LilPhantum/Crop-Yield-Prediction-System

Machine learning-based crop yield prediction system for West Africa using ensemble models Random Forest, Gradient Boosting, Ridge Regression. Trained on 12 African countries 2,776 samples, 1990-2013 with 20+ engineered features. Achieves 81% average R² across 8 crops. Includes web interface for interactive predictions and climate scenario analysis # ClimateCropSim: Crop Yield Prediction Under Climate Variability ClimateCropSim is a machine learning-based crop yield prediction system designed to estimate crop yield under projected climate conditions. The system uses rainfall, temperature, pesticide input, crop type, country selection, and land size to generate crop-specific yield predictions, visual charts, and AI-assisted recommendations. ## Overview This project supports crop yield simulation for major crops used in the system: Maize, Rice, Sorghum, Cassava, Wheat, Yam, Potatoes, and Soybeans. The prediction engine is built with trained machine learning models and deployed through a web interface that allows users to run simulations without interacting directly with the model code. The system focuses on selected African countries and was developed as a computational simulation system for predicting crop yield under climate variability. ## Key Features - Crop yield prediction under projected 2026 climate conditions - Crop-specific machine learning models - Support for multiple crop selection - Climate-based prediction using rainfall and temperature values - Feature engineering for climate stress and interaction variables - Total production calculation based on land size - Interactive climate and yield visualizations using ApexCharts - AI-based recommendation generated from completed prediction output - FastAPI backend for API-based prediction - Responsive frontend interface using HTML, CSS, and JavaScript ## Supported Crops The system currently supports the following crops: - Maize - Rice - Sorghum - Cassava - Wheat - Yam - Potatoes - Soybeans ## Supported Countries The deployed interface currently supports predictions for the following countries: - Cameroon - Ghana - Kenya - Mali - Niger - Senegal - Burkina Faso - South Africa - Egypt - Morocco ## Model Performance The table below shows the available model evaluation results. The coefficient of determination, represented as R², measures how wel …