Feature Selection Optimization Using Hybrid Genetic Algorithm and Recursive Feature Elimination for High-Dimensional Data.

Abstract

High-dimensional datasets, commonly encountered in domains such as bioinformatics, text mining, and financial analytics, often suffer from the curse of dimensionality, leading to increased computational complexity and reduced model performance. Effective feature selection is therefore essential to enhance classification accuracy while minimizing redundancy and irrelevant information. This study proposes a hybrid feature selection framework that integrates a Genetic Algorithm (GA) with Recursive Feature Elimination (RFE) to address the limitations of traditional methods. The Genetic Algorithm is employed as a global search strategy to explore the feature space and identify promising subsets, while RFE is utilized as a local optimization technique to iteratively eliminate less significant features based on model-specific importance scores. The proposed GA–RFE hybrid approach aims to balance exploration and exploitation, enabling efficient identification of optimal feature subsets in high-dimensional environments. The framework is evaluated on multiple benchmark datasets characterized by large feature spaces. Performance is assessed using standard evaluation metrics, including classification accuracy, precision, recall, and F1-score, alongside computational efficiency. Experimental results demonstrate that the hybrid method significantly outperforms standalone GA, RFE, and traditional dimensionality reduction techniques such as Principal Component Analysis (PCA) and LASSO in terms of both predictive performance and feature reduction. Furthermore, the proposed approach achieves a substantial reduction in feature dimensionality without compromising model robustness, thereby facilitating faster training and improved generalization. The findings highlight the effectiveness of combining evolutionary optimization with recursive elimination for feature selection in complex datasets. This research contributes to the development of scalable and intelligent feature selection techniques and provides a practical solution for handling high-dimensional data in real-world machine learning applications.