A Hybrid Filter-Wrapper Approach for FeatureSelection

Feature selection is the task of selecting a small subset from original featuresthat can achieve maximum classification accuracy. This subset of features hassome very important benefits like, it reduces computational complexity of learningalgorithms, saves time, improve accuracy and the selected features can beinsightful for the people involved in problem domain. This makes feature selectionas an indispensable task in classification task.This dissertation presents a two phase approach for feature selection. In thefirst phase a filter method is used with “correlation coefficient” and “mutualinformation” as statistical measure of similarity. This phase helps in improvingthe classification performance by removing redundant and unimportantfeatures. A wrapper method is used in the second phase with the sequentialforward selection and sequential backward elimination. This phase helps in selectingrelevant feature subset that produce maximum accuracy according tothe underlying classifier. The Support Vector Machine (SVM) classifier (linearand nonlinear) is used to evaluate the classification accuracy of our approach.This empirical results of commonly used data sets from the University ofCalifornia, Irvine repository and microarray data sets showed that the proposedmethod performs better in terms of classification accuracy, number of selectedfeatures, and computational efficiency.7