🌾 Food Supply Chain AI Platform

Quick Start

Installation

pip install -r requirements.txt

Run Application

python app.py

Access Dashboard

Open browser to: http://localhost:5003

Features

✨ AI-Powered Analytics Dashboard 💡 AI Insights on KPI Cards
🤖 AI Recommendation Boxes 📊 Farm Comparison Overview 🏭 4 Farms (Farm A, B, C, D) with Different Performance Levels 📋 Detailed View Pages with Pagination 📱 Fully Responsive Design ⚡ Real-Time Chart.js Visualizations 🔍 Performance Scoring System 🎯 Crop Recommendation Engine - AI-driven crop allocation to maximize profit 📈 Price Prediction - 6-month crop price forecasts using SARIMA models 🤖 Multilingual AI Chatbot - English, Hindi, and Kannada support with Gemini API

Farm Data

The application uses 4 separate CSV files for 4 different farms:

• Farm A: Excellent performance (high yield, low spoilage, high satisfaction)
• Farm B: Average performance (moderate metrics)
• Farm C: Needs attention (low yield, high spoilage, low satisfaction)
• Farm D: Good performance (above average metrics)

Each farm has 125 records with comprehensive data across all supply chain stages.

File Structure

AI-Food-Chain/
├── app.py                          # Flask backend with all API endpoints
├── requirements.txt                # Python dependencies
├── README.md                       # This file
├── QUICK_SETUP.md                 # Quick setup guide
├── SETUP_INSTRUCTIONS.md          # Detailed setup instructions
├── farm_a_data.csv                # Farm A data
├── farm_b_data.csv                # Farm B data
├── farm_c_data.csv                # Farm C data
├── farm_d_data.csv                # Farm D data
├── food_supply_chain_data.csv     # Combined supply chain data
├── models/
│   ├── price_predictor.py         # Price prediction model utilities
│   ├── sarima_wheat_price_model.pkl
│   ├── sarima_corn_price_model.pkl
│   ├── sarima_lettuce_price_model.pkl
│   └── sarima_tomato_price_model.pkl
├── updated_farm_data/             # Updated farm data snapshots
│   ├── updated_farm_a_data.csv
│   ├── updated_farm_b_data.csv
│   ├── updated_farm_c_data.csv
│   └── updated_farm_d_data.csv
└── templates/
    └── index.html                 # Main dashboard (single-page app)

Dashboard Sections

🎯 Crop Recommendation (NEW!)

The intelligent crop recommendation system analyzes:

• Predicted crop prices from SARIMA models
• Farm's historical performance with each crop
• Profitability scores based on yield, spoilage, defects, shelf life, and pest risk
• Cross-farm optimization to prevent market saturation

Features:

• View all farms' optimal crop allocations
• See profitability scores (0-100)
• Get detailed reasoning for each recommendation
• Click to view farm-specific recommendations with profit analysis
• Understand how crop allocation prevents market overlap

Example Allocations:

• FarmA → Tomato (Profitability: 133.74)
• FarmB → Lettuce (Profitability: 77.71)
• FarmC → Wheat (Profitability: 45.56)
• FarmD → Corn (Profitability: 109.88)

📈 Price Prediction

• 6-Month Forecasts for all 4 crops (Wheat, Corn, Lettuce, Tomato)
• SARIMA Models with Box-Cox transformations for accurate predictions
• Confidence Intervals showing upper and lower bounds
• Interactive Charts with Chart.js visualizations
• Price/Quintal predictions for market planning

Overview Tab

• Farm Comparison: View all 4 farms side-by-side
• Performance Scores: See which farms are doing well and which need attention
• Comparative Charts: Yield, spoilage, defects, and performance score comparisons
• AI Insights: Recommendations on which farms need immediate attention

Individual Farm Tabs (Farm A, B, C, D)

1. 📊 Overview - Farm-specific key metrics and performance score
2. 🌱 Production - Yield, pest risk, harvest and machinery uptime
3. ❄️ Storage - Temperature, humidity, spoilage rate, shelf life
4. ⚙️ Processing - Defect rates, machinery uptime, packaging speed
5. 🚚 Transportation - Distance, fuel usage, delivery time, delays
6. 🏪 Retail - Inventory, sales velocity, pricing, waste
7. 👨‍👩‍👧‍👦 Consumption - Household waste, recipe accuracy, satisfaction
8. ♻️ Waste - Segregation, upcycling, biogas output
9. 🎯 Crop Recommendation - Farm-specific crop recommendations

API Endpoints

Crop Recommendation

• GET /api/farm/crop-recommendations-all - Optimal allocation for all farms
• GET /api/farm/<farm_name>/crop-recommendation - Detailed recommendation for specific farm

Price Prediction

• GET /api/prediction/price/<crop_name> - Price forecast for a crop (JSON)
• GET /api/prediction/price - All crop price predictions (HTML page)

Overview

• GET /api/overview - Compare all farms performance

Farm-Specific

• GET /api/farm/<farm_name>/kpis - Farm KPI metrics
• GET /api/farm/<farm_name>/production - Production data
• GET /api/farm/<farm_name>/storage - Storage data
• GET /api/farm/<farm_name>/processing - Processing data
• GET /api/farm/<farm_name>/transportation - Transportation data
• GET /api/farm/<farm_name>/retail - Retail data
• GET /api/farm/<farm_name>/consumption - Consumption data
• GET /api/farm/<farm_name>/waste - Waste data

