An Improved Ensemble Approach to Predict Cardiovascular Problems

Abstract

Cardiovascular problems have fully dominated the world of humankind. Millions of people in the world die every year because of diverse existing cardiovascular problems. Data mining techniques plays imperative role to curb the death rate because diverse cardiac prediction models are devised using different mining techniques/algorithms. In this research work five models are devised to attain the performance measures related to the heart diseases prediction problems which include decision tree, J48, neural network, naïve Bayes and support vector machine. Among these employed classifiers J48 and neural network (MLP) depicts enhanced performance measures in comparison to other employed techniques. This experimented approach employed ensemble methodology to combine the predictive power of the two acceptable techniques i.e. J48 and neural networks (MLP) using WEKA simulation tool. Ensemble algorithms are tallied among the topmost machine learning techniques to combine the prediction power from desired multiple techniques. Ensemble includes several techniques but we engaged bagging method of ensemble to devise the prediction models i.e. model 1 and model 2 using J48 and neural networks. Both the developed models are satisfactory to employ for the prediction of cardiovascular problems however model 1 conquers to model 2 for the prediction of cardiovascular problems.