Heart Disease Prediction and Its Important Parameters Analysis Using Machine Learning

In recent time the number of people dying and suffering from heart disease is very high also these cases are increasing day by day. In heart disease the diagnosis at early phase is very crucial. Machine learning can help in the prediction of heart diseases very accurately just by providing patient data as input. To have good accuracy of prediction the only requirement is that the learning dataset must be as big as possible. The purpose of this research is to compare the accuracies of classification models used to predict heart disease by using the largest dataset available for heart disease prediction, check out the heatmap to find the correlation between different parameters involved in the dataset and to find the most important parameter with importance factor for predicting the heart disease. Random Forest model is having the highest 92.44% accuracy followed by SVM and decision tree in predicting heart disease. Logistic regression had the least accuracy among all the models used. ST slope and oldpeak has very high correlation factor of 0.52 and exercise angina and max heart rate have very low correlation of -0.38. ST slope is the most important factor for predicting heart disease among all other factors given in the dataset and fasting blood sugar is the least important parameter used in prediction of heart disease.