🌾 MANDI-MITRA XAI

Explainable Market Intelligence & Agricultural Price Forecasting for Madhya Pradesh Farmers

MandiMitra XAI is a full-stack,production-grade Explainable Artificial Intelligence (XAI) platform designed to assist farmers in Madhya Pradesh (India) with data-driven agricultural decisions.

Unlike conventional black-box AI systems, MandiMitra XAI emphasizes transparency by providing SHAP-based explanations for every prediction, enabling farmers to understand why the AI recommends a particular action.

✨ Key Feature

• 🔮 Probabilistic Forecasting: 14-day price predictions using PyTorch LSTM with Bahdanau attention for 10 major crops across 10 districts.
• 🧠 Explainability (XAI): SHAP (Shapley Additive exPlanations) provides transparent reasoning behind AI predictions in simple language.
• 💰 Sell vs. Store Optimizer: A decision engine that recommends whether to sell now or store, factoring in weather-dependent storage costs and crop spoilage rates.
• 🌦️ Harvest Window Optimizer: Recommends optimal harvest timing based on live data from Open-Meteo API.
• 📜 Bilingual Interface: Full support for English and Hindi (i18n ready) via a modern, farmer-friendly React dashboard.
• 📡 Live Market Data: Actively syncs real-time prices from the Indian Government's Agmarknet (data.gov.in) API.

🛠️ Technology Stack

Backend & Machine Learning

• Framework: FastAPI, Uvicorn, Motor (Async MongoDB)
• ML Core: PyTorch (LSTM + Attention), SHAP, XGBoost, scikit-learn
• Data Pipeline: httpx, openmeteo-requests for live APIs

Frontend

• Framework: React 18, Vite
• Libraries: Recharts (Data Viz), Axios, date-fns, React Router
• Styling: Premium modern CSS, responsive design with interactive elements.

🚀 Local Setup Instructions

1. Environment Configuration

Clone the repository and set up environment variables:

git clone https://github.com/Nightkilller/MANDI-MITRA.git
cd MANDI-MITRA
cp .env.example .env

(Fill in your local instances or mock cluster URIs in .env)

2. Backend & ML Setup

# Create and activate virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt
pip install -r requirements-dev.txt

# Seed mock database (optional)
python scripts/seed_mock_data.py 

# Start FastAPI server
uvicorn backend.main:app --reload

The API will be running at http://localhost:8000

3. Frontend Setup

# In a new terminal
cd frontend

# Install dependencies and start Web UI
npm install
npm run dev

The React App will be accessible at http://localhost:5173

📚 Documentation

• For detailed architecture, models, and results, read the PROJECT_REPORT.md.
• For Render/Vercel free-tier deployment strategies, read the DEPLOYMENT.md.