Comparison of Decision Tree and Linear Regression Algorithms in the Case of Spread Prediction of COVID-19 in Indonesia

Abstract

COVID-19 is a disease that was first discovered in Wuhan, China and caused the 2019-2020 coronavirus pandemic. This virus can cause respiratory tract infections such as flu when infecting humans. According to Ministry of Health of the Republic of Indonesia, the number of confirmed cases of COVID-19 in Indonesia at March 2021 is 1,511,712 with 40,858 deaths and 1,348,330 recovered. For that, Indonesia is declared to have the highest confirmed cases in ASEAN. Several studies have been carried out to handle some cases by using the data mining techniques such as Decision Tree or Linear Regression algorithm, as example to classify the respiratory diseases and predict pregnancy hypertension. In this study, we tried to analyze COVID-19 cases in Indonesia and conducted an experiment of predicting COVID-19 new cases with the Decision Tree (CART) and Linear Regression algorithms. Then we will compare the values of these two algorithms by using R2 Score to evaluate the prediction performance. The results of this analysis state that DKI Jakarta province has the highest number of positive cases, cures and deaths in Indonesia. The value of the comparison results from the R2 Score obtained in the Decision Tree algorithm reached 95.69% (training) and 92.15% (testing) while the Linear Regression algorithm reached 79.93% (training) and 77.25% (testing).