Performance Evaluation of the Supervised Machine Learning Algorithms Using R

Abstract

Today, big data occupies a significant place in each one’s life. Various supervised and unsupervised machine learning algorithms exist which can be applied on this big data to get the desired information. A wide application of the classification algorithms is seen in analyzing the data of medicine or health where it has been useful in predicting the disease based on the given features. In this paper, the Cleveland database from the dataset of heart disease publicly available on UCI machine learning Web site was taken, and three different classification algorithms were applied on it to predict the presence or absence of heart disease in patients. The performance of the three well-known algorithms—support vector machine, decision tree and random forest—has been compared based on various metrics accuracy, sensitivity, and specificity and error rate. The experiment has been performed on R language, a programming language which is widely used for data visualization, and the result shows that random forest algorithm outperforms the rest of the two algorithms in successfully classifying healthy and unhealthy patients. For this experiment, fourteen different features like age, resting electrocardiographic results, resting blood pressure, chest pain type, etc., of an individual have been considered.