Karmane/cpsc-consumer-product-recalls-enriched

CPSC Consumer Product Recalls Enriched Dataset

This dataset consolidates official U.S. Consumer Product Safety Commission recall announcements into a single analysis-ready dataframe covering June 7, 1973 through May 21, 2026.

Each row represents one CPSC recall announcement enriched with parsed units and pricing ranges, sale period dates, responsible-firm rollups, retailer and marketplace signals, country-of-manufacture normalization, remedy flags, and hazard-category indicators. The result is easier to use for product compliance screening, marketplace trust and safety review, importer diligence, insurer monitoring, consumer-risk research, and retail operations analysis than the raw announcement export alone.

The source data comes directly from the CPSC recall listing CSV and reproducible row-level SaferProducts API URLs.

---

Dataset Preview

record_id	recall_number	product_safety_warning_number	has_product_safety_warning	announcement_date	announcement_year	announcement_month	announcement_quarter	recall_heading	product_name	product_description	hazard_description	remedy_description	consumer_action	original_product_safety_warning_announcement	remedy_type	remedy_type_count	remedy_refund	remedy_repair	remedy_replace	remedy_dispose	remedy_inspect	remedy_new_instructions	remedy_no_remedy_available	units_text	units_affected_value	units_affected_bucket	incidents_text	no_incidents_reported	reported_incident_count_lower_bound	reported_injury_count_lower_bound	reported_death_count_lower_bound	sold_at_label	sold_at	sale_start_date	sale_end_date	sale_duration_days	price_low_usd	price_high_usd	importers	manufacturers	distributors	responsible_firm	responsible_firm_role	manufactured_in	manufactured_in_normalized	imported_product_flag	child_product_flag	mandatory_standard_violation_flag	serious_injury_or_death_flag	hazard_fire	hazard_drowning	hazard_fall	hazard_tip_over	hazard_battery_ingestion	hazard_choking	hazard_suffocation	hazard_entrapment	hazard_magnet	hazard_electrocution	hazard_laceration	hazard_amputation	hazard_crash	hazard_keyword_count	primary_hazard_category	sold_online_flag	sold_on_amazon	sold_on_walmart	sold_on_target	sold_on_qvc	sold_on_costco	sold_on_ebay	sold_on_wayfair	sold_on_shein	source_csv_url	source_api_url	source_page_url	source_domain	last_collected_at
cpsc-recall-73-003	73-003		False	1973-06-07 00:00:00+00:00	1973	6	2	Consumer Product Safety Commission Says Some Tappan Built-In Ovens May Pose Hazard To Consumers	Tappan Built-In Oven	WASHINGTON, D.C. (June 8) - The Consumer Product Safety Commission Friday said it has learned from t... [1,965 more chars]	A possible loose fitting connection could result in a hazard to consumers from gas leakage.	CPSC Chairman Richard 0. Simpson said the Commission is asking wholesalers, distributors, and retail... [143 more chars]			No Remedy Available	1	False	False	False	False	False	False	True			Unknown		False																		False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	0	other	False	False	False	False	False	False	False	False	False	https://www.cpsc.gov/s3fs-public/recall-data/recalls_recall_listing.csv	https://www.saferproducts.gov/RestWebServices/Recall?RecallNumber=73003&format=json	https://www.cpsc.gov/Recalls	cpsc.gov	2026-05-31 16:44:20+00:00
