, {"amount": 1780.34, "company": "Wonka Co", "name": "Frank Cruz", "country": "American Samoa", "city": "Nanyamba", "status": "Churned", "currency": "AED"}, {"amount": 375.32, "company": "Vehement", "name": "Ivan Novak", "country": "Serbia", "city": "Paris", "status": "Churned", "currency": "ALL"}, {"amount": NaN, "company": "Soylent Corp", "name": "Karl Park", "country": "American Samoa", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": 4790.43, "company": "Cyberdyne", "name": "Lena Johnson", "country": "Belarus", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": 2841.16, "company": "Stark Industries", "name": "Priya Fischer", "country": "Belarus", "city": "Nanyamba", "status": "Churned", "currency": "ALL"}, {"amount": 3385.31, "company": "Vehement", "name": "David Fischer", "country": "American Samoa", "city": "Paris", "status": "Trial", "currency": "AED"}, {"amount": 590.68, "company": "Acme Inc", "name": "Grace Johnson", "country": "American Samoa", "city": "Nanyamba", "status": "Churned", "currency": "AED"}, {"amount": NaN, "company": "Acme Inc", "name": "Judy Park", "country": "American Samoa", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": 8269.69, "company": "Wonka Co", "name": "Alice Wong", "country": "Serbia", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 6487.76, "company": "Stark Industries", "name": "Carol Khan", "country": "Serbia", "city": "Nanyamba", "status": "Churned", "currency": "ALL"}, {"amount": 6425.47, "company": "Stark Industries", "name": "Priya Wong", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "AED"}, {"amount": 6019.2, "company": "Stark Industries", "name": "Bob Lee", "country": "American Samoa", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 3516.95, "company": "Globex", "name": "Karl Adams", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 2198.89, "company": "Globex", "name": "Eve Petrov", "country": "Serbia", "city": "Paris", "status": "Trial", "currency": "FJD"}, {"amount": NaN, "company": "Soylent Corp", "name": "Omar Wong", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "FJD"}, {"amount": 5281.75, "company": "Vehement", "name": "Frank Park", "country": "American Samoa", "city": "Nanyamba", "status": "Trial", "currency": "ALL"}, {"amount": NaN, "company": "Globex", "name": "Bob Wong", "country": "Belarus", "city": "Paris", "status": "Trial", "currency": "FJD"}, {"amount": NaN, "company": "Hooli", "name": "Bob Moore", "country": "Serbia", "city": "Nanyamba", "status": "Churned", "currency": "AED"}, {"amount": 138.89, "company": "Cyberdyne", "name": "Bob Khan", "country": "Belarus", "city": "Paris", "status": "Churned", "currency": "FJD"}, {"amount": 8953.45, "company": "Acme Inc", "name": "Mona Novak", "country": "American Samoa", "city": "Paris", "status": "Churned", "currency": "ALL"}, {"amount": 3163.99, "company": "Stark Industries", "name": "Alice Brandt", "country": "American Samoa", "city": "Paris", "status": "Trial", "currency": "ALL"}, {"amount": 8765.17, "company": "Wonka Co", "name": "Carol Wong", "country": "Serbia", "city": "Nanyamba", "status": "Trial", "currency": "FJD"}, {"amount": 7243.21, "company": "Hooli", "name": "Omar Adams", "country": "American Samoa", "city": "Paris", "status": "Churned", "currency": "ALL"}], "dirty_cols": ["amount", "company", "name", "country", "city", "status", "currency"], "clean_cols": ["amount", "company", "name", "country", "city", "status", "currency"], "plan": {"dataset_summary": "59 rows × 7 columns. 7 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"srb": "Serbia", "RS": "Serbia", "BY": "Belarus", "BLR": "Belarus", "as": "American Samoa", "AS": "American Samoa", "ASM": "American Samoa", "SRB": "Serbia", "AMERICAN SAMOA": "American Samoa", "republic of serbia": "Serbia", "Republic of Serbia": "Serbia", "Sebia": "Serbia", "republic of belarus": "Belarus", "AMERICAN SMOA": "American Samoa", "Republic of Belarus": "Belarus", "Asm": "American Samoa", "Belaus": "Belarus", "serbia": "Serbia", "belarus": "Belarus", "Republic o Serbia": "Serbia", "Seerbia": "Serbia", "by": "Belarus", "american samoa": "American Samoa"}, "rationale": "Unified 23 variant spelling(s) into canonical labels."