AI-powered forecasting system for agriculture market stability and decision-making.
Agricultural commodity prices are highly volatile due to seasonal changes, market demand, and supply uncertainties. Traditional models often fail, leading to poor buffer stock management and ineffective government interventions.
This project leverages AI and time-series forecasting models (ARIMA, XGBoost, Random Forest) to build a robust price prediction system. By integrating real-time data from open government sources, it aims to:
- Enhance price forecasting accuracy
- Improve farmer decision-making
- Support government market interventions
- Reduce economic uncertainty in the agri-sector
- Develop an AI-driven forecasting model for agricultural & horticultural commodities.
- Integrate historical trends, seasonal patterns, and market signals.
- Provide data-driven insights for farmers, traders, and policymakers.
- Enable better buffer stock management and risk mitigation.
✅ Data Collection & Preprocessing
- Collects real-time commodity prices from Agmarknet and Data.gov.in.
- Cleans missing values, removes outliers, and normalizes datasets.
✅ Prediction Models
- ARIMA – Baseline time-series forecasting.
- XGBoost & Random Forest – ML models for improved accuracy.
- Deep Learning (Future Work) – Enhancing long-term trend predictions.
✅ Explainability
- Feature engineering to capture seasonality, inflation, and market shocks.
- Feature reduction with MRMR for optimal accuracy.
✅ Frontend (Prototype)
- Login system for users (farmers, traders, policymakers).
- Select crop & region → get future price predictions instantly.
Raw Data → Preprocessing → Model Building → Prediction → Frontend
Modules
- Data Preprocessing: Cleans & structures price datasets.
- Model Building: Trains ARIMA & ML models.
- Frontend: Simple UI to input crop & region → outputs predicted price.
- Python (Pandas, NumPy, Scikit-learn, Statsmodels, XGBoost)
- Data Sources: agmarknet.gov.in, data.gov.in
- Visualization: Matplotlib, Seaborn
- Frontend: Basic web login + selection form
- Deployment (Future): Streamlit / Taipy dashboard
- Mohanty, Thakurta, Kar (2023) – Agricultural Commodity Price Prediction Model: A Machine Learning Framework
- Zhang, Chen, Ling, Xia (2020) – Forecasting Agricultural Commodity Prices Using Model Selection Framework With Time Series Features and Forecast Horizons
- Elbasi, Mostafa, Alarnaout (2022) – AI in Agriculture: A Systematic Review
- Add deep learning models (LSTM, GRU) for sequence forecasting.
- Deploy as an interactive dashboard (Streamlit/Taipy).
- Add real-time alerts (SMS/Email/Slack) for sudden price spikes.
- Extend prediction to multi-region & multi-crop analysis.
🌱 Farmers – Choose the best time to sell crops. 📦 Traders – Optimize buying & storage strategies. 🏛️ Policymakers – Enable better intervention & stabilize markets.
- Sharanya T -727622BAD007
- Santhosh S -727622BAD073
- Aashif Shadin K N -727622BAD099
✨ Empowering agriculture with AI for smarter markets and stronger communities. 🌍