In early August 2022, Indonesia's Investment Minister, Mr.Bahlil Lahadalia, announced that the skyrocketing fuel prices have caused a national increase in fuel prices, and the difference between the current price and the one listed in the APBN has sparked discussions on removing fuel subsidies to save on the state budget. This has generated both pros and cons among the public, and one of the mediums used to express opinions on this matter is Twitter. Therefore, in this study, the author used the Bernoulli Naive Bayes Classifier classification method, which began with data collection using the Tweepy Library combined with the Twitter API, followed by data preprocessing to, then formed a text classification system with the help of N-Gram TF-IDF feature extraction, and implemented the Bernoulli Naive Bayes Classifier algorithm with the help of the Sklearn Library. With a total of 710 tweet data, divided into 80% training data and 20% test data, and passed through 3 testing scenarios (Unigram, Bigram, and Trigram), the resulting accuracies are 0.69, 0.57, and 0.57, respectively. Thus, it has been proven that the use of Unigram N-Gram TF-IDF combined with the Bernoulli Naive Bayes produces better results, particularly for sentiment analysis research.
Copyrights © 2023