Comparative Study between Two-Class SVM and One-Class SVM Classifiers for Outlier Detection for Disease Diagnosis

Abstract

Data mining techniques can be effectively used for major disease diagnosis. Outlier detection based on classification techniques is widely used in disease diagnosis. support vector machine (SVM)-based outlier detection can be used for detecting rare and abnormal objects from disease dataset. SVM-based outlier detection can be possible in two ways: two-class SVM and one-class SVM. Both classifiers have their own advantages and limitations. In this paper, we have applied both classifiers on multiple disease datasets and compared these classifier performances with various measures. Experimental result shows that two-class SVM gives better performance than one-class SVM with respect to accuracy, sensitivity, specificity and RMSE but it has a long training time compared with oneclass SVM.