Analyzing Public Opinion on Electrical Vehicles in Indonesia Using Sentiment Analysis and Topic Modeling

Abstract

Electrical vehicles (EVs) are one of the solutions to tackle the issues of greenhouse gas emissions and climate change in the world. In Indonesia, the government has made regulations supporting the implementation of EVs through various incentive programs and infrastructure developments, which are expected to increase public interest in the use of EVs. However, there are still many pros and cons found in the use of EVs in Indonesia, especially in social media. In this paper, we discuss the implementation of sentiment analysis models through social media, Twitter. It uses supervised learning methods, such as Support Vector Machine (SVM), Logistic Regression, Random Forest, Gradient Boosting Algorithm, Convolutional Neural Networks (CNN), and Recurrent Neural Networks (RNN). The total data used is 7102 tweets with 2847 tweet samples to become labeling data. The results of the analysis are as many as 1586 tweets (55,71%) responded positively and 1261 (44,29%) responded negatively to EVs. SVM is the best model with 75.08% accuracy and the most topics that support EVs to appear were the temporary G20 activities and the benefit of EVs with positive support of tweets. And others tend to prioritize primary needs than own EVs. We utilize Latent Dirichlet Allocation (LDA) to examine topics related to EVs in Indonesia. Finally, this paper contributes to extending knowledge of sentiment methods from the discussion that sticks out on social media, and suitable techniques for conducting research related to sentiment analysis as well as topics of discussion that are closely related to the issue of EVs.