Crop Yield Prediction in Maharashtra Using Machine Learning: A Comparative Analysis of Multiple Models
TL;DR
The results indicate that ensemble models outperform traditional approaches in crop yield prediction, with XGBoost achieving the highest performance, and the effectiveness of machine learning techniques, particularly ensemble methods, in improving crop yield prediction.
Abstract
This study aims to develop and evaluate machine learning models for crop yield prediction using environmental and soil-related factors. The dataset includes key variables such as rainfall, temperature, humidity, soil pH, and crop production data collected from different regions. Various machine learning algorithms, including Linear Regression, Decision Tree, Random Forest, and XGBoost, were implemented and compared to identify the most effective model for accurate yield prediction. The results indicate that ensemble models outperform traditional approaches, with XGBoost achieving the highest performance with an R² score of 0.86 and RMSE 224.59, followed by Random Forest with 0.85 and RMSE 238.11. Linear Regression R² score 0.573 and RMSE 404.75 showed comparatively lower performance due to its limitation in capturing non-linear relationships in agricultural data. Feature importance analysis revealed that rainfall, humidity, and temperature are the most influential factors affecting crop yield. Crop-wise analysis further demonstrated that prediction accuracy varies across different crops, with chickpea showing higher accuracy compared to rice and maize, which exhibited greater variability. The comparison between actual and predicted values confirms that the models are capable of capturing overall yield trends with reasonable accuracy. The findings highlight the effectiveness of machine learning techniques, particularly ensemble methods, in improving crop yield prediction. This study provides useful insights for agricultural planning, resource management, and decision-making, contributing towards sustainable farming practices.