Impact of Feature Selection on Support Vector Machine Using Microarray Gene Expression Data

Abstract

Recent researches have investigated the impact of feature selection methods on the performance of support vector machine (SVM) and claimed that no feature selection methods improve it in high dimension. However, they have based this argument on their experiments with simulated data. We have taken this claim as a research issue and investigated different feature selection methods on the real time micro array gene expression data. Our research outcome indicates that feature selection methods do have a positive impact on the performance of SVM in classifying micro array gene expression data.