vikkirkobane/kenya-crops-yield-prediction

# Kenya Crop Yield Prediction **AISIP Cohort 1 — Pathway 4: AI Engineering Capstone** **Africa AI Hub | Victor Chogo | May 2026** --- ## Problem Statement Kenya's 7 million+ smallholder farming households produce 75% of the country's food but have almost no access to data-driven yield forecasting tools. Without yield estimates, farmers cannot plan storage, negotiate fair prices, or make informed decisions about fertilizer investment. This project builds a machine learning model that predicts crop yield (kg/ha) before harvest, using climate, soil, and agronomic inputs available to any farmer. --- ## Live Demo > **Deployed app:** kenya-crops-yield-predictio… > **GitHub:** github.com > Run locally: `streamlit run app.py` --- ## Quick Start ```bash # Clone the repo git clone github.com cd kenya-crops-yield-prediction # Install dependencies pip install -r requirements.txt # Generate dataset python data/generate_data.py # Train models (creates models/ and plots/) python train_models.py # Launch the Streamlit app streamlit run app.py ``` --- ## Project Structure ``` kenya-crop-yield-prediction/ ├── data/ │ ├── generate_data.py # Synthetic dataset generator │ └── crop_yield_kenya.csv # Generated dataset (2,500 records) ├── models/ # Saved model artifacts (auto-created) │ ├── best_model.pkl │ ├── model_meta.json │ └── le_*.pkl # Label encoders ├── plots/ # Visualisations (auto-created) │ ├── model_comparison.png │ ├── actual_vs_predicted.png │ ├── feature_importance.png │ └── yield_distributions.png ├── app.py # Streamlit web application ├── train_models.py # Full training pipeline ├── model_card.md # Model documentation ├── requirements.txt └── README.md ``` --- ## Dataset | Field | Desc …