Classification of Heart Disease: Comparative Analysis using KNN, Random Forest, Gaussian Naive Bayes, XGBoost, SVM, Decision Tree, and Logistic Regression
Yoshiven Boer, Lianca Valencia, Melisa Rachel Setiadi, Karli Eka Setiawan, Muhammad Fikri Hasani
Abstract
The heart is the center of the human body, and when it is not functioning well, the body cannot operate effectively. This research aims to compare some of the classifier models to find the best model for classifying heart disease by its symptoms for early detection, so users will be able to know whether their symptoms belong to a type of heart disease that requires them to undergo a check-up or not to carry out early heart disease detection. The classifier models used in this research are K-NN, Random Forest, Naive Bayes, XGBoost, SVM, Decision Tree, and Logistic Regression algorithms. As for the results, the Logistic Regression model shows the best result with its metrics on average being 92.61%, and the second best result was achieved by Gaussian Naive Bayes with its metrics on average being 90.78%, while the other models results were below 90%. So, this research shows that Logistic Regression is the best model to classify heart disease for early detection because of its higher results than other classifier models.