Automobile-Manufacturers Data Guide Data: Dataset Structure and Field Coverage

Europe anchors the category. ACEA alone contributes two quality-score-9 sources - the ACEA European Automobile Statistics Hub and ACEA Passenger Car Registrations by Fuel Type and Manufacturer - whose monthly PDF releases cover roughly 30 markets across BEV, PHEV, hybrid, petrol and diesel powertrains with manufacturer market shares. The OICA World Motor Vehicle Production Statistics supply the worldwide counterweight: country-level annual unit counts running from 1999 to 2026.

Eighteen primary records is a small enough universe to learn, but the sources are not interchangeable. Grouped by publisher type, here is the field with the quality score Datadory assigns each record.

Trade-association series (3 datasets). The ACEA European Automobile Statistics Hub scores 9 and is free without login: monthly passenger-car and quarterly commercial-vehicle registration releases plus figure pages on production, trade, taxation, fleet and electrification for EU27, EFTA and the UK. Its companion, ACEA Passenger Car Registrations by Fuel Type and Manufacturer - also a 9 - packs about 30 markets x 7 fuel columns plus roughly 20 manufacturer groups and 40+ brands into six-page monthly PDFs, twelve releases a year, with cumulative year-to-date columns and archives online back through at least the 2010s. OICA World Motor Vehicle Production Statistics (quality 8) adds roughly 40+ countries/regions per year across ~27 years of annual tables plus partial-year quarterly drops such as 2026 Q1, each year a small free spreadsheet.

Government statistics and APIs (10 datasets). NHTSA's vPIC API leads at 10. Eurostat STS_INPR_M (9) has published a monthly series since 17 March 2010, served through June 2026 in JSON, TSV and SDMX. The EPA Fuel Economy Dataset (1984-2026) (9) combines 43 model years in one CSV, with per-year ZIP archives from 1978 including a 2027 preliminary file. The Washington State Electric Vehicle Population Data (9) refreshes 294,193 BEV/PHEV registration records weekly. The EPA FuelEconomy.gov Web Services API (8) mirrors that universe over keyless REST calls returning JSON or XML. The Bureau of Transportation Statistics Data Inventory (8) hosts 155 public-domain datasets on Socrata. Japan's e-Stat (6), data.gov (6, searching a 552,271-dataset federal catalog), the EU Open Data Portal (5, 4,108 hits for 'car production sales') and India's OGD platform (4) complete the government tier.

Machine-learning benchmarks (3). The Stanford Cars Dataset on HuggingFace (7) ships 72,472 parquet rows across 196 make-model-year classes; the UCI Automobile Dataset (6) holds 205 rows x 26 attributes from the 1985 Ward's Automotive Yearbook under commercial delivery terms 4.0; and a Kaggle mirror of the same snapshot (6) has accumulated 229 community notebooks.

Track European demand and brand share month by month. The two ACEA releases report new passenger-car registrations across EU27, EFTA and the UK by country, power source and roughly 20 manufacturer groups with 40+ brands, including cumulative year-to-date figures.

Benchmark national production footprints. The OICA World Motor Vehicle Production Statistics give annual units by country and vehicle type from 1999 through 2026, while Eurostat's STS_INPR_M index turns monthly NACE C29 output into seasonally adjusted cycle indicators.

Build fuel-economy and emissions analyses. The EPA Fuel Economy Dataset (1984-2026) and the EPA FuelEconomy.gov Web Services API supply MPG, CO2, engine and transmission specifications for every US configuration, as CSV bulk files or keyless JSON/XML calls.

Power VIN decoding and entity resolution. The NHTSA vPIC API maps WMI codes and decodes VINs into up to 154 fields with no integration key, supporting fitment lookups, insurance quoting and lead enrichment.

Measure EV adoption at neighborhood level. The Washington State Electric Vehicle Population Data publishes 294,193 registered BEV/PHEV records refreshed weekly, geocoded from county down to census tract under commercial delivery terms.

Train computer-vision make-model recognition. The Stanford Cars Dataset (HuggingFace) provides 196 make-model-year classes across 49+ manufacturers in 72,472 parquet rows, and the commercial delivery terms 4.0 UCI Automobile Dataset adds 205 labeled rows for tabular benchmarks.

Compare sales rankings across markets. GoodCarBadCar Automotive Sales Data tracks automaker, brand, model and segment volumes for the US from 2006 to 2026 plus Canada, Europe, China and 180+ Rest of World country pages.

Contextualize vehicle flows and logistics exposure. The Bureau of Transportation Statistics Data Inventory contributes 155 public-domain datasets spanning freight movement, truck travel times, vehicle miles traveled and border crossings.

Data scientists arrive searching for a "vin decoding api without integration key" or a "car image classification dataset benchmark". Their stack is already here: vPIC resolves makes, models and WMI codes into up to 154 fields per VIN for entity resolution, Stanford Cars supplies 8,144 training and 8,041 test images (plus seven corrupted test variants) for fine-grained vision models, and the UCI Automobile Dataset - 205 rows, 26 attributes - remains the classic tabular teaching set, with 229 community notebooks on its Kaggle mirror.

Market researchers search "european car registration statistics monthly" and "monthly car sales by model". ACEA's monthly releases hand them brand-share tables across EU27, EFTA and the UK; GoodCarBadCar extends the picture to US model-level rankings back to 2006 and 180+ other markets; OICA's 1999-2026 production tables set the supply-side baseline.

Aftermarket catalog and parts teams hunt "aces pies fitment data subscription": the Auto Care Association's VCdb covers 175.5K+ vehicle applications and 2.4M+ configurations across North and South America plus nine off-highway segments.

Policy and logistics analysts query "electric vehicle population data washington" or pull from BTS's 155 public-domain datasets on truck travel times and border crossings, while regional planners watch the weekly-refreshed EV registry geocode adoption down to census tract.

1. Anchor identity on vPIC. Start every pipeline with NHTSA's keyless API: DecodeVinValues normalizes VINs into up to 154 fields, GetAllManufacturers pages 100 manufacturer records at a time, and responses come back as JSON, XML, CSV or ZIP. NHTSA applies an automated traffic-rate control mechanism, so cache responses rather than re-decoding.

1. Watch electrification locally. Schedule a weekly pull of the Washington State EV registry as CSV or GeoJSON - it is the only record in the slice refreshed that often at census-tract granularity.

1. Benchmark against the classics. Keep Stanford Cars (6.02 GB across nine parquet splits) and UCI (an 8.3 KB ZIP) handy as regression baselines for new make-model models.

Nine figures from the catalog capture this slice at a glance, as of August 2026.

The eight questions below are the ones these records answer most directly - from where monthly European registration tables live to whether Eurostat's production index can be reused commercially.

Three adjacent records round out the pool from neighboring industries: Eurostat PRODCOM production statistics for building product industries, the USDOT Open Data Portal's national transportation datasets, and the Wikidata SPARQL Query Service.