Skip to content
RAG Repo

FAOSTAT is the statistical database of the Food and Agriculture Organization of the United Nations, maintained by the FAO Statistics Division. It is one of the largest public collections of food and agriculture data, covering over 245 countries and territories, with most time series running from 1961 to the present and some census material reaching back further. The domains span crop and livestock production, trade, food balance sheets, prices, land use and inputs, agrifood emissions, population and employment, investment, macroeconomic indicators, forestry, and food security and nutrition, including the cost and affordability of a healthy diet.

You access it through the web portal at fao.org/faostat, which lets you browse by domain, country and year and export selections, and through a bulk download service and a REST API for programmatic retrieval. Most data comes as CSV or JSON tables organised by domain codes, so it is a structured statistical database rather than a packaged machine-learning dataset.

For RAG and AI work the practical caveat is the same as for World Bank Open Data and Our World in Data. A language model reasons poorly over raw number tables, so the reliable pattern is to turn the series you need into short natural-language statements (for example a sentence stating a country, year, commodity and value) and embed those, or to keep the figures in a database the model can query, rather than dumping tables into a vector index. It grounds assistants for agriculture, food policy, trade and development research with authoritative, internationally comparable figures.

Since 1 July 2024 the statistical data is released under CC BY 4.0, so commercial use is permitted provided you credit FAO using the citation format it supplies. Two caveats are worth noting. Some third-party data embedded in the databases carries the original provider terms, and older FAO INFOODS publications are CC BY-NC-SA, so check the metadata for anything beyond the core statistics.

agriculturefoodstatisticstime-seriesglobaltradeopen-data

Related sources