<![CDATA[In the digital era, user reviews on application platforms play a crucial role in evaluating service quality and customer satisfaction. This study aims to compare two sentiment analysis methods, namely Naive Bayes and Support Vector Machine (SVM), in classifying the sentiment of Ferizy app reviews on PlayStore into positive, negative, and neutral categories. Naive Bayes, known for its simplicity, efficiency on small datasets, and fast training, is compared to SVM, which is recognized for its high performance on complex data with non-linear distributions and its flexibility in kernel usage. This study also evaluates the performance of both methods based on accuracy, precision, recall, and F1-score metrics, particularly in handling class imbalance and noise in the data. The dataset consists of user reviews of the Ferizy application, which are analyzed to identify sentiment patterns and trends. The implementation results show that Naive Bayes achieves an accuracy of 79.27%, while SVM reaches an accuracy of 82.62%. This difference indicates that SVM is superior in handling more complex patterns in review data, although the margin is relatively small. The findings also reveal significant differences between the two methods, particularly in sentiment classification accuracy. Factