A polyhedral conic functions based embedded feature selection method
Optimization Eruditorum, Volume 1, Issue 2, December 2024, Pages 88–100
REFAIL KASIMBEYLI
Department of Industrial Engineering, Faculty of Engineering, Eskisehir Technical University, Eskisehir 26555, Türkiye
OZNUR AY
UNEC Mathematical Modeling and Optimization Research Center, Azerbaijan State University of Economics (UNEC), Istiqlaliyyat Str. 6, Baku, 1001, Azerbaijan
Abstract
In this paper, a polyhedral conic functions based embedded feature selection method is proposed. The original PCF algorithm developed for classification, is reformulated such that, both feature selection and classification are performed simultaneously. The proposed algorithm is tested on some currently available real world data sets and compared with other well known feature selection and classification algorithms. It is shown that classifiers obtained by the new method, have better test accuracies on some test problems, exhibit comparable prediction performance on all data sets tested, and increase the generalization performance.
Cite this Article as
Refail Kasimbeyli and Oznur Ay, A polyhedral conic functions based embedded feature selection method, Optimization Eruditorum, 1(2), 88–100, 2024