cpsc-recall-73-014	73-014		False	1973-08-19 00:00:00+00:00	1973	8	3	CPSC Bans Three Spray Adhesives-Asks Manufacturers Of Others To Halt Production	Foil Art Spray Adhesive Scotch Brand Spra-Ment Krylon Spray Adhesive	WASHINGTON, D.C. --The Consumer Product Safety Commission Monday, acting under the Hazardous Substan... [2,048 more chars]	Possibly linked to chromosome breakage and resulting multiple birth defects.	Manufacturers of all other aerosol spray adhesives are to voluntarily discontinue production and dis... [92 more chars]			Dispose	1	False	False	False	True	False	False	False			Unknown		False																		False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	0	other	False	False	False	False	False	False	False	False	False	https://www.cpsc.gov/s3fs-public/recall-data/recalls_recall_listing.csv	https://www.saferproducts.gov/RestWebServices/Recall?RecallNumber=73014&format=json	https://www.cpsc.gov/Recalls	cpsc.gov	2026-05-31 16:44:20+00:00
cpsc-recall-73-015	73-015		False	1973-08-20 00:00:00+00:00	1973	8	3	CPSC Warns Of Electric Shock Dangers From "Little Wonder TV Antenna"	Little Wonder TV Antenna	WASHINGTON, D.C. (August 21)--Citing the potential for severe electric shock hazard, the Consumer Pr... [1,900 more chars]	The product "contains no safety device" to prevent electrical shock and possible electrocution.	Consumers to immediately unplug and then disconnect the "Little Wonder TV Antenna."			Dispose	1	False	False	False	True	False	False	False			Unknown		False																		False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	1	electrocution	False	False	False	False	False	False	False	False	False	https://www.cpsc.gov/s3fs-public/recall-data/recalls_recall_listing.csv	https://www.saferproducts.gov/RestWebServices/Recall?RecallNumber=73015&format=json	https://www.cpsc.gov/Recalls	cpsc.gov	2026-05-31 16:44:20+00:00
cpsc-recall-73-017	73-017		False	1973-08-27 00:00:00+00:00	1973	8	3	CPSC Bans Nine Additional Aerosol Spray Adhesives-Reaffirms Recommendation To Consumers To Discontin... [29 more chars]			The Consumer Product Safety Commission today banned nine additional aerosol spray adhesives as "pres... [223 more chars]					0	False	False	False	False	False	False	False			Unknown		False																		False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	0	other	False	False	False	False	False	False	False	False	False	https://www.cpsc.gov/s3fs-public/recall-data/recalls_recall_listing.csv	https://www.saferproducts.gov/RestWebServices/Recall?RecallNumber=73017&format=json	https://www.cpsc.gov/Recalls	cpsc.gov	2026-05-31 16:44:20+00:00
cpsc-recall-73-021	73-021		False	1973-09-14 00:00:00+00:00	1973	9	3	CPSC Warns Of Possible Fire Hazard	Zenith 19-inch Table Model Color Television Set	WASHINGTON, D.C. (Sept. 14) --The Consumer Product Safety Commission-- acting on a report to it from... [1,997 more chars]	Possible fire hazard	The Commission said that consumers should immediately unplug the sets and discontinue use until they... [13 more chars]			Inspect	1	False	False	False	False	True	False	False	12,000	12000.0	10K-99,999	Zenith learned of the problem following a fire in one of the sets in Milwaukee, Wisconsin. There wer... [48 more chars]	False																		False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	1	fire	False	False	False	False	False	False	False	False	False	https://www.cpsc.gov/s3fs-public/recall-data/recalls_recall_listing.csv	https://www.saferproducts.gov/RestWebServices/Recall?RecallNumber=73021&format=json	https://www.cpsc.gov/Recalls	cpsc.gov	2026-05-31 16:44:20+00:00

