Datadory notebook
Steel Data Guide Data: Dataset Structure and Field Coverage
Datadory delivers steel data guide data covering comprehensive field definitions, entity mappings, and historical time series — structured for direct analytics and delivered on demand.
1,744 datasets. Pick your catch.
What does the steel data landscape look like in 2026?
Steel data splits into four usable families: official production and trade statistics, price assessments, government import monitors, and machine-learning benchmarks. Datadory catalogs 25 datasets for the industry — 23 primary sources plus 2 related records from adjacent coal and coke coverage (the Metallurgical Coal Exit List and the SteelHome Database for coke, coal and steel raw materials). No single publisher covers the cycle end to end, so most working stacks blend at least two families.
Trade associations fill the gaps the statistical agencies leave. AISI Industry Data publishes weekly US raw steel production and capability utilization; worldsteel's Monthly crude steel production (71 countries) release covers roughly 70 countries representing about 98% of world output every month; and EUROFER European Steel data, JISF Japan steel statistics and WV Stahl German steel statistics give the EU, Japan and Germany recurring national views.
Which core datasets should every analyst know?
Government and international statistics. The three USGS National Minerals Information Center pages are the cheapest high-quality foundation anywhere in the slice. Iron and Steel Statistics and Information carries three Data Series 140 sheets spanning 120-plus years of annual observations each, about 30 Mineral Commodity Summaries editions and roughly 30 Minerals Yearbook chapters, all US Government works in the commercial delivery terms. Iron Ore Statistics and Information is the only slice source covering mine production, reserves and direct reduced iron for roughly 40 producing countries annually — its world table notes that Michigan, Minnesota and Utah account for essentially all usable US ore output. UNSD Industrial Commodity Statistics (steel) adds annual crude-steel series in both physical quantity and monetary value (about 2,300 records for the crude steel line alone), with 1995–2016 online and the 1950–2003 historical series available on request.
What do teams actually use steel data for?
Track global monthly crude steel output. worldsteel's Monthly crude steel production (71 countries) release covers about 98% of world production across roughly 70 reporting countries, and UNSD Industrial Commodity Statistics (steel) adds annual national series in physical quantity and value for market sizing and cycle timing. Together they answer the most-searched question in the niche — crude steel production by country — without a terminal subscription.
Research decarbonization and plant-level energy intensity. IEA Iron & Steel sector data and analysis scenario charts project the sector's energy demand and CO2 pathway to 2050, and the commercial delivery terms 4.0 Steel Industry Energy Consumption set — 35,040 fifteen-minute readings from one South Korean plant across 2018 — feeds emissions baselines and load-forecasting models.
Train computer-vision quality inspection models. NEU Metal Surface Defects Database, Severstal Steel Defect Detection and Steel Plates Faults (UCI) supply labeled strip imagery, pixel masks and tabular fault features respectively, so classification, segmentation and structured-fault work each have a purpose-built benchmark.
Size and compare national steel markets. The Steel Statistical Yearbook, EUROFER European Steel data interactive tools and Iron Ore Statistics and Information country tables supply production, trade, reserves and apparent-consumption inputs economy by economy.
Follow national sectors month by month. India's Ministry of Steel Monthly Summary PDF pairs with Steel Import Management System (SIMS) data transaction-level imports; Japan's JISF Japan steel statistics workbooks break production out by furnace type and orders by consuming sector; Germany gets monthly crude steel figures and the annual factbook from WV Stahl German steel statistics; and EUROFER European Steel data covers the EU aggregate.
Build backtestable industrial-cycle indicators and country profiles. Eurostat NACE C24 short-term statistics from January 1953 and JISF Japan steel statistics order-book series provide long monthly histories for quant and macro research, and WV Stahl German steel statistics factbooks pair with Eurostat structural business statistics for consulting-grade employment and turnover views.
Who uses steel data?
