Your objective is to predict the agricultural yield (Rendement_kg_ha, in kg/ha) for each province, agricultural season, and crop using historical production, satellite vegetation, and climate data.
The data covers:
- 19 provinces of Chad (Sahelian and Sudanian zones, excluding non-agricultural desert provinces)
- 10 agricultural seasons (June–May), from 2016/2017 to 2025/2026
- 14 crops: Millet, Sorghum, Maize, Rice, Berbere, Wheat, Fonio (cereals); Groundnut, Sesame, Cowpea, BambaraGroundnut, Cassava, SweetPotato, Taro (other crops)
Prohibited external data list:
- FAO-ASIS / MVHI (Weighted Vegetation Health Index, Chad) -- our satellite feature; if available in the open, it would allow reconstructing wVHI_moyen_saison / wVHI_min_saison directly
- NASA POWER (climate data, 12N/15E point) -- exact source of Temp_moy_C, Precip_totale_mm, Humidite_rel_moy_pct
- FAOSTAT (Chad, production/yield/area series) -- allows reconstructing national totals
- Chad's official agricultural statistics (DPSA/ANADER/SODELAC reports, "Final results of the agricultural campaign") -- this is our raw data source, the most critical one to explicitly prohibit