🌾 Crop Yield Prediction System

Author: Pushkarjay Ajay

🌱 About This Project

The Crop Yield Prediction System is a comprehensive, end-to-end machine learning solution designed to revolutionize agricultural decision-making by providing accurate, data-driven crop yield forecasts. In an era where climate change and population growth pose significant challenges to global food security, this project addresses the critical need for intelligent farming tools that empower farmers, agricultural researchers, and policymakers with actionable insights.

This system leverages advanced machine learning algorithms, specifically Gradient Boosting Regression, to analyze complex relationships between environmental factors, soil characteristics, and farm management practices. By processing data from 22 different crop types across 20 Indian states, the model has been trained on a robust synthetic dataset of 75,000 samples, achieving an impressive 96.27% accuracy (R² Score)—demonstrating its reliability for real-world agricultural applications.

The project encompasses the complete data science lifecycle: from synthetic data generation that mirrors real agricultural patterns, through exploratory data analysis and feature engineering, to model training, evaluation, and deployment. The production-ready REST API enables seamless integration with existing agricultural management systems, while the responsive web dashboard provides an intuitive interface for farmers and agricultural officers to obtain instant yield predictions without technical expertise.

Key innovations include multi-crop comparison capabilities, total production calculators with support for various land measurement units (Hectares, Acres, Bigha), and comprehensive technical documentation. The modular architecture ensures maintainability and extensibility, allowing for future enhancements such as real-time weather API integration, geo-spatial prediction maps, and mobile applications. Whether you're a smallholder farmer seeking to optimize your harvest or an agricultural enterprise planning large-scale operations, this system provides the predictive intelligence needed to make informed decisions and maximize agricultural productivity.

🎯 Overview

This project implements an end-to-end machine learning pipeline for predicting crop yield. It combines data from multiple agricultural datasets, performs intelligent feature engineering, trains ensemble models, and serves predictions via a REST API with an interactive dashboard.

Key Highlights

📊 75,000 samples synthetic dataset with 22 crops across 20 Indian states
🤖 96.27% accuracy (R² Score) with Gradient Boosting
🌐 REST API for seamless integration
📱 Responsive Dashboard with real-time predictions
📈 Multi-crop comparison feature
🔢 Total production calculator with unit conversions

📁 Project Structure

Crop-Yield-Prediction/
├── Dataset/                      # Raw data files
│   ├── agricultural_yield_test.csv
│   ├── Data.csv
│   └── indiancrop_dataset.csv
│
├── Phase-1/                      # Initial exploration & documentation
│   ├── crop yield prediction.ipynb
│   ├── README.md
│   └── *.pdf, *.pptx            # Project documentation
│
├── Phase-2/                      # Production ML pipeline
│   ├── src/                     # Source code modules
│   │   ├── config.py            # Configuration & constants
│   │   ├── utils.py             # Logging utilities
│   │   ├── data_generation.py   # Synthetic dataset generator
│   │   ├── visualization.py     # Plot generation
│   │   ├── training.py          # ML training pipeline
│   │   └── outlier_analysis.py  # Outlier detection
│   ├── api/                      # Flask REST API
│   │   ├── app.py
│   │   └── requirements.txt
│   ├── dashboard/                # Frontend UI
│   │   ├── index.html           # Main dashboard
│   │   ├── technical.html       # Technical documentation
│   │   ├── report.html          # Quick view presentation
│   │   ├── script.js
│   │   └── style.css
│   ├── model/                    # Trained model artifacts
│   ├── plots/                    # EDA visualizations (8 plots)
│   ├── logs/                     # Terminal output logs
│   ├── run_pipeline.py           # Master pipeline runner
│   ├── unified_dataset.csv       # 75K synthetic dataset
│   └── README.md
│
└── README.md                     # This file

✨ Features

🌡️ Prediction Inputs

Weather: Rainfall, Temperature, Humidity, Sunshine hours, GDD, Pressure, Wind Speed
Soil: Quality, pH, Organic Carbon, Soil Moisture, N-P-K nutrients
Management: Fertilizer amount, Irrigation type, Seed variety, Pesticide usage
Location: State, Crop type, Year, Crop price

📊 Dashboard Features

Real-time yield predictions
Backend status indicator
Multi-crop yield comparison
Total production calculator (Hectare/Acre/Bigha)
Data field status toggle (shows which fields are ML-active)
Interactive visualizations
Technical documentation
PPT-style quick view report

🛠️ Tech Stack

Category	Technologies
Backend	Python 3.11, Flask 3.1, scikit-learn
ML Model	Gradient Boosting Regressor
Data	pandas, numpy, joblib
Visualization	matplotlib, seaborn
Frontend	HTML5, CSS3, JavaScript (ES6+)
API	REST, JSON, CORS-enabled

