Microarray Data Analysis Using Masked Sequential Backward Selection

Abstract

In any microarray data analysis, selecting the optimal number of features is a very important task without information loss and removing unwanted data. Also, the coverage of training samples should be high considering only the best features in a dataset. In this paper, a new method called masked sequential backward selection technique is proposed and its performance is compared with other popular feature selection techniques. The method selects the most differentially expressed genes based on its expression value distribution. Also, it allows the user to select the number of features to be retrieved to get best classification accuracy and study a larger set of high relevant genes for the target disease.