Primary competition visual

Prédiction IA des rendements agricoles par télédétection et séries temporelles by Chad Institute of Algorithms

Helping Chad
500 Zindi Points
Under code review
GIS
Machine Learning
37 joined
27 active
Starti
12 Aug 26
Closei
15 Aug 26
Reveali
20 Aug 26
About

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
Files
Description
Files
This is the data required for training your model.
This is the testing dataset for evaluating your model.
This file shows the format of your submission file.