}]}, {"name": "city", "detected_semantic_type": "city", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"paris": "Paris", "parsi": "Paris", "nanyamba": "Nanyamba", "prais": "Paris", "NANYAMBA": "Nanyamba", "paxis": "Paris", "PARIS": "Paris", "NANYMABA": "Nanyamba", "PARIIS": "Paris", "Paaris": "Paris"}, "rationale": "Unified 10 variant spelling(s) into canonical labels."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"trial": "Trial", "churned": "Churned", "Chrned": "Churned", "FREE TRIAL": "Trial", "canceled": "Churned", "TRIAL": "Trial", "CHURNED": "Churned", "cancelled": "Churned", "free trial": "Trial", "FREE TRRIAL": "Trial", "canceeld": "Churned", "CANCEPED": "Churned"}, "rationale": "Unified 12 variant spelling(s) into canonical labels."}]}, {"name": "currency", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"Lek": "ALL", "lek": "ALL", "Fiji Dollar": "FJD", "UAE DIRHAM": "AED", "all": "ALL", "UAE Dirham": "AED", "uae diraam": "AED", "Uae Dirham": "AED", "fiji dollar": "FJD", "aed": "AED", "UAE Driham": "AED", "LEK": "ALL", "fjd": "FJD"}, "rationale": "Unified 13 variant spelling(s) into canonical labels."}]}], "flags": [{"column": "amount", "issue": "out_of_range", "action": "flag_only", "rationale": "Value -100 is implausible for 'amount'; flagged for human review, not auto-changed."}]}}
{"dirty": [{"amount": "6.153,50", "is_active": "T", "email": "jgkjpil@mail.com", "unit": "centimeter"}, {"amount": "$4,840.44", "is_active": "TRUE", "email": "JMJNG@TEST.ORG", "unit": "centimeter"}, {"amount": "2.380,75", "is_active": "1", "email": "  gfhgmmp@test.org", "unit": "Centimeter"}, {"amount": "--", "is_active": "true", "email": "  AGCBHDJ@TEST.ORG", "unit": "Cm"}, {"amount": "N/A", "is_active": "Yes", "email": " CELB@CORP.IO ", "unit": "kilogram"}, {"amount": "8.119,92", "is_active": "TRUE", "email": "PJOLNMO@MAIL.COM", "unit": "kilograms   "}, {"amount": "$6,960.00", "is_active": "Y", "email": "OEHCNF@MAIL.COM", "unit": "CM  "}, {"amount": "-", "is_active": "F", "email": "aheeil@example.com", "unit": " kilograms"}, {"amount": "$8,358.59", "is_active": "N", "email": "amobke@test.org", "unit": "cm   "}, {"amount": "TBD", "is_active": "TRUE", "email": "efpkne@corp.io", "unit": "   Kilograms"}, {"amount": "3.434,27", "is_active": "1", "email": "FKIDO@EXAMPLE.COM   ", "unit": "Centimeter"}, {"amount": "2.531,90", "is_active": "T", "email": "epalda@example.com", "unit": "cm"}, {"amount": "$6,234.01", "is_active": "1", "email": "AINEGO@EXAMPLE.COM", "unit": "cm"}, {"amount": "$3,163.62", "is_active": "T", "email": "MHFHHFC@CORP.IO", "unit": "Kg"}, {"amount": "7.614,06", "is_active": "No", "email": "jmdkk@corp.io ", "unit": "cm"}, {"amount": "427,45", "is_active": "true", "email": "KJKLCKN@TEST.ORG", "unit": "centimeters"}, {"amount": "1.801,74", "is_active": "true", "email": "pgml@test.org", "unit": "centimeters"}, {"amount": "2.752,50", "is_active": "0", "email": "  fbpdjf@mail.com", "unit": "KG"}, {"amount": "$2,281.92", "is_active": "0", "email": " cmdangb@corp.io ", "unit": "CENTIMETER"}, {"amount": "$5,205.53", "is_active": "TRUE", "email": "oaebkch@example.com", "unit": "kg"}, {"amount": "$1,799.41", "is_active": "No", "email": "ocpgpb@mail.com", "unit": " kg "}, {"amount": "$8,721.64", "is_active": "No", "email": "EIAOKD@TEST.ORG", "unit": "   kilogram"}, {"amount": "8.024,00", "is_active": "T", "email": " blcgmb@example.com ", "unit": "kilogram "}, {"amount": "-", "is_active": "FALSE", "email": "OGHP@CORP.IO", "unit": "Cm   "}, {"amount": "$4,916.49", "is_active": "T", "email": "ajoepc@test.org", "unit": "KG"}, {"amount": "", "is_active": "", "email": "", "unit": ""}], "clean": [{"amount": 