Open Access

ARTICLE

SCChOA: Hybrid Sine-Cosine Chimp Optimization Algorithm for Feature Selection

by Shanshan Wang^1,2,3, Quan Yuan¹, Weiwei Tan¹, Tengfei Yang¹, Liang Zeng^1,2,3,*

1 School of Electrical and Electronic Engineering, Hubei University of Technology, Wuhan, 430068, China
2 Hubei Key Laboratory for High-Efficiency Utilization of Solar Energy and Operation Control of Energy Storage System, Hubei University of Technology, Wuhan, 430068, China
3 Xiangyang Industrial Institute of Hubei University of Technology, Xiangyang, 441100, China

* Corresponding Author: Liang Zeng. Email:

Computers, Materials & Continua 2023, 77(3), 3057-3075. https://doi.org/10.32604/cmc.2023.044807

Received 09 August 2023; Accepted 19 October 2023; Issue published 26 December 2023

Abstract

Feature Selection (FS) is an important problem that involves selecting the most informative subset of features from a dataset to improve classification accuracy. However, due to the high dimensionality and complexity of the dataset, most optimization algorithms for feature selection suffer from a balance issue during the search process. Therefore, the present paper proposes a hybrid Sine-Cosine Chimp Optimization Algorithm (SCChOA) to address the feature selection problem. In this approach, firstly, a multi-cycle iterative strategy is designed to better combine the Sine-Cosine Algorithm (SCA) and the Chimp Optimization Algorithm (ChOA), enabling a more effective search in the objective space. Secondly, an S-shaped transfer function is introduced to perform binary transformation on SCChOA. Finally, the binary SCChOA is combined with the K-Nearest Neighbor (KNN) classifier to form a novel binary hybrid wrapper feature selection method. To evaluate the performance of the proposed method, 16 datasets from different dimensions of the UCI repository along with four evaluation metrics of average fitness value, average classification accuracy, average feature selection number, and average running time are considered. Meanwhile, seven state-of-the-art metaheuristic algorithms for solving the feature selection problem are chosen for comparison. Experimental results demonstrate that the proposed method outperforms other compared algorithms in solving the feature selection problem. It is capable of maximizing the reduction in the number of selected features while maintaining a high classification accuracy. Furthermore, the results of statistical tests also confirm the significant effectiveness of this method.

---

Keywords

---