A diagnostic model based on hybrid features selection method for the diagnosis of clinical diseases

Abstract

This paper developed a diagnosis model based on hybrid feature selection method to diagnose erythemato-squamous diseases. Our hybrid feature selection method, named HCELFS (Hybrid Correlation Evaluator and Linear Forward Selection), combines the advantages of filters and wrappers to select the optimal feature subsets from the original feature set. In HCELFS, Correlation Attribute Evaluator acts as filters to remove redundant features and Linear Forward Selection with some machine learning algorithms acts as the wrappers to select the ideal feature subset from the remaining features. Several experiments using WEKA had been conducted, utilizing 10 fold cross validations. The experimental results with erythemato-squamous diseases data set demonstrate that our proposed model has a better performance than some well-known feature selection algorithms with optimal classification accuracy with no more than 16 features for erythemato-squamous diseases