An intelligent price forecasting solution for farmers and policymakers.
Every year, over 10,000 farmers in India die by suicide, often due to unpredictable price drops, weather shocks, or disasters. AgriPredict aims to empower farmers with foresight — helping them make better financial decisions using AI-driven price forecasting.
Agriculture is one of the most critical yet riskiest occupations globally. Most farmer losses stem from a lack of reliable, region-specific price predictions. This project tackles that problem by providing hyper-local, data-driven forecasts tailored for different regions across India.
- 🔄 Customized Forecasting for 100 Regional Centers: India’s vastness means price trends vary greatly. AgriPredict trains separate models for 100 simulated centers to reflect realistic, region-specific market behavior.
- 📉 SARIMA for Seasonal Trends: SARIMA (Seasonal AutoRegressive Integrated Moving Average) models help capture repeating seasonal patterns in agricultural prices and also detect sudden drops caused by shocks.
- 🔺 Boosted Accuracy with Gradient Boosting: XGBoost is layered on top of SARIMA predictions to refine and improve forecasting accuracy.
- 📊 Core Prediction System Implemented: Built using Flask, this web app allows users to choose a commodity and center, then view its upcoming price trends via dynamic graphs.
⚠️ Future Expansion Plans: Weather and disaster forecasts are planned to be integrated soon to enhance prediction accuracy.
The dataset is artificially generated but mimics real-world agricultural price behavior — including seasonal patterns and abrupt fluctuations caused by simulated disasters.
- Python
- Flask
- Pandas, NumPy
- SARIMAX (statsmodels)
- XGBoost
- Matplotlib
- HTML/CSS (for frontend)
