Classification of Heart Disease Using Machine Learning Techniques

The most crucial task in the medical field is diagnosing an illness. If a disease is determined at the early stage then many lives can be saved. The purpose of this paper is to use the medical data to predict cardiovascular heart disease using both supervised and unsupervised learning techniques and to show the effects of feature correlation on the classification model with over four different algorithms namely, Logistic Regression, Naive Bayes, Random Forest and Artificial Neural Networks. For the performance assessment, it incorporates F1-score, precision, Area under curve and recall. Overall, Logistic Regression algorithm tends to perform well for both Hungary and Statlog dataset whereas for Cleveland dataset, Artificial Neural Networks performs better than Logistic Regression in terms of accuracy. In terms of area under curve score, Logistic Regression performance is higher in all the dataset compared to Naive Bayes, Random Forest and Artificial Neural Networks. The results tabulated evidently prove that the designed diagnostic system is capable of predicting the risk level of heart disease effectively when compared to other approaches