Comparisons of Logistic Regression and Support Vector Machines in Classification of Echocardiogram Dataset

Abstract

Echocardiography is a test that uses sound waves to produce an image of our heart. This image is called an echocardiogram. This paper uses Echocardiogram Dataset, in which the problem is to classify from 7 features whether the patient will survive or not. In this study, the classification method is used to solve this problem. Some classification methods can be applied to classify category response variables, such as Logistic regression and Support Vector Machines (SVM). The method for predicting best accuracy used holdout and cross-validation. Before doing classification, some preprocessing procedures were applied to this dataset. The preprocessing procedures include missing value imputation using median imputation, outliers detection in univariate and multivariate procedures, and feature selection using the backward method. The result of classification in the analysis showed that SVM with unstratified holdout gave the best accuracy, that is 91.54%.