Abstract—Features’ selection is a dimension reduction technique that aims to enhance classification accuracy by removing unrelated and redundant features. The Wrapper approach, one features selection strategy, provides an accurate estimation of classification performance. In view of this, we propose a new model of Evolutionary Wrapper Feature selection. This model exploits Extreme Learning Machines (ELM) to evaluate selected subsets, comprising a Genetic Algorithm (GA) as a search algorithm to find a set of feature subsets. A priority was assigned to each feature when GA had explored the space of feature combinations. The use of priority avoids replacing one feature with another of higher priority. The goal of this model is to investigate the accuracy rate of using feature selection methods and the impact of using priority with the features. Two machine learning classifiers are considered: the ELM and the Support Vector Machine (SVM). The proposed model is piloted based on a Chronic Kidney Disease dataset (CKD) from UCI. Experimental results indicate that the proposed model can achieve a better accuracy rate with these two classifiers. In addition, it requires much less time to find the best subset of features.
 

Index Terms—extreme learning machine, evolutionary, wrapper feature selection, genetic algorithm, features priority

Cite: Methaq Kadhum, Saher Manaseer, and Abdel Latif Abu Dalhoum, "Evaluation Feature Selection Technique on Classification by Using Evolutionary ELM Wrapper Method with Features Priorities," Journal of Advances in Information Technology, Vol. 12, No. 1, pp. 21-28, February 2021. doi: 10.12720/jait.12.1.21-28

Copyright © 2021 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.