Abstrait

CLASSIFICATION AND COMPARATIVE STUDY OF DATA MINING CLASSIFIERS WITH FEATURE SELECTION ON BINOMIAL DATA SET

Pushpalata Pujari

This paper describes about the performance analysis of different data mining classifiers before and after feature selection on binomial data set. Three data mining classifiers Logistic Regression, SVM and Neural Network classifiers are considered in this paper for classification. The Congressional Voting Records data set is a binomial data set investigated in this study is taken from UCI machine learning repository. The classification performance of all classifiers is presented by using statistical performance measures like accuracy, specificity and sensitivity. Gain chart and R.O.C (Receiver Operating Characteristics) chart are also used to measure the performances of the classifiers. A comparative study is carried out among the data mining classifiers. Experimental result showed that without feature selection Logistic Regression and SVM classifiers provides 100% accuracy and neural network provides 98.13 % accuracy on test data set. With feature selection SVM classifier provides 100% accuracy. The performance of SVM classifier is found to be the best among all the classifiers with reduced number of features.