Enhancing Crop Yield Prediction Using Machine Learning: A Comprehensive Study for Sustainable Agriculture

Precision agriculture has seen a change in philosophy lately thanks to the addition of machine learning (ML) methods, making data-driven choices to boost crop performance possible. In this study, the authors analyze crop yield prediction models using several ML methods known as Random Forest, XGBoost, Support Vector Regression (SVR) and LightGBM. The evaluation applies a dataset formed by looking at soil characteristics and the main crops grown in Maharashtra, India. Preparation of the dataset for learning was done using data and feature engineering methods. Each model’s accuracy was assessed using Measure Squared Error (MSE), Mean Absolute Error (MAE) and R². Random Forest and XGBoost turned out to be the leading models, almost reaching perfect correlation (R²), but SVR performed badly with a negative R² score, showing it misses out on important connections in agriculture. The results highlight how ensemble-based machine learning helps improve predictions and decision-making for agribusinesses. The research compares how algorithms work and also highlights the effect of specific soil features on what is grown. Thanks to these findings, agronomists, farmers and policymakers will find it easier to plan crops and divide available resources..