Heart disease prediction using ensemble methods

Abstract

Nowadays, people are suffering from many health issues. One of them is heart disease among the worldwide population. This causes due to imbalance lifestyle and unhealthy food consumption. The data generated by hospitals is huge and complex by nature which store patients medical and demographic information. Accurate and prompt diagnosis of heart diseases are becoming more challenging task in medical domain due to the complex data. Therefore, the computer aided systems are useful to store this complex and multivariate data to generate useful decisions. Machine learning techniques are used to classify and to predict the diseases. In this study, Majority voting classifier and Bagging ensemble method both have been evaluated. These ensemble methods combined the five base classifiers including DT (Decision Tree), LR (Logistic Regression), ANN (Artificial Neural Network), NB (Naïve Bayesian), and KNN (K-Nearest Neighbour). Bagging ensemble approach is used to combine the multiple classifiers prediction abilities for better performance. Experimental work is performed on Cleveland dataset using 14 attributes which is available online on UCI Repository. The results showed that the Bagging ensemble method is performed better to achieve higher accuracy of 87.78 %.