English

Harmonised public data on France and EU regions

Prisme brings together official statistics from Insee, Eurostat, the French tax administration (DGFiP), the World Bank and other public producers, expresses them in a single geography and publishes them as one open data cube, with the reason for every missing value. The site itself is in French; the data, the SQL console and the code work in any language.

Download the data cube Open the SQL console

What is in the cube

Indicators on population, income and poverty, housing and property prices, employment, health and life expectancy, education and the economy, for French communes, départements and régions, EU NUTS 2 regions and countries. Every French figure is expressed in a single reference year of the official geographic code (communes merge or split every year), and every European figure in the NUTS 2024 classification.

• One file per geographic level, in long format: one row per indicator, territory and year. Parquet for every level, CSV as well except for communes.
• A data dictionary (CSV and JSON) gives each indicator's unit, source, licence and required attribution.
• A missing value is never a zero. Rows without a value carry a statut explaining why.

Columns

indicateur_id

Indicator identifier (see the dictionary).

geo_niveau

Geographic level: commune, arrondissement_municipal (Paris, Lyon, Marseille), departement, region, nuts2, pays (country), zone (aggregates such as the EU).

geo_code

Territory code: Insee code in the reference geography, NUTS code or country code.

territoire

Territory name.

code_departement, code_region, code_pays

Parent département, région (France) and country (Eurostat code for NUTS regions).

annee

Year of the data, which is not the year of the geography.

valeur, unite

Value and unit. The value is empty when it is not published.

statut

ok when published; otherwise masque (suppressed by Eurostat), secret (statistical confidentiality), ventes_insuffisantes (too few property sales to publish a price), hors_champ (territory outside the source's scope), couverture_partielle (source covers only part of the territory) or absent (not published).

drapeau

Eurostat flag: p provisional, e estimated, b break in series, u low reliability, d definition differs.

source_id

Source, described in the source catalogue.

Query it without installing anything

The SQL console runs DuckDB in your browser and reads the Parquet files directly, fetching only the parts a query needs: nothing is sent to a server. The same files can be read from DuckDB, Python (pandas, polars) or R, locally or straight from their URL, for example select * from 'https://www.prisme-donnees.fr/donnees/cube-departement.parquet' limit 10.

Browse by place or by indicator

Every French département and région and every EU NUTS 2 region has its own page: latest values, rank among its peers, trends and what the sources do not publish. Every indicator has a page too, with its definition, licence and full ranking. The comparison tool puts two territories side by side, indicator by indicator, and its address can be shared. These pages are in French, but their tables and figures read the same in any language.

Rules the data follows

• No individual property sale is ever published. Prices from the French property transactions file (DVF) are medians over a minimum number of sales; below that threshold, the cell says ventes_insuffisantes.
• Rates and medians are never added up or split. When communes merge or split, counts are reallocated with weights; a rate or a median only crosses a one-to-one link (an unchanged or recoded commune).
• Each value keeps its producer's licence, listed in the dictionary: mostly the French Licence Ouverte 2.0 and CC BY 4.0. EU region boundaries from GISCO are not part of the cube: they are restricted to non-commercial use.

The analyses

The analyses (in French) use the same data: every figure is computed at build time from the published cube, comes with its confidence interval and its limits, and a claim that the data stop supporting fails the build. Topics include life expectancy across French départements, the gap between GDP and household income in EU regions, falling property prices, vacant housing, fertility in Europe and the ageing of EU regions. The method page describes the statistics used.

Frequently asked questions

Is the Prisme data free to reuse?

Yes, under each producer's licence, listed for every indicator in the data dictionary: mostly Licence Ouverte 2.0 (Etalab) and CC BY 4.0, which allow commercial reuse with attribution.

Why is a value missing for some territories?

Because the producer did not publish it (statistical confidentiality, low reliability, territory outside the source's scope) or because too few property sales took place. The statut column gives the reason; a missing value is never replaced by zero.

Which geography are French communes in?

All French data are expressed in one reference year of the official geographic code (Code officiel géographique), so that series from different years and sources can be joined.