Caleb403/maize-yield-prediction

Maize Yield Prediction — Uasin Gishu County, Kenya — Machine Learning Models with LOOCV RMSE 0.364 t/ha # Maize Yield Prediction — Uasin Gishu County, Kenya **IBM SkillsBuild Data Analytics Bootcamp** ## Live Dashboard Open the interactive dashboard ## Project Summary This project predicts annual maize yield (t/ha) in Uasin Gishu County, Kenya using seasonal weather, soil, and seed management data from 2012–2023. ## Key Findings - Rainfall timing matters more than total volume - Pre-planting conditions (Sep–Nov) shape the following harvest - Soil acidity (pH 5.7) explains the structural yield gap of ~6.5 t/ha ## Model Performance - Algorithm: Random Forest + SVR - Validation: Leave-One-Out Cross-Validation (n=12) - LOO RMSE: 0.364 t/ha (10% of county average) - LOO R²: 0.304 ## Data Sources - Ministry of Agriculture & Livestock Development, Kenya - NASA POWER Climate Data - Lomurut (2014), Purdue University - CIMMYT / Tegemeo Institute