AI Insights

• GET /api/ai-insights/<farm_name>/<section> - AI-generated insights for farm and section
• POST /api/chatbot - Multilingual AI chatbot (supports English, Hindi, Kannada)

Details

• GET /details/<farm_name>/<stage> - Detailed view pages with pagination
• GET /details/all/<stage> - Comparison details for all farms

Performance Scoring

Farms are scored on a 0-100 scale based on:

• Yield (20 points)
• Spoilage rate (15 points)
• Defect rate (15 points)
• Delivery delays (10 points)
• Waste percentage (10 points)
• Customer satisfaction (15 points)
• Pest risk (10 points)
• Machinery uptime (5 points)

Performance Levels:

• 80-100: Excellent Performance ✅
• 65-79: Good Performance ✅
• 50-64: Average Performance ⚠️
• 0-49: Needs Attention ❌

Crop Recommendation Algorithm

The system uses a greedy optimization algorithm:

1. Price Prediction: SARIMA models predict next 6 months average price
2. Profitability Calculation: Score = Price × Yield - (Spoilage + Defects + Waste) + Quality Factors
3. Allocation Matrix: Creates score matrix for all farm-crop combinations
4. Greedy Selection: Assigns highest-scoring farm-crop pairs while preventing overlap
5. Market Protection: Each crop allocated to exactly one farm to maximize price

Default Prices (Fallback)

• Wheat: ₹2200/quintal
• Corn: ₹1800/quintal
• Lettuce: ₹1200/quintal
• Tomato: ₹1500/quintal

Sample Data Fields

Each farm CSV contains 125 records with 39 fields covering:

• Production: Crop type, soil moisture, temperature, rainfall, fertilizer, yield, pest risk, harvest robot uptime
• Storage: Temperature, humidity, spoilage rate, grading score, shelf life, storage days
• Processing: Process type, packaging type, packaging speed, defect rate, machinery uptime
• Transportation: Transport mode, distance, fuel usage, delivery time, delays, spoilage in transit
• Retail: Inventory, sales velocity, pricing index, waste percentage
• Household: Household waste, recipe accuracy, satisfaction score
• Waste: Waste type, segregation accuracy, upcycling rate, biogas output

AI Chatbot with Gemini Integration

The platform includes a multilingual AI chatbot powered by Google's Gemini API.

Setting Up Gemini API

1. Get API Key:

   • Visit Google AI Studio
   • Create a new API key for the Gemini API

2. Set Environment Variable:

   # Windows PowerShell
   $env:GEMINI_API_KEY="your-api-key-here"
   
   # Linux/Mac
   export GEMINI_API_KEY="your-api-key-here"

3. Or Add to .env file:

   GEMINI_API_KEY=your-api-key-here
   

Using the Chatbot

• Click the chatbot button to open the chat interface
• Select language: English, Hindi, or Kannada
• Ask questions about farms, metrics, performance, comparisons, etc.
• The chatbot uses Gemini with full context of all farm data
• Farmer-focused expertise with natural conversational tone

Supported Languages

• 🇬🇧 English - Professional farming English
• 🇮🇳 Hindi - हिंदी (Devanagari script)
• 🇮🇳 Kannada - ಕನ್ನಡ (Kannada script)

Example Questions

• "Which farm is best for tomato cultivation?"
• "Compare yields across all farms"
• "What crop should Farm A grow next season?"
• "Show me storage conditions for all farms"
• "Which farm needs the most attention?"
• "What's the expected profit for lettuce in Farm D?"

Text-to-Speech (TTS)

The chatbot supports multilingual text-to-speech:

• English & Hindi: Browser native TTS (Web Speech API)
• Kannada: Google Translate TTS (via gTTS)
• Click speaker icon to hear chatbot responses

Environment Variables

GEMINI_API_KEY=<your-gemini-api-key>
OLLAMA_MODEL=llama2  # Optional, for local LLM fallback

Technology Stack

• Backend: Flask (Python web framework)
• Frontend: HTML5, CSS3, JavaScript (vanilla)
• Charts: Chart.js (data visualization)
• ML Models: SARIMA (price prediction), Statsmodels
• AI: Google Gemini API (chatbot)
• TTS: Web Speech API, gTTS
• Data: Pandas, NumPy (data processing)

Troubleshooting

Models Not Loading

# Test model loading
python -c "from app import load_models, trained_models; load_models(); print(trained_models)"

API Endpoints Not Responding

# Check if Flask app is running
curl http://localhost:5003/api/overview

Crop Recommendations Not Showing

• Ensure models/ directory contains all 4 .pkl files
• Check browser console for JavaScript errors (F12)
• Verify /api/farm/crop-recommendations-all returns valid JSON

Chatbot Not Working

• Verify GEMINI_API_KEY is set: echo $env:GEMINI_API_KEY
• Check if API key has Gemini API enabled
• Review browser console for errors

Performance Tips

• Caching: Farm data is cached in memory for efficiency
• Pagination: Large datasets use pagination (50 records per page)
• Lazy Loading: Charts and data load on-demand when tabs are selected
• Responsive Design: Optimized for desktop, tablet, and mobile

Future Enhancements

• Real-time data integration with IoT sensors
• Machine learning for anomaly detection
• Advanced crop rotation recommendations
• Supply chain optimization with ML
• Multi-year trend analysis
• Weather integration for crop planning
• Market price tracking and alerts

Enjoy! 🌾✨📊