Can Machine Learning Forecast Rice Yields in Data-Constrained Settings? Satellite Climate Data, National Crop Statistics, and Lessons from Sierra Leone

In three linesML study on rice yield forecasting in data-constrained Sierra Leone. XGBoost + satellite climate data (CHIRPS, NASA POWER) reduces error by 34% (RMSE 284 vs 428 kg/ha). May-June rainfall is dominant predictor. Open-source pipeline provided.

Summary generated by Claude — human-verified