Eurostat 🇪🇺 — European regional data, open and queryable

GDP, population, crime, health, education, climate — at NUTS2 (regional) and NUTS3 (provincial) level for every EU country.

Eurostat for DataCivicLab downloads, normalises and publishes Eurostat SDMX data as columnar parquet files on Google Cloud Storage. All data is open, free, and SQL-queryable.

What's inside

Published datasets	21 NUTS3 + 9 NUTS2 (30 total)
Period	1980 — 2025 (varies by dataset)
Coverage	All EU + EFTA + candidate countries
Format	Parquet on public GCS

By theme

Theme	Examples
💼 Economy	GDP, GVA, Employment, Productivity
👥 Demography	Population, Deaths, Births, Ageing
🚓 Crime	Recorded offences by ICCS
🏨 Tourism	Nights spent at accommodation
🚗 Transport	Road accidents
🌡️ Climate	Heating/cooling degree days
🌱 Environment	Soil erosion
🏢 Business	Enterprise births/deaths
🏥 Health	Physicians, Hospital beds
📊 Social	Poverty risk, Income inequality
📚 Education	Early school leaving, Tertiary attainment
🔬 Innovation	R&D expenditure

Full list: docs/dataset-registry.md

Note: all 28 datasets use the declarative SDMX pipeline (type: sdmx, agency ESTAT). The two historical script-based connectors (eurostat-bd-hgnace2-r3-nuts3, eurostat-pop-nuts3) were migrated to SDMX and now produce the same analytical marts as the rest of the catalog.

Questions you can answer

• Which Italian province has the highest GDP per capita? And the lowest?
• How does the crime rate compare between European regions?
• Which regions have the oldest population? And the youngest?
• Where is R&D spending growing fastest in Europe?
• How does the poverty risk differ between northern and southern Europe?

Three ways to access

1. Via MCP (toolkit)

All datasets are accessible via the DataCivicLab toolkit MCP server — the single MCP entry point for every Lab dataset. Connect it to your AI client:

-- GDP per capita of Italian provinces (2024)
SELECT geo_label_en, ROUND(value) AS gdp_pc
FROM eurostat_gdp_nuts3
WHERE geo LIKE 'IT%' AND unit='EUR_HAB' AND year=2024
ORDER BY value DESC;

Available tools: toolkit_find, toolkit_dataset_overview, toolkit_layer (SQL query; mart tables via table=mart_sintesi|mart_trend| mart_geo_benchmark), toolkit_registry_show (codelists).

2. Via DuckDB directly

import duckdb
duckdb.sql("""
    SELECT geo_label_en, ROUND(value) AS gdp_pc
    FROM read_parquet('gs://dataciviclab-clean/eurostat/eurostat_gdp_nuts3/*_clean.parquet')
    WHERE geo LIKE 'IT%' AND unit='EUR_HAB' AND year=2024
    ORDER BY value DESC
""").show()

3. Via download parquet

Public GCS buckets:

• Clean: gs://dataciviclab-clean/eurostat/{slug}/
• Mart: gs://dataciviclab-mart/eurostat/{slug}/

Discussions by theme

Theme	Discussion
💼 Economy	Labour productivity across the EU
📚 Education	Early school leaving
🏥 Health	Hospital bed capacity
🏢 Business	Enterprise births
🔬 R&D	R&D spending

Contribute

• Discussions — questions, findings, ideas
• Good first issues
• Contributing guide — how to add a dataset

Technical docs

• Dataset registry — published + planned
• License: MIT

Part of DataCivicLab.