Electric Vehicle Sentiment Analysis Using a Comparison of Naïve Bayes and Support Vector Machine

Abstract

The development of electric vehicles in Indonesia has sparked various opinions from the public, which are often shared on social media, especially X. These opinions need to be analyzed to understand how the public views the policies and implementation of environmentally friendly vehicles. This study aims to examine public sentiment toward electric vehicles by comparing two types of text classification algorithms, namely Naïve Bayes and Support Vector Machine (SVM), using the Term Frequency–Inverse Document Frequency (TF-IDF) approach. The data used is Indonesian-language tweets collected through a crawling process, which then undergoes several pre-processing stages such as cleaning, case folding, normalization, tokenizing, stopword removal, and stemming. After that, the data was labeled for sentiment into three categories: positive, negative, and neutral, before being processed using a classification algorithm. To evaluate the model's performance, a confusion matrix was used, which shows the algorithm's performance based on accuracy, precision, recall, and F1-score values. The research results show that the Naïve Bayes algorithm has better results with an accuracy of 92%, while SVM achieves an accuracy of 76%. Therefore, the Naïve Bayes algorithm is considered more suitable for analyzing the sentiment of tweets related to electric vehicles in Indonesia.