<![CDATA[The advancement of digital technology has encouraged the increasing use of online learning applications such as Ruangguru, while simultaneously fostering various innovations in the field of education. Ruangguru, as one of the most popular educational applications in Indonesia, receives thousands of user comments that can be analyzed to reflect user satisfaction and perception. This study aims to automatically classify user comments based on the sentiments they contain using the Naïve Bayes Classifier algorithm. This approach is expected to help Ruangguru developers better understand user needs and preferences, thereby improving service quality. The dataset was obtained from the Google Play Store platform, consisting of approximately 5,000 comments collected during the period from October 28 to December 31, using the google-play-scraper tool. The application of the Multinomial Naïve Bayes algorithm with TF-IDF weighting was employed to analyze the data, resulting in four sentiment categories: Baik Sekali, Baik, Cukup Baik, and Kurang Baik. Evaluation of the model was conducted using accuracy, precision, recall, and F1-score metrics. With an accuracy rate of 84.83%, the model correctly predicted the actual labels in approximately 85% of the test data. The model also achieved an F1-score of 85%, a precision of 86%, and a recall of 85%. The classification results revealed that the “Baik” category dominated with a proportion of 28.3%, followed by “Baik Sekali” at 24.3%, “Cukup Baik” at 24.0%, and “Kurang Baik” at 23.4%. These findings indicate that the model maintains a reasonable balance between sensitivity and accuracy in sentiment classification. Therefore, the Naïve Bayes Classifier method is capable of automatically identifying user opinions and has the potential to serve as a valuable tool in sentiment analysis for online learning services.]]>