Access Requirements (Paid Dataset)

This dataset is behind manual gated access.

To obtain access:

1. Purchase the dataset here:
   https://thearticulated.gumroad.com/l/cpsc-consumer-product-recalls-enriched

2. Provide your Hugging Face username at checkout.

3. Return to this Hugging Face page and click:
   "Request Access"

4. Your access will be approved within 1-12 hours.

Once approved, you can use the Python snippet at the bottom of this README to load the dataset.

---

Dataset Structure

Total rows: 9,866

Total columns: 79

Splits

• data: 9,866 rows

Data Files

• data: data/data.parquet

Data Dictionary

The table below describes the columns included in this dataset.

column	pandas_dtype	dataset_type	description
record_id	object	string	Text column. Stable unique identifier generated from the CPSC recall number.
recall_number	object	string	Text column. Primary CPSC recall number for the announcement.
product_safety_warning_number	object	string	Text column. Product safety warning number linked to the recall announcement when one is listed.
has_product_safety_warning	bool	boolean	True/false column. Boolean flag indicating that the announcement references a linked CPSC product safety warning.
announcement_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Date the recall announcement was published by CPSC.
announcement_year	int32	integer	Whole-number numeric column. Calendar year extracted from the announcement date.
announcement_month	int32	integer	Whole-number numeric column. Calendar month extracted from the announcement date.
announcement_quarter	int32	integer	Whole-number numeric column. Calendar quarter extracted from the announcement date.
recall_heading	object	string	Text column. Official CPSC recall headline for the announcement.
product_name	object	string	Text column. Product name reported by CPSC for the recalled item.
product_description	object	long text	Free-text column containing longer text values. Cleaned CPSC product description text.
hazard_description	object	long text	Free-text column containing longer text values. Cleaned CPSC hazard narrative explaining the risk.
remedy_description	object	long text	Free-text column containing longer text values. Cleaned CPSC remedy narrative describing what consumers should do.
consumer_action	object	long text	Free-text column containing longer text values. Consumer action text supplied by the source export when present.
original_product_safety_warning_announcement	object	long text	Free-text column containing longer text values. Reference text for the original product safety warning announcement when present.
remedy_type	object	string	Text column. Raw remedy type text from the CPSC export.
remedy_type_count	int64	integer	Whole-number numeric column. Count of distinct remedy actions listed in the remedy type field.
remedy_refund	bool	boolean	True/false column. Boolean flag indicating that the recall offers a refund remedy.
remedy_repair	bool	boolean	True/false column. Boolean flag indicating that the recall offers a repair remedy.
remedy_replace	bool	boolean	True/false column. Boolean flag indicating that the recall offers a replacement remedy.
remedy_dispose	bool	boolean	True/false column. Boolean flag indicating that the recall instructs consumers to dispose of the product.
remedy_inspect	bool	boolean	True/false column. Boolean flag indicating that the recall includes an inspection step.
remedy_new_instructions	bool	boolean	True/false column. Boolean flag indicating that the recall issues new instructions rather than only a physical remedy.
remedy_no_remedy_available	bool	boolean	True/false column. Boolean flag indicating that the recall states no remedy is currently available.
units_text	object	string	Text column. Raw units field from the recall export.
units_affected_value	float64	float	Decimal numeric column. Parsed numeric lower-bound estimate of affected units when a quantity can be extracted.
units_affected_bucket	object	string	Text column. Bucketed estimate of affected units for easier filtering and segmentation.
incidents_text	object	long text	Free-text column containing longer text values. Raw incidents and injuries narrative from the source export.
no_incidents_reported	bool	boolean	True/false column. Boolean flag indicating that the incidents field explicitly states none were reported.
reported_incident_count_lower_bound	float64	float	Decimal numeric column. Parsed lower-bound count of reported incidents, complaints, or reports when stated in the text.
reported_injury_count_lower_bound	float64	float	Decimal numeric column. Parsed lower-bound count of injuries when stated in the text.
reported_death_count_lower_bound	float64	float	Decimal numeric column. Parsed lower-bound count of deaths or fatalities when stated in the text.
sold_at_label	object	string	Text column. Raw sales-channel label from the CPSC export, such as Sold At or Sold Online At.
sold_at	object	long text	Free-text column containing longer text values. Raw retailer and sale-period narrative from the CPSC export.
sale_start_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Parsed start date of the sale period when the sold-at narrative contains a recognizable range.
sale_end_date	datetime64[ns, UTC]	datetime	Date or timestamp column. Parsed end date of the sale period when the sold-at narrative contains a recognizable range.
sale_duration_days	float64	float	Decimal numeric column. Number of days between parsed sale start and sale end dates when both are available.
price_low_usd	float64	float	Decimal numeric column. Parsed low-end U.S. dollar price from the sold-at narrative when available.
