amanpandey8120/mandi-prize

# 🌾 AgroAid Mandi Price Prediction API ## Overview The AgroAid Mandi Price Prediction API is a Machine Learning-powered service that predicts agricultural commodity prices using historical mandi market data. The API is designed to integrate with the AgroAid platform and provide: * Real-time price prediction * 7-day price forecasting * Market trend analysis * Confidence scoring * Sell/Hold recommendations The model is trained on historical APMC mandi data and deployed using FastAPI and Render. --- # Project Architecture ``` AgroAid App │ ▼ Supabase Database │ ▼ Feature Generation │ ▼ Mandi Prediction API │ ▼ LightGBM Model │ ▼ Prediction Results ``` --- # Technology Stack ### Backend * Python * FastAPI * Pandas * NumPy ### Machine Learning * LightGBM * Scikit-Learn * Joblib ### Database * Supabase PostgreSQL ### Deployment * Render --- # Project Structure ``` mandi-price-api/ │ ├── main.py ├── requirements.txt ├── district_mapping.json ├── commodity_mapping.json ├── README.md │ ├── mandi_price_model.pkl │ └── training/ ├── train_model.ipynb ├── clean_apmc_data.csv └── feature_engineering.py ``` --- # File Descriptions ## main.py Main FastAPI application. Responsibilities: * Load trained model * Accept API requests * Generate predictions * Generate forecasts * Calculate trends * Calculate confidence scores * Generate recommendations Available endpoints: * GET / * POST /predict * POST /forecast * POST /trend * POST /confidence * POST /recommendation --- ## mandi_price_model.pkl Serialized LightGBM model. Generated after training. Used for: * Single day prediction * Forecast generation --- ## district_mapping.json District encoder mapping. Example: ```json { "Rewa": 47, "Bhopal": 12 } ``` Purpose: Convert district names into numerical values used by the model. --- ## commodity_mapping.json Commodity encoder mapping. Example: ```json { "Wheat": 58, "Soybean": 125 } ``` Purpose: Convert commodity names into numerical values …