6153.5, "is_active": true, "email": "jgkjpil@mail.com", "unit": "cm"}, {"amount": 4840.44, "is_active": true, "email": "jmjng@test.org", "unit": "cm"}, {"amount": 2380.75, "is_active": true, "email": "gfhgmmp@test.org", "unit": "cm"}, {"amount": NaN, "is_active": true, "email": "agcbhdj@test.org", "unit": "cm"}, {"amount": NaN, "is_active": true, "email": "celb@corp.io", "unit": "kg"}, {"amount": 8119.92, "is_active": true, "email": "pjolnmo@mail.com", "unit": "kg"}, {"amount": 6960.0, "is_active": true, "email": "oehcnf@mail.com", "unit": "cm"}, {"amount": NaN, "is_active": false, "email": "aheeil@example.com", "unit": "kg"}, {"amount": 8358.59, "is_active": false, "email": "amobke@test.org", "unit": "cm"}, {"amount": NaN, "is_active": true, "email": "efpkne@corp.io", "unit": "kg"}, {"amount": 3434.27, "is_active": true, "email": "fkido@example.com", "unit": "cm"}, {"amount": 2531.9, "is_active": true, "email": "epalda@example.com", "unit": "cm"}, {"amount": 6234.01, "is_active": true, "email": "ainego@example.com", "unit": "cm"}, {"amount": 3163.62, "is_active": true, "email": "mhfhhfc@corp.io", "unit": "kg"}, {"amount": 7614.06, "is_active": false, "email": "jmdkk@corp.io", "unit": "cm"}, {"amount": 427.45, "is_active": true, "email": "kjklckn@test.org", "unit": "cm"}, {"amount": 1801.74, "is_active": true, "email": "pgml@test.org", "unit": "cm"}, {"amount": 2752.5, "is_active": false, "email": "fbpdjf@mail.com", "unit": "kg"}, {"amount": 2281.92, "is_active": false, "email": "cmdangb@corp.io", "unit": "cm"}, {"amount": 5205.53, "is_active": true, "email": "oaebkch@example.com", "unit": "kg"}, {"amount": 1799.41, "is_active": false, "email": "ocpgpb@mail.com", "unit": "kg"}, {"amount": 8721.64, "is_active": false, "email": "eiaokd@test.org", "unit": "kg"}, {"amount": 8024.0, "is_active": true, "email": "blcgmb@example.com", "unit": "kg"}, {"amount": NaN, "is_active": false, "email": "oghp@corp.io", "unit": "cm"}, {"amount": 4916.49, "is_active": true, "email": "ajoepc@test.org", "unit": "kg"}], "dirty_cols": ["amount", "is_active", "email", "unit"], "clean_cols": ["amount", "is_active", "email", "unit"], "plan": {"dataset_summary": "26 rows × 4 columns. 4 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}], "columns": [{"name": "amount", "detected_semantic_type": "currency", "issues": ["numeric_stored_as_text", "currency_symbols", "disguised_nulls"], "operations": [{"op": "normalize_disguised_nulls", "rationale": "Converted N/A, '-', 'null' etc. to true missing."}, {"op": "parse_currency", "rationale": "Stripped currency symbols/grouping; parsed to number."}]}, {"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "email", "detected_semantic_type": "email", "issues": ["casing", "whitespace"], "operations": [{"op": "normalize_email", "rationale": "Lowercased and trimmed email addresses."}]}, {"name": "unit", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"centimeter": "cm", "Centimeter": "cm", "Cm": "cm", "kilogram": "kg", "kilograms": "kg", "CM": "cm", "Kilograms": "kg", "Kg": "kg", "centimeters": "cm", "KG": "kg", "CENTIMETER": "cm"}, "rationale": "Unified 11 variant spelling(s) into canonical labels."}]}], "flags": []}}
{"dirty": [{"is_active": "true", "status": "In Progress", "name": "Mona Lee"}, {"is_active": "Yes", "status": "   close-dwon", "name": "   Lena Ali"}, {"is_active": "Yes", "status": "  in prgress", "name": "Bob Cruz"}, {"is_active": "TRUE", "status": "closed-won", "name": "Grace Diaz"}, {"is_active": "FALSE", "status": "won", "name": "Grace Park"}, {"is_active": "true", "status": "won", "name": "Sara Adams   "}, {"is_active": "true", "status": "OPEN", "name": "Heidi Novak"}, {"is_active": "1", "status": "Open", "name": "Bob Ali"}, {"is_active": "T", "status": "Open", "name": "  Frank Adams"}, {"is_active": "Yes", "status": "in-progress", "name": "Heidi Novak"}, {"is_active": "1", "status": "ongoing", "name": "Mona Petrov "}, {"is_active": "TRUE", "status": " Open ", "name": "Omar Moore"}, {"is_active": "Y", "status": "open", "name": "Judy Ali"}, {"is_active": "1", "status": " open", "name": "David Petrov  "}, {"is_active": "Yes", "status": "   close-dwon", "name": "   Lena Ali"}], "clean": [{"is_active": true, "status": "In Progress", "name": "Mona Lee"}, {"is_active": true, "status": "Won", "name": "Lena Ali"}, {"is_active": true, "status": "In Progress", "name": "Bob Cruz"}, {"is_active": true, "status": "Won", "name": "Grace Diaz"}, {"is_active": false, "status": "Won", "name": "Grace Park"}, {"is_active": true, "status": "Won", "name": "Sara Adams"}, {"is_active": true, "status": "Open", "name": "Heidi Novak"}, {"is_active": true, "status": "Open", "name": "Bob Ali"}, {"is_active": true, "status": "Open", "name": "Frank Adams"}, {"is_active": true, "status": "In Progress", "name": "Heidi Novak"}, {"is_active": true, "status": "In Progress", "name": "Mona Petrov"}, {"is_active": true, "status": "Open", "name": "Omar Moore"}, {"is_active": true, "status": "Open", "name": "Judy Ali"}, {"is_active": true, "status": "Open", "name": "David Petrov"}], "dirty_cols": ["is_active", "status", "name"], "clean_cols": ["is_active", "status", "name"], "plan": {"dataset_summary": "15 rows × 3 columns. 3 column(s) need cleanup, 1 table-level fix(es).", "table_operations": [{"op": "drop_exact_duplicates", "rationale": "Removed 1 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"close-dwon": "Won", "in prgress": "In Progress", "closed-won": "Won", "won": "Won", "OPEN": "Open", "in-progress": "In Progress", "ongoing": "In Progress", "open": "Open"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "name", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}], "flags": []}}
{"dirty": [{"is_active": "1", "country": "TJK  ", "company": "Acme Inc", "rate": "74.8%"}, {"is_active": "N", "country": "TTO", "company": "  Umbrella", "rate": "4.2%"}, {"is_active": "true", "country": "trinidad and tobago", "company": " Initech", "rate": "14.1%"}, {"is_active": "Y", "country": "  Trinidad And Tobago", "company": "   Globex", "rate": "90.7%"}, {"is_active": "No", "country": " republic of tajikistan", "company": "Vehement", "rate": "72.2%"}, {"is_active": "false", "country": "tj", "company": "Initech", "rate": "34.4%"}, {"is_active": "N", "country": "TTO", "company": "   Initech", "rate": "79.2%"}, {"is_active": "false", "country": "TJK", "company": "Acme Inc", "rate": "41.8%"}, {"is_active": "Y", "country": "TTO   ", "company": "  Umbrella", "rate": "25.8%"}, {"is_active": "Y", "country": "republic of trinidad and tobago", "company": "  Initech", "rate": "60.6%"}, {"is_active": "true", "country": "Republic of Trinidad and Tobago", "company": " Hooli ", "rate": "43.1%"}, {"is_active": "N", "country": "TTO", "company": "   Initech", "rate": "79.2%"}, {"is_active": "true", "country": "trinidad and tobago", "company": " Initech", "rate": "14.1%"}, {"is_active": "", "country": "", "company": "", "rate": ""}], "clean": [{"is_active": true, "country": "Tajikistan", "company": "Acme Inc", "rate": 0.748}, {"is_active": false, "country": "Trinidad and Tobago", "company": "Umbrella", "rate": 0.042}, {"is_active": true, "country": "Trinidad and Tobago", "company": "Initech", "rate": 0.141}, {"is_active": true, "country": "Trinidad and Tobago", "company": "Globex", "rate": 0.907}, {"is_active": false, "country": "Tajikistan", "company": "Vehement", "rate": 0.722}, {"is_active": false, "country": "Tajikistan", "company": "Initech", "rate": 0.344}, {"is_active": false, "country": "Trinidad and Tobago", "company": "Initech", "rate": 0.792}, {"is_active": false, "country": "Tajikistan", "company": "Acme Inc", "rate": 0.418}, {"is_active": true, "country": "Trinidad and Tobago", "company": "Umbrella", "rate": 0.258}, {"is_active": true, "country": "Trinidad and Tobago", "company": "Initech", "rate": 0.606}, {"is_active": true, "country": "Trinidad and Tobago", "company": "Hooli", "rate": 0.431}], "dirty_cols": ["is_active", "country", "company", "rate"], "clean_cols": ["is_active", "country", "company", "rate"], "plan": {"dataset_summary": "14 rows × 4 columns. 