price_high_usd	float64	float	Decimal numeric column. Parsed high-end U.S. dollar price from the sold-at narrative when available.
importers	object	string	Text column. Importer text supplied by the CPSC announcement when present.
manufacturers	object	long text	Free-text column containing longer text values. Manufacturer text supplied by the CPSC announcement when present.
distributors	object	string	Text column. Distributor text supplied by the CPSC announcement when present.
responsible_firm	object	string	Text column. Best-effort primary firm name selected from manufacturer, importer, or distributor fields.
responsible_firm_role	object	string	Text column. Role of the selected responsible firm, such as manufacturer, importer, or distributor.
manufactured_in	object	string	Text column. Raw country or geography text for where the product was manufactured.
manufactured_in_normalized	object	string	Text column. Normalized manufacturing-country label for easier grouping and joins.
imported_product_flag	bool	boolean	True/false column. Boolean flag indicating that the product appears to be imported based on country or importer signals.
child_product_flag	bool	boolean	True/false column. Boolean flag indicating that the recall text suggests a child or infant-oriented product.
mandatory_standard_violation_flag	bool	boolean	True/false column. Boolean flag indicating that the announcement mentions violation of a mandatory standard or required warnings.
serious_injury_or_death_flag	bool	boolean	True/false column. Boolean flag indicating that the announcement explicitly references serious injury or death risk.
hazard_fire	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions fire, overheating, or burning hazards.
hazard_drowning	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions drowning or submersion hazards.
hazard_fall	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions fall hazards.
hazard_tip_over	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions tip-over or toppling hazards.
hazard_battery_ingestion	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions button-cell, coin-battery, or ingestion hazards.
hazard_choking	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions choking hazards.
hazard_suffocation	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions suffocation, asphyxiation, or strangulation hazards.
hazard_entrapment	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions entrapment, trapping, or pinching hazards.
hazard_magnet	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions magnets or magnetic components.
hazard_electrocution	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions electric shock or electrocution hazards.
hazard_laceration	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions cuts or laceration hazards.
hazard_amputation	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions amputation hazards.
hazard_crash	bool	boolean	True/false column. Boolean flag indicating that the recall text mentions crash or collision hazards.
hazard_keyword_count	int64	integer	Whole-number numeric column. Count of hazard-category keyword flags triggered for the row.
primary_hazard_category	object	string	Text column. Best-effort primary hazard category selected from the triggered hazard flags.
sold_online_flag	bool	boolean	True/false column. Boolean flag indicating that the sales narrative suggests online distribution.
sold_on_amazon	bool	boolean	True/false column. Boolean flag indicating that Amazon is mentioned in the sales narrative.
sold_on_walmart	bool	boolean	True/false column. Boolean flag indicating that Walmart is mentioned in the sales narrative.
sold_on_target	bool	boolean	True/false column. Boolean flag indicating that Target is mentioned in the sales narrative.
sold_on_qvc	bool	boolean	True/false column. Boolean flag indicating that QVC is mentioned in the sales narrative.
sold_on_costco	bool	boolean	True/false column. Boolean flag indicating that Costco is mentioned in the sales narrative.
sold_on_ebay	bool	boolean	True/false column. Boolean flag indicating that eBay is mentioned in the sales narrative.
sold_on_wayfair	bool	boolean	True/false column. Boolean flag indicating that Wayfair is mentioned in the sales narrative.
sold_on_shein	bool	boolean	True/false column. Boolean flag indicating that SHEIN is mentioned in the sales narrative.
source_csv_url	object	string	Text column. Official CPSC CSV download URL used to build the dataset.
source_api_url	object	string	Text column. Row-level SaferProducts API URL that reproduces the announcement record by recall number.
source_page_url	object	string	Text column. Top-level CPSC recalls page associated with the source dataset.
source_domain	object	string	Text column. Domain name of the source system used for the row.
last_collected_at	datetime64[us, UTC]	datetime	Date or timestamp column. UTC timestamp when the dataset build collected the source data.

Intended Use

This dataset is intended for research, experimentation, analysis, and model prototyping.

Loading the Dataset

import os
from datasets import load_dataset

HUGGINGFACE_API_KEY_KARMANE = os.environ.get("HUGGINGFACE_API_KEY_KARMANE")

dataset = load_dataset(
    "Karmane/cpsc-consumer-product-recalls-enriched",
    token=HUGGINGFACE_API_KEY_KARMANE,
)

print(dataset)
print(dataset[list(dataset.keys())[0]][0])

# getting the DataFrame itself
# df = dataset[list(dataset.keys())[0]].to_pandas()

---

Marker, Mason. (2025). CPSC Consumer Product Recalls Enriched Dataset. Hugging Face. https://huggingface.co/datasets/Karmane/cpsc-consumer-product-recalls-enriched