On selecting features for binary classification in microarray data analyses

Feature selection has become the most interesting challenge in processing the analysis of high-dimensional microarray data. It addresses the issue of dimensionality reduction by obtaining important features to construct a good model, especially for classification. There are many different feature selection methods that have been proposed, developed, and, eventually, commonly used. Some of these methods are discussed by many researchers to be excellent to improve the accuracy of classification by taking care of redundant and irrelevant instances. The complicated associations among the genes in microarrays data tend to make works more difficult, but removing less important features can improve the accuracy. In this study, basic feature selection techniques were compared based on support vector machine performance in classifying binary classification problems. The experiments were based on high-dimensional microarray datasets which were preprocessed by reducing its dimensionality using correlation based feature selection and fast correlation based filter and evaluated based on classification accuracy resulted from support vector machine standard.