# CPCB Clean Dataset -- Final Data Dictionary

Generated from: `merged_cpcb_clean.csv`  
File size on disk: **151.6 MB**

## Dataset Overview
- Rows: **972,562**
- Columns: **20**
- Date range: **2009-01-01 to 2025-12-31** (0 rows with unparseable date)

## Column-by-Column Dictionary
Every `dtype` below is read directly from `df.dtypes` on the actual loaded file -- not assumed. Every `unit` is sourced as noted; where no unit applies (categorical/text/boolean fields) that's stated explicitly instead of left blank.

### `date`
- **dtype (as loaded by pandas):** `object`
- **unit:** calendar date (IST local, no time-of-day)  _(source: clean_cpcb.py docstring: 'CPCB timestamps are IST local')_
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 6,209
- **top 5 values:** {'2025-10-21': 514, '2025-10-24': 514, '2025-10-22': 513, '2025-10-20': 513, '2025-10-23': 512}
- **sample values:** ['2009-01-01', '2009-01-01', '2009-01-01']

### `station`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- free-text station name, parsed from source filename
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 532
- **top 5 values:** {'Nehru Nagar': 8005, 'Shastri Nagar': 7302, 'Collectorate': 6577, 'Anand Vihar': 6327, 'Shadipur': 5941}
- **sample values:** ['DTU', 'ITO', 'NSIT Dwarka']

### `city`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- free-text city name, parsed from source filename
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 289
- **top 5 values:** {'Delhi': 116163, 'Mumbai': 46430, 'Bengaluru': 39281, 'Hyderabad': 30930, 'Chennai': 25919}
- **sample values:** ['Delhi', 'Delhi', 'Delhi']

### `state`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- categorical -- from State_Geocoded (Nominatim's returned state), PRIMARY state field
- **non-null:** 850,652 / 972,562 (87.5%)
- **unique values:** 30
- **top 5 values:** {'Maharashtra': 127095, nan: 121910, 'Uttar Pradesh': 120153, 'Karnataka': 86218, 'Haryana': 68098}
- **sample values:** ['Karnataka', 'Karnataka', 'Karnataka']

### `state_hint`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- categorical -- from State_Hint, canonical state mapped from the source folder name
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 31
- **top 5 values:** {'Maharashtra': 127095, 'Delhi': 121910, 'Uttar Pradesh': 120153, 'Karnataka': 86218, 'Bihar': 72673}
- **sample values:** ['Delhi', 'Delhi', 'Delhi']

### `state_mismatch_flag`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- boolean -- True if state_hint disagrees with the geocoded state
- **non-null:** 850,652 / 972,562 (87.5%)
- **unique values:** 2
- **top 5 values:** {False: 841064, nan: 121910, True: 9588}
- **sample values:** [False, False, False]

### `geocode_confidence`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- categorical enum: high / medium / medium_state_mismatch / low_ambiguous / not_found
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 3
- **top 5 values:** {'high': 841064, 'medium_state_mismatch': 126158, 'low_ambiguous': 5340}
- **sample values:** ['medium_state_mismatch', 'medium_state_mismatch', 'medium_state_mismatch']

### `agency`
- **dtype (as loaded by pandas):** `object`
- **unit:** N/A -- free-text monitoring agency code (e.g. CPCB, UPPCB, MPCB), parsed from source filename
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 49
- **top 5 values:** {'UPPCB': 99695, 'MPCB': 98974, 'CPCB': 80590, 'DPCC': 73243, 'KSPCB': 70341}
- **sample values:** ['CPCB', 'CPCB', 'CPCB']

### `lat`
- **dtype (as loaded by pandas):** `float64`
- **unit:** decimal degrees (WGS84)  _(source: Nominatim/OpenStreetMap geocoder output format)_
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 289
- **min:** 8.4882  **max:** 34.0747  **mean:** 22.9735  **median:** 24.7964

### `lon`
- **dtype (as loaded by pandas):** `float64`
- **unit:** decimal degrees (WGS84)  _(source: Nominatim/OpenStreetMap geocoder output format)_
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 289
- **min:** 70.7775  **max:** 94.6394  **mean:** 78.5007  **median:** 77.3091

### `coverage_area_km2`
- **dtype (as loaded by pandas):** `float64`
- **unit:** km²  _(source: computed from Nominatim bounding box by cpcb_pipeline.py)_
- **non-null:** 972,562 / 972,562 (100.0%)
- **unique values:** 266
- **min:** 0  **max:** 63,238.2000  **mean:** 2,805.6955  **median:** 1,141.7000

### `pm25`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'PM2.5 (µg/m³)')_
- **non-null:** 906,236 / 972,562 (93.2%)
- **unique values:** 156,644
- **min:** 0.0100  **max:** 1,000  **mean:** 57.8282  **median:** 41.0100

### `pm10`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'PM10 (µg/m³)')_
- **non-null:** 861,062 / 972,562 (88.5%)
- **unique values:** 179,248
- **min:** 0.0300  **max:** 1,000  **mean:** 118.8644  **median:** 92.7900

### `no2`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'NO2 (µg/m³)')_
- **non-null:** 939,539 / 972,562 (96.6%)
- **unique values:** 153,689
- **min:** 0.0100  **max:** 500  **mean:** 25.9318  **median:** 18.8000

### `so2`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'SO2 (µg/m³)')_
- **non-null:** 915,167 / 972,562 (94.1%)
- **unique values:** 138,954
- **min:** 0.0100  **max:** 199.8300  **mean:** 12.6577  **median:** 9.0200

### `co`
- **dtype (as loaded by pandas):** `float64`
- **unit:** mg/m³  _(source: from raw column 'CO (mg/m³)' -- NOTE different unit family than other pollutants)_
- **non-null:** 920,104 / 972,562 (94.6%)
- **unique values:** 79,333
- **min:** 0  **max:** 50  **mean:** 1.1384  **median:** 0.7300

### `o3`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'Ozone (µg/m³)')_
- **non-null:** 892,019 / 972,562 (91.7%)
- **unique values:** 151,934
- **min:** 0.0100  **max:** 499.6300  **mean:** 30.9691  **median:** 25.8400

### `nh3`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'NH3 (µg/m³)')_
- **non-null:** 779,161 / 972,562 (80.1%)
- **unique values:** 137,757
- **min:** 0.0100  **max:** 498.4000  **mean:** 24.3772  **median:** 17.8800

### `no`
- **dtype (as loaded by pandas):** `float64`
- **unit:** µg/m³  _(source: from raw column 'NO (µg/m³)')_
- **non-null:** 923,787 / 972,562 (95.0%)
- **unique values:** 141,244
- **min:** 0.0100  **max:** 500  **mean:** 14.9131  **median:** 7.1064

### `nox_ppb`
- **dtype (as loaded by pandas):** `float64`
- **unit:** ppb  _(source: from raw column 'NOx (ppb)' -- NOTE ppb, not µg/m³ like the others)_
- **non-null:** 933,951 / 972,562 (96.0%)
- **unique values:** 157,146
- **min:** 0  **max:** 499.4400  **mean:** 29.7808  **median:** 19.9900