Machine learning methods for predicting cardiovascular diseases analyzing a hybrid dataset

Cardiovascular disease (CVD) remains one of the world’s leading causes of death. Early diagnosis is one of the first and most important treatment steps to prevent fatal cardiovascular diseases. To this end, the data analysis is the most important step to prevent the death from the CVD. In this study, we have been considered the machine learning techniques for the early diagnosis of the cardiovascular diseases. A hybrid dataset called Sathvi which combined The Hungarian, the Switzerland, the Cleveland, and the Long Beach datasets has been analysed using Naive Bayes (NB), Logistic Regression (LR), k-nearest neighbour (k-NN) and Random Forest Classifier(RFC) machine learning techniques. The Sathvi dataset has 531 instances and 12 attributes. There is no missing value in the Sathvi dataset. We have been obtained the 85.63%, 81.68%, 80.00% and 77.50% accuracy value using the NB, LR, k-NN and RFC machine learning techniques respectively. We have also obtained close accuracy values of the Naive Bayes method's applying the other machine learning methods considered in this study. We can say that the precision, recall and F -score values and the ROC and AUC score scaled are very good agreement with the accuracy values of the aforesaid methods in this study. The AUC values of the NB, LR, k-NN and RFC machine learning techniques are 84.29%, 80.74%, 79.91% and 76.00% respectively. The comparison of the some attributes has been given with the pair plots.