Feature Scaled Element Balancing with Random Boosting for Heart Disease Prediction using Machine Learning

Abstract

In the current scenario, the researchers are focusing towards health care project for the prediction of the disease and its type. In addition to the prediction, there exists a need to find the influencing parameter that directly related to the disease prediction. The analysis of the parameters needed to the prediction of the disease still remains a challenging issue. With this view, we focus on predicting the heart disease by applying the dataset with boosting the parameters of the dataset. The heart disease data set extracted from UCI Machine Learning Repository is used for implementation. The anaconda Navigator IDE along with Spyder is used for implementing the Python code. Our contribution is folded is folded in three ways. First, the data preprocessing is done and the attribute relationship is identified by the correlation values. Second, the data set is fitted to random boost regressor and the important features are identified. Third, the dataset is feature scaled reduced and then fitted to random forest classifier, decision tree classifier, Naïve bayes classifier, logistic regression classifier, kernel support vector machine and KNN classifier. Fourth, the dataset is reduced with principal component analysis with five components and then fitted to the above mentioned classifiers. Fifth, the performance of the classifiers is analyzed with the metrics like accuracy, recall, fscore and precision. Experimental results shows that, the Naïve bayes classifier is more effective with the precision, Recall and Fscore of 0.89 without random boost, 0.88 with random boosting and 0.90 with principal component analysis. Experimental results show, the Naïve bayes classifier is more effective with the accuracy of 89% without random boost, 90% with random boosting and 91% with principal component analysis.