Title: A condensed hybrid feature selector for enhancing classifier performance using TOPSIS and improved Rao optimisation

Authors: A.S. Karthik Kannan; S. Appavu alias Balamurugan; Millie Pant

Addresses: Department of Applied Mathematics and Scientific Computing, Indian Institute of Technology Roorkee (IITR), Saharanpur Campus, India ' Department of Applied Mathematics and Scientific Computing, Indian Institute of Technology Roorkee (IITR), Saharanpur Campus, India ' Department of Applied Mathematics and Scientific Computing, Indian Institute of Technology Roorkee (IITR), Saharanpur Campus, India

Abstract: A wide range of fields are increasingly utilising high-dimensional data, such as text mining, bioinformatics, image processing, and pattern recognition. A classification system is less efficient as a result of the curse of dimensionality problem, which incurs high computational costs. With its efficacy to identify the optimal features from the feature space, an approach to selecting the more informative features based on MCDM and improved Rao optimisation methods is proposed. As a measure of the candidate solution's fitness, the proposed work uses the classifier error rate and feature selection ratio. A comparison of the proposed method with state-of-the-art methods is conducted using 12 popular benchmark datasets. A hybrid approach outperforms the standard strategy in terms of selected features and classification accuracy, according to the results of the experiment.

Keywords: hybrid feature selector; classification; multi-criteria decision making; MCDM; TOPSIS; improved Rao optimisation.

DOI: 10.1504/IJBIDM.2024.140240

International Journal of Business Intelligence and Data Mining, 2024 Vol.25 No.2, pp.210 - 238

Received: 17 Dec 2022
Accepted: 20 Nov 2023
Published online: 31 Jul 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article