Enhancement of Embedded Feature Selection Method for 3D Molecular Structure of Amphetamine-Type Stimulants (ATS) Drugs

Abstract

The fundamental of this research paper is to propose the enhancement of the embedded feature selection method between Sequential Forward Floating Selection (SFFS) and Support Vector Machine-Recursive Feature Elimination (SVM- RFE). These two feature selection methods were primarily embedded to enhance the effectiveness and quality of data identification. In this research, three feature selection variables from MATLAB were used as a mechanism to compare and evaluate the best feature throughout the embedded feature selection process. Those features are fscmrmr, relieff and chisquare. Lastly, a standard evaluation technique, a crossvalidation process in WEKA, is used to systematically run repeated percentage splits. This study ran selected classifiers with 10 times cross-validations to capture each experiment’s accuracy.