Prediction of Heart Disease Using Machine Learning Algorithms

Heart disease has become one of the alarming issues of death. It is accountable for fatty plaques in the arteries. If this fatal condition can be identified early, we can preserve many people’s arteries. Different types of supervised machine learning algorithms are applied in our research paper in order to predict heart disease existence in patient body. Besides this, we have focused on an efficient way to improve the performance of our applied classifiers. Imputing mean value technique is applied to handle null values present in our dataset. The features which are unnecessary are removed by using the info-gain feature selection technique. In order to calculate prediction accuracy, K-Nearest Neighbors (KNN), Naive Bayes and Random Forest are applied to the heart disease dataset. Accuracy, precision, recall, F1-score, and ROC are calculated which help us to compare the performance of the classification models. Handling null values on a particular column by imputing mean values of that column and our applied info-gain feature selection technique has aided us in improving the accuracy of our prediction models. Random Forest among all has given the best classification accuracy which is 95.63% with precision, recall, F1-score and ROC are 0.93, 0.92, 0.92 and 0.9, respectively.