Optimizing Crop Yield Prediction - A Comparative Analysis and Development of a Hybrid Algorithm for Regional Agricultural Data

Abstract

Accurate crop yield prediction is essential for optimizing resource allocation, managing risks, and ensuring sustainable agricultural practices. This study introduces a novel hybrid algorithm that integrates multiple predictive models, including Linear Regression, Decision Trees, Random Forests, and Neural Networks, with a Gradient Boosting Machine (GBM) as the meta-model, to improve the accuracy of region-specific crop yield predictions. Using required dataset, covering environmental, agricultural, and economic factors, the hybrid algorithm demonstrated superior performance compared to individual models. It achieved an RMSE of 17.55 tons/ha, MAE of 13.80 tons/ha, and an R² of 0.87, outperforming state-of-the-art models. The study’s findings underscore the hybrid algorithm’s ability to capture complex, non-linear relationships in agricultural data, improving the precision of crop yield forecasts. This enhanced predictive capability can support farmers and policymakers in making informed decisions, optimizing resource use, and mitigating risks associated with climate variability. However, limitations such as dataset specificity and increased computational complexity highlight the need for further refinement. Future research should focus on expanding the dataset to diverse geographical regions and optimizing the algorithm for broader applicability.