An Automated Parameter Selection Approach for Simultaneous Clustering and Feature Selection

Abstract

In this paper, an improvisation in Niching Memetic Algorithm for Simultaneous Clustering and Feature Selection (NMA_CFS) is proposed. In NMA_CFS, the parameters such as replacement group size, selection group size and population size are determined empirically and are manually obtained after hit and trial experimentation. An automated approach is proposed to determine these parameters of NMA_CFS. The experimental results reveal that this modified NMA_CFS does not deteriorate the performance of NMA_CFS due to automation, compare to the original NMA_CFS.