Hybrid framework of ID3 with multivariate attribute selection for heart disease analysis

Abstract

Heart disease is the heart- and blood-vein condition. It is very difficult for medical practitioners and physicians to assess accurately on the diagnosis of heart disease. There are many cardiovascular risk factors. Recognizing the importance of heart disease test variables for categorizing patients based on the results is significant. This research aimed at gaining a deeper understanding by applying machine learning techniques in R to examine the heart disease risk factors. In this study we proposed a new hybrid feature selection algorithm that combines the heuristic approach of Information Gain (IG) of Decision Tree Induction with Multivariate Feature Selection's Recursive Feature Elimination (RFE) for sub-set selection of attributes. Experimental findings show that the method of selection proposed by HIGRFE greatly reduces the redundancies among features. Compared to traditional information gain and recursive feature removal algorithms, the subset of features provides higher classification accuracy. We have implemented SVM classifier to evaluate the efficiency of the reduced feature subset, which has proved the power of the proposed HIGRFE-Selection algorithm