onkelzwerg/tanzania-agri-prices-data

Weekly wholesale food-crop prices for Tanzanian regions, extracted from Ministry of Agriculture bulletins. CC BY 4.0. Code: tanzania-agri-prices. # Tanzania Agricultural Prices — Data Weekly wholesale prices for seven food crops across Tanzanian mainland regions, extracted from the **Weekly Market Bulletins** published by the Tanzanian Ministry of Agriculture. This repository holds the growing, machine-readable dataset. The extraction code (PDF parser + web app) lives in the companion repository `tanzania-agri-prices`. ## Coverage | | | |---|---| | Weeks | 67 (2024-12-02 … 2026-05-25, plus one new bulletin per week) | | Regions | 26 mainland regions + "National Average" | | Crops | maize, rice, beans, sorghum, bulrush millet, finger millet, round potato | | Rows | ~10,300 (about 80 % carry a price; the rest were not reported) | Not every region reports every week, and some weeks were never published. Known gaps (no bulletin available): 2025-03-17, 2025-06-30, 2025-10-27, 2025-12-22, 2026-01-12, 2026-01-26, 2026-02-02, 2026-02-23, 2026-03-02, 2026-04-06, 2026-04-27. ## Layout & schema ``` data/ raw/ / .csv one file per bulletin week (long format, source of truth) processed/prices.csv all weeks concatenated (generated; what the web app fetches) provenance.csv source PDF, SHA-256 and import date per week schema/datapackage.json machine-readable schema (Frictionless-style) scripts/validate.ts validation, runs in CI on every push/PR scripts/build-processed.ts regenerates data/processed/prices.csv ``` `data/processed/prices.csv` is a generated artifact — do not edit it by hand. Regenerate it with `bun scripts/build-processed.ts` after changing any week file; CI fails if the committed copy is stale. The companion web app loads it over jsDelivr: `cdn.jsdelivr.net` Each week file is long-format CSV: ```csv week_start,week_end,region,crop,price 2026-05-11,2026-05-15,Arusha,maize,800 2026-05-11,2026-05-15,Arusha,bulrush_millet, ``` - **Unit:** TZS (Tanzanian Shilling) per kilogram …