The Performance of Machine Learning Model Bernoulli Naïve Bayes, Support Vector Machine, and Logistic Regression on COVID-19 in Indonesia using Sentiment Analysis

Abstract

The COVID-19 pandemic has significantly impacted Indonesia, necessitating a deeper understanding of public sentiment towards the crisis. This study investigates the performance of three prominent machine learning models: Bernoulli Naïve Bayes, Support Vector Machine (SVM), and Logistic Regression, in analyzing sentiments related to COVID-19 in Indonesia. Utilizing a dataset comprising social media posts, the research aims to classify sentiments into positive, and negative categories, providing insights into the public's perception of the pandemic and associated measures. Sentiment analysis serves as a powerful tool to capture the collective emotions and opinions of the populace, which are pivotal in shaping public health responses and policies. The accuracy of LR and SVM is 99%, whereas Bayesian has an accuracy of 98%. We conclude that Logistic Regression and Support Vector Machine are the best model for the above dataset. This research evaluates these models' accuracy and reliability in the context of the Indonesian language, which influence sentiment interpretation. The findings of this study will contribute to the fields of natural language processing and public health by highlighting the efficacy of machine learning models in sentiment analysis during a health crisis. Moreover, the results will assist policymakers and health officials in understanding public sentiment, enabling them to tailor communication and interventions more effectively.