View in EDS

Feature Selection Method based on Neighborhood Rough Sets and Crazy Adaptive Whale Optimization Algorithm.

Saved in:
Bibliographic Details
Title: Feature Selection Method based on Neighborhood Rough Sets and Crazy Adaptive Whale Optimization Algorithm.
Authors: Zhou, Yun, Sitjongsataporn, Suchada
Source: International Journal of Intelligent Engineering & Systems; 2026, Vol. 19 Issue 1, p368-388, 21p
Subject Terms: FEATURE selection, ROUGH sets, DATA reduction, OPTIMIZATION algorithms, STATISTICS, K-nearest neighbor classification, METAHEURISTIC algorithms
Abstract: This paper proposes a feature selection method that integrates Crazy Adaptive Whale Optimization Algorithm (CAWOA) with Neighborhood Rough Sets (CAWOA-NRS), eliminating redundant features from datasets. The algorithm proposed in this paper first constructs the original decision table; Subsequently, the population is initialized using symmetric PWLCM chaotic mapping algorithm, and prey search position of traditional whale optimization algorithm is improved by using adaptive dynamic weight and crazy perturbation factor. Crazy adaptive whale optimization algorithm was tested and verified using six benchmark functions in the CEC 2005 test set. CAWOA algorithm optimized the neighborhood rough sets with the same initial conditions and was compared with other improved rough sets feature selection algorithms. Finally, the KNN classifier is randomly sampled for training to calculate the classification accuracy, reduction rate and processing time. Taking the heart dataset in the UCI repository as an example, CAWOA-NRS algorithm was analyzed for performance compared with the other five population intelligence algorithms, maintaining the highest classification accuracy of 82.78%, Processing time of 2.807s, and feature selection rate of 46.15%. Meanwhile, statistical analysis was conducted using the Friedman Test and Bonferroni-Dunn test. The average ranking of the CAWOA-NRS algorithm was 1.17. The chi-square statistic x²2 ≈19.50 calculated in the experiment is greater than 11.07, indicating that the null hypothesis is rejected at a significance level of 0.05. The experimental results show that the proposed CAWOA-NRS algorithm can obtain the minimum attribute reduction set while also having superior decision-making ability after reduction. The proposed method has certain superiority in practical applications [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Intelligent Engineering & Systems is the property of Intelligent Networks & Systems Society and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:This paper proposes a feature selection method that integrates Crazy Adaptive Whale Optimization Algorithm (CAWOA) with Neighborhood Rough Sets (CAWOA-NRS), eliminating redundant features from datasets. The algorithm proposed in this paper first constructs the original decision table; Subsequently, the population is initialized using symmetric PWLCM chaotic mapping algorithm, and prey search position of traditional whale optimization algorithm is improved by using adaptive dynamic weight and crazy perturbation factor. Crazy adaptive whale optimization algorithm was tested and verified using six benchmark functions in the CEC 2005 test set. CAWOA algorithm optimized the neighborhood rough sets with the same initial conditions and was compared with other improved rough sets feature selection algorithms. Finally, the KNN classifier is randomly sampled for training to calculate the classification accuracy, reduction rate and processing time. Taking the heart dataset in the UCI repository as an example, CAWOA-NRS algorithm was analyzed for performance compared with the other five population intelligence algorithms, maintaining the highest classification accuracy of 82.78%, Processing time of 2.807s, and feature selection rate of 46.15%. Meanwhile, statistical analysis was conducted using the Friedman Test and Bonferroni-Dunn test. The average ranking of the CAWOA-NRS algorithm was 1.17. The chi-square statistic x²<subscript>2</subscript> ≈19.50 calculated in the experiment is greater than 11.07, indicating that the null hypothesis is rejected at a significance level of 0.05. The experimental results show that the proposed CAWOA-NRS algorithm can obtain the minimum attribute reduction set while also having superior decision-making ability after reduction. The proposed method has certain superiority in practical applications [ABSTRACT FROM AUTHOR]
ISSN:2185310X
DOI:10.22266/ijies2026.0131.23