Hybrid Grey Butterfly Optimizer-Based Feature Selection for Enhancing Agricultural Commodity Price Prediction Using Machine Learning Classifiers

Abstract

One of the most important tasks for economic planning and decision-making in the agriculture industry is the prediction of agricultural commodity prices. However, high dimensionality and redundancy are common in this field's datasets, increasing computing cost and decreasing prediction accuracy. To address these problems, this study proposes a novel Hybrid Grey Butterfly Optimizer (HGBO) for feature selection that combines the exploratory nature of the Butterfly Optimization Algorithm (BOA) with the leadership-driven search capabilities of Grey Wolf Optimization (GWO). The proposed hybrid approach enhances the feature selection process by avoiding local optima and preventing premature convergence. The selected feature subsets were assessed using machine learning classifiers including XGBoost, Random Forest, Gradient Boosting, and Support Vector Machines (SVM) to know the efficacy of the HGBO strategy. The performance of the proposed method was assessed in comparison to alternative optimization techniques, utilizing a range of evaluation metrics such as accuracy, precision, recall, and the F1-score. The findings indicate that HGBO demonstrates superior performance by identifying the most pertinent feature subsets, thereby enhancing predictive accuracy. This research underscores the HGBO algorithm's capacity for forecasting agricultural prices, presenting a valuable approach for addressing high-dimensional data challenges within agricultural analytics. .