4 column(s) need cleanup, 2 table-level fix(es).", "table_operations": [{"op": "drop_empty_rows", "rationale": "Removed 1 fully-empty row."}, {"op": "drop_exact_duplicates", "rationale": "Removed 2 exact duplicate row(s)."}], "columns": [{"name": "is_active", "detected_semantic_type": "boolean", "issues": ["inconsistent_booleans"], "operations": [{"op": "standardize_boolean", "rationale": "Mapped Yes/Y/1/TRUE → true, No/N/0/FALSE → false."}]}, {"name": "country", "detected_semantic_type": "country", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"TJK": "Tajikistan", "TTO": "Trinidad and Tobago", "trinidad and tobago": "Trinidad and Tobago", "Trinidad And Tobago": "Trinidad and Tobago", "republic of tajikistan": "Tajikistan", "tj": "Tajikistan", "republic of trinidad and tobago": "Trinidad and Tobago", "Republic of Trinidad and Tobago": "Trinidad and Tobago"}, "rationale": "Unified 8 variant spelling(s) into canonical labels."}]}, {"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}], "flags": []}}
{"dirty": [{"company": "Cyberdyne", "rate": "38.9%", "status": "In Progress"}, {"company": "Soylent Corp", "rate": "3.0%", "status": "ongoing"}, {"company": "Cyberdyne", "rate": "75.7%", "status": "Open"}, {"company": "Stark Industries", "rate": "57.9%", "status": "WIP"}, {"company": "Vehement", "rate": "61.5%", "status": "NEW"}, {"company": "Hooli", "rate": "93.2%", "status": "open"}, {"company": "  Initech", "rate": "61.9%", "status": "new"}, {"company": "   Initech", "rate": "53.4%", "status": "in progress"}, {"company": " Globex ", "rate": "69.3%", "status": "in-progress"}, {"company": "Initech", "rate": "29.9%", "status": "open   "}, {"company": "Globex", "rate": "98.5%", "status": "  in progress"}, {"company": "Soylent Corp", "rate": "3.1%", "status": "new"}, {"company": "Wonka Co", "rate": "21.0%", "status": "new"}], "clean": [{"company": "Cyberdyne", "rate": 0.389, "status": "In Progress"}, {"company": "Soylent Corp", "rate": 0.03, "status": "In Progress"}, {"company": "Cyberdyne", "rate": 0.757, "status": "Open"}, {"company": "Stark Industries", "rate": 0.579, "status": "In Progress"}, {"company": "Vehement", "rate": 0.615, "status": "Open"}, {"company": "Hooli", "rate": 0.932, "status": "Open"}, {"company": "Initech", "rate": 0.619, "status": "Open"}, {"company": "Initech", "rate": 0.534, "status": "In Progress"}, {"company": "Globex", "rate": 0.693, "status": "In Progress"}, {"company": "Initech", "rate": 0.299, "status": "Open"}, {"company": "Globex", "rate": 0.985, "status": "In Progress"}, {"company": "Soylent Corp", "rate": 0.031, "status": "Open"}, {"company": "Wonka Co", "rate": 0.21, "status": "Open"}], "dirty_cols": ["company", "rate", "status"], "clean_cols": ["company", "rate", "status"], "plan": {"dataset_summary": "13 rows × 3 columns. 3 column(s) need cleanup, 0 table-level fix(es).", "table_operations": [], "columns": [{"name": "company", "detected_semantic_type": "text", "issues": ["whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed leading/trailing and doubled spaces."}]}, {"name": "rate", "detected_semantic_type": "percent", "issues": ["numeric_stored_as_text"], "operations": [{"op": "parse_percent", "rationale": "Parsed percent text to a fraction."}]}, {"name": "status", "detected_semantic_type": "categorical", "issues": ["inconsistent_categories", "casing", "whitespace"], "operations": [{"op": "strip_whitespace", "rationale": "Trimmed surrounding/doubled spaces."}, {"op": "canonicalize_categories", "mapping": {"ongoing": "In Progress", "WIP": "In Progress", "NEW": "Open", "open": "Open", "new": "Open", "in progress": "In Progress", "in-progress": "In Progress"}, "rationale": "Unified 7 variant spelling(s) into canonical labels."}]}], "flags": []}}