Machine Learning

Climate and Soil-Based Agricultural Crop Yield Forecasting System

The Climate and Soil-Based Agricultural Crop Yield Forecasting System is a comprehensive machine learning system that predicts crop yields using historical agricultural data. This project is ideal for B.Tech, MCA, CSE, and engineering students as their final year project. It implements Lasso, Ridge, and ElasticNet regression with robust feature engineering and 5-fold cross-validation, achieving 78.2% R² accuracy.

The system utilizes agricultural crop yield data from the FAO spanning 1960-2020, containing over 15,000 records with features including crop type, country, and time-based engineered features like year_since_1960, year_squared, year_cubic, and interaction terms (year_crop_interaction, year_country_interaction). It provides a web-based interface for data upload, exploratory data analysis, model training, and yield prediction, making sophisticated forecasting accessible to agricultural stakeholders.

• Multi-Model Crop Yield Forecasting
• Ridge Regression (78.2% R²)
• Lasso Regression (77.6% R²)
• ElasticNet Regression (76.9% R²)
• Feature Engineering (temporal trends, interactions)
• Interactive Visualizations
• Real-time Yield Predictions
• Model Performance Comparison
• Feature Importance Analysis
• Flask Web Application