🚀 Installation

Prerequisites

Python 3.9 or higher
pip package manager
Git

Clone Repository

git clone https://github.com/Pushkarjay/Crop-Yield-Prediction.git
cd Crop-Yield-Prediction

Setup Virtual Environment

# Create virtual environment
python -m venv .venv

# Activate (Windows)
.venv\Scripts\activate

# Activate (Linux/Mac)
source .venv/bin/activate

Install Dependencies

cd Phase-2/api
pip install -r requirements.txt

💻 Usage

Start Backend API

cd Phase-2/api
python app.py

The API will run at http://localhost:5000

Start Frontend (Development)

cd Phase-2/dashboard
python -m http.server 8080

Open http://localhost:8080 in your browser

Make a Prediction (API)

curl -X POST http://localhost:5000/predict \
  -H "Content-Type: application/json" \
  -d '{
    "Rainfall_mm": 500,
    "Temperature_C": 25,
    "Humidity": 70,
    "Soil_Quality": 80,
    "Nitrogen": 50,
    "Phosphorus": 30,
    "Potassium": 40,
    "Fertilizer_Amount_kg_per_hectare": 150,
    "Sunshine_hours": 6,
    "Soil_Humidity": 45,
    "Irrigation_Schedule": 5,
    "Seed_Variety": 2
  }'

🌐 Live Demo

GitHub Pages (Frontend Only)

🔗 Live Dashboard

⚠️ Note: The live demo frontend is hosted on GitHub Pages. The prediction functionality requires the backend API to be running locally. Follow these steps:

Clone the repository
Start the backend API (python app.py)
Use the live dashboard OR open local index.html
The dashboard will show "✓ Backend Online" when API is ready

Hosting Frontend on GitHub Pages

To host the frontend yourself:

Enable GitHub Pages:
- Go to repository Settings → Pages
- Source: Deploy from branch
- Branch: main → /root or /docs
- Save

Access URL:

https://<username>.github.io/Crop-Yield-Prediction/Phase-2/dashboard/

Backend Requirement:
- The frontend alone cannot make predictions
- Users must run python Phase-2/api/app.py locally
- Dashboard shows backend status indicator

📡 API Reference

Endpoint	Method	Description
`/`	GET	API information
`/health`	GET	Health check & model status
`/features`	GET	List of required features
`/model-info`	GET	Model metrics & info
`/predict`	POST	Single prediction
`/predict-batch`	POST	Batch predictions

Response Example

{
  "status": "success",
  "prediction": {
    "yield_kg_per_hectare": 636.21,
    "yield_tons_per_hectare": 0.636,
    "bags_per_hectare_50kg": 12.7
  },
  "crop_info": {
    "crop_type": "Rice",
    "state": "Bihar"
  },
  "total_production": {
    "total_kg": 3181.05,
    "total_tons": 3.181
  }
}

📈 Model Performance

Model	R² Score	MAE (kg/ha)	RMSE (kg/ha)
Gradient Boosting 🏆	0.9627	1,610	3,574
Random Forest	0.9594	1,782	3,728
Decision Tree	0.8834	2,514	6,314
Linear Regression	0.7156	5,221	9,866

Why Gradient Boosting? Selected for best generalization (CV R² = 0.9603), faster inference (~3x), smaller model size (~4x), and better outlier handling compared to Random Forest.

Feature Importance (Top 5)

Crop Type (~25%)
Year (~18%)
Rainfall (~12%)
Fertilizer Amount (~11%)
Temperature (~9%)

📸 Screenshots

Dashboard

Feature Analysis

Model Performance

🚀 Future Enhancements

Add crop type as ML feature (requires model retraining)
Integrate weather API for real-time data
Add geo-spatial prediction maps
Mobile application development
Cloud deployment (AWS/GCP)
Historical yield tracking
Seasonal pattern analysis

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Fork the project
Create your feature branch (git checkout -b feature/AmazingFeature)
Commit your changes (git commit -m 'Add some AmazingFeature')
Push to the branch (git push origin feature/AmazingFeature)
Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

👨‍💻 Author

Pushkarjay Ajay

GitHub: @Pushkarjay

🙏 Acknowledgments

Agricultural datasets from various open sources
scikit-learn for ML algorithms
Flask for API framework
The open-source community

Made with ❤️ for Agriculture

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
Dataset		Dataset
Phase-1		Phase-1
Phase-2		Phase-2
.gitattributes		.gitattributes
README.md		README.md

Pushkarjay/Crop-Yield-Prediction

Folders and files

Latest commit

History

Repository files navigation