datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
nhtsa-vehicle-complaints
NHTSA Vehicle Complaints 1995–March 2026
One-time dated snapshot.
Vehicle complaint records with narratives and component fields.
Verified coverage: Added dates: 1995-01-03 through 2026-03-26.
Records: 2,189,824 component rows; 1,582,110 complaints.
This repository contains a 1,000-row public sample, not the full package. A sample does not establish complete historical coverage.
Limitations
Failure dates include implausible values through 2203 and model years… See the full description on the dataset page: https://huggingface.co/datasets/claritystorm/nhtsa-vehicle-complaints.nhtsa-vehicle-recalls
NHTSA Vehicle Recalls 1966–March 2026
One-time dated snapshot.
Recall campaign and affected-product records with defect and remedy text.
Verified coverage: Recall dates: 1966-01-19 through 2026-03-26.
Records: 176,073 product rows; 29,865 distinct campaign IDs.
This repository contains a 1,000-row public sample, not the full package. A sample does not establish complete historical coverage.
Limitations
Multiple product/model rows can belong to the same campaign.… See the full description on the dataset page: https://huggingface.co/datasets/claritystorm/nhtsa-vehicle-recalls.nhtsa-fars-1975-2023
NHTSA FARS Fatal Crashes 1975–2023
One-time dated snapshot.
Fatal crash, vehicle and person records in CSV and Parquet.
Verified coverage: 1975–2023; all 49 years present in each table.
Records: 1,836,052 crashes; 8,344,160 rows across three tables.
This repository contains a 1,000-row public sample, not the full package. A sample does not establish complete historical coverage.
Limitations
Crash, vehicle and person rows represent different units; do not add them… See the full description on the dataset page: https://huggingface.co/datasets/claritystorm/nhtsa-fars-1975-2023.NHTSAnhtsa-investigations
NHTSA Defect Investigations
Every NHTSA Office of Defects Investigation (ODI) action touching a tracked vehicle — open and historical. The leading indicator of a recall: a Preliminary Evaluation (PE) can escalate to an Engineering Analysis (EA) and then a recall. Includes the recall campaign number for investigations that became recalls.
Columns
column
meaning
action_number
NHTSA investigation ID (prefix = type)
type
PE | EA | RQ | DP | AQ —… See the full description on the dataset page: https://huggingface.co/datasets/ProblemsByVin/nhtsa-investigations.