SamuelDasaolu/nigeria-crop-yield-predictor

Machine Learning application that predicts crop yields in Nigeria based on historical climate and soil data. Built with XGBoost & Streamlit. # 🇳🇬 Nigeria Crop Yield Prediction System **Live Demo:** Click here to test the app ## 📖 Project Overview This project is a machine learning solution designed to support Precision Agriculture in Nigeria. It predicts the expected crop yield (kg/hectare) for major Nigerian crops based on environmental factors (rainfall, temperature) and farm size. Unlike standard datasets, the data for this project was **custom-engineered** by merging agricultural output data from the **FAO (Food and Agriculture Organization)** with historical climate data from the **World Bank/NIMET**, covering the years 1990–2024. ## 🚀 Key Features * **Custom Data Pipeline:** Scripts to scrape, clean, pivot, and merge disparate data sources into a usable ML dataset. * **High-Performance Model:** Trained an **XGBoost Regressor** achieving an **R² Score of ~0.98** on test data. * **Interactive Dashboard:** A Streamlit frontend allowing farmers and policymakers to simulate climate scenarios. * **Robust Preprocessing:** Automated handling of categorical crop data using Scikit-Learn Pipelines. ## 🛠️ Tech Stack * **Language:** Python * **Modeling:** XGBoost (Gradient Boosting), Scikit-Learn * **Data Engineering:** Pandas, NumPy * **Deployment:** Streamlit Cloud * **Serialization:** Joblib ## 📂 Project Structure ```text ├── data/ # Raw FAO data and processed CSVs ├── models/ # Serialized model (.pkl) artifacts ├── notebooks/ # Jupyter notebooks for EDA and experimentation ├── streamlit_app.py # The frontend application ├── train_model.py # Reproducible training pipeline script ├── requirements.txt # Project dependencies └── README.md # Documentation