PERFORMANCE COMPARISON OF TWITTER SENTIMENT ANALYSIS USING FASTTEXT SVM AND TF-IDF SVM: A CASE STUDY ON ELECTRIC MOTORCYCLES

Electric motorcycles are trending on Twitter as two-wheeled vehicles different from those using fossil fuels. Electric motorcycles rely on batteries charged using electricity. However, there are many opinions about electric motorcycles on social media, especially Twitter. Yet, tweets and comments on Twitter often contain irrelevant words that can affect sentiment analysis. In this study, sentiment analysis was conducted on 8,000 data from Twitter using FastText and TF-IDF as word embedding techniques, along with Support Vector Machine (SVM) as the classification technique. The aim of this research is to compare the performance of SVM using different feature extraction techniques, namely FastText and TF-IDF. The results of this study are expected to be beneficial for electric vehicle manufacturers and individuals interested in electric vehicles. In this comparison, the performance of TF-IDF and FastText feature extraction in sentiment classification with SVM will be evaluated. SVM performance is assessed based on accuracy, precision, recall, and F1-score for each feature extraction technique used. The test results show an average accuracy above 83%, with the highest values being 86% for accuracy, 79% for precision, 52% for recall, and 58% for F1-score.