Data scientists and ML engineers search phrasings like 'steel surface defect detection dataset' and 'steel plant energy consumption dataset'. They land on Severstal Steel Defect Detection for 12,568 masked sheet images, Steel Plates Faults (UCI) for training-permissive tabular faults and Steel Industry Energy Consumption for 15-minute load series. Catalog-wide, 666 of 1,744 datasets reach top-tier relevance for this profile and 620 of those are completely free — the steel slice's three vision benchmarks and one energy set all fit the free, downloadable pattern. See the dedicated page on the data scientists profile and its ranked picks for the best data for steel teams.
Investors and quant researchers hunt backtestable industrial-cycle indicators: Eurostat's index from January 1953, JISF order-book workbooks and AISI utilization rates form a monthly-to-weekly cycle dashboard. Journalists and academics gravitate to the citable public-domain tier — USGS yearbooks, SIMA monitors and UN statistics — where provenance is institutional and redistribution is unrestricted.
What do the numbers say about steel data?
The table below condenses the slice's measurable properties, with catalog-wide context from the full 1,744-dataset catalog computed on 2026-08-22.
Frequently asked questions about steel data
Eight questions account for most steel-data searches; each answer below names the specific source and the exact access terms, so nothing needs a follow-up click.
Keep reading
These pages continue the thread:
| Number | What it measures |
|---|---|
| 25 | Datasets cataloged for steel: 23 primary sources plus 2 related records from coal/coke coverage |
| 7.52 of 10 | Mean Datadory quality score across the 23 primary records (catalog-wide average: 7.81 across 1,744 datasets) |
| 11 of 23 | Primary datasets scoring 8 or higher on the 10-point quality rubric |
| 1900 and January 1953 | Start of the longest free annual series (USGS Data Series 140) and of Eurostat's monthly NACE C24 production index |
| ~98% | Share of world crude steel output covered by worldsteel's 71-country monthly release |
| 82.4% | Share of all 1,744 cataloged datasets that are free to access, for comparison with steel's 74% |
| If you need to... | Start with | Why |
|---|---|---|
| Track global output monthly | Monthly crude steel production (71 countries) | ~70 countries, ~98% of world output, free press releases 3–4 weeks after month end |
| Train a defect-inspection model | Severstal Steel Defect Detection | 12,568 labeled 256x1600 px sheet images with RLE masks in four classes |
| Forecast plant electricity load | Steel Industry Energy Consumption | 35,040 fifteen-minute readings across calendar year 2018, commercial delivery terms 4.0 |
Pick up where this leaves off
Every one of these ships with sample rows before you commit to anything.
Iron Ore Statistics and Information Data
Iron and Steel Statistics and Information Data
Iron and Steel Scrap Statistics and Information Data
Steel Import Monitor (SIMA) Data
EU steel & basic-metals production statistics (dissemination API)
UNSD Industrial Commodity Statistics (Steel) Data
Want rows instead of a pitch? Name the datasets.
API, files, or your warehouse. Daily, weekly, or hourly.
Get a sampleQuestions worth asking
Where can I get weekly US steel production data?
'AISI Industry Data' publishes weekly US raw steel production and capability utilization by district free on steel.org, with monthly shipment and SIMA import releases alongside. The detailed AIS 7 monthly report and the Annual Statistical Report are sold separately through S&P Global's Steel Data and Analysis product.
What datasets are available for steel surface defect detection?
Three benchmarks cover the task: 'NEU Metal Surface Defects Database' offers 1,800 grayscale images across six classes, 'Severstal Steel Defect Detection' has 12,568 labeled sheet images with run-length-encoded masks, and 'Steel Plates Faults (UCI)' gives 1,941 tabular samples with 27 features and seven fault labels.
Where can I compare crude steel production across countries?
worldsteel's 'Monthly crude steel production (71 countries)' release covers roughly 70 countries representing about 98% of world output as free press releases. 'UNSD Industrial Commodity Statistics (steel)' adds annual crude steel series for about 103 countries, while the subscription-only 'Steel Statistical Yearbook' extends to roughly 90 countries.
Where can I find long historical steel production statistics?
'Iron and Steel Statistics and Information' from USGS includes Data Series 140 annual series running from 1900, with Mineral Commodity Summaries editions back to 1996. Eurostat's sts_inpr_m production index extends from January 1953 to the current month, and UN historical commodity series reach back to 1950 on request.