Sentiment Analysis of “Hepatitis of Unknown Origin” on Social Media Using Machine Learning

Abstract

Hepatitis of Unknown Origin worries civil society around the world. The public's fear of the emergence of the mysterious hepatitis disease caused panic and expressed views and criticisms and poured them on social media Twitter. The government's policy in considering the spread of unknown hepatitis can be taken from the public's view, especially on social media, which accommodates various public sentiments. This study aims to analyze public sentiment on “hepatitis of the unknown origin” using the Twitter dataset. We conducted sentiment analysis using various machine learning algorithms to get the best model accuracy in analyzing public sentiment against “hepatitis of unknown origin”. The algorithms used in this research are Random Forest, Extra Tree Classifier, Decision Tree, Naïve Bayes, Support Vector Machine, and XGBoost Classifier. We tested the model's accuracy against the data set using 20-fold cross-validation. The results showed that the public discussed the spread of unknown hepatitis to children, with each algorithm detecting a greater number of negative sentiments than neutral and positive sentiments. In addition, the results of the comparison of the Decision Tree algorithm obtained a model accuracy of 81.92%%, Random Forest 78.99%, Extra Tree Classifier 79.82%, Naïve Bayes 77.79%, Support Vector Machine 75.52%, and XGBoost Classifier 71.49%. We conclude that the Decision Tree is the best model for analyzing public sentiment about “Hepatitis of Unknown Origin”.