Comparison of Naïve Bayes Algorithm, Support Vector Machine and Decision Tree in Analyzing Public Opinion on COVID-19 Vaccination in Indonesia

Abstract

The spread of COVID-19 in Indonesia has caused many negative impacts. Therefore, the government is taking vaccination measures to suppress the spread of COVID-19. Public response to vaccinations on Twitter has been mixed, with some supporting it and some not. The data for this study comes from the Twitter feed of the drone portal Emprit Academy (dea). Classification is performed using SVM, decision tree and Naive Bayes algorithm. The purpose of this study is to inform the public about whether vaccination against COVID-19 is inclined toward positive, neutral, or negative opinions. Moreover, this study compares the accuracy of the three algorithms used, namely Naive Bayes (NB), Support Vector Machine (SVM) and Decision Tree, and the validation performed using the K-Fold Cross-Validation method, AdaBoost feature selection, and the TF-IDF Transformer feature extraction test. The result obtained from this study is that the accuracy of the 90:10 data keeps improving, dividing by 82.86% on the SVM algorithm, 81.43% on the Naive Bayes and 78.57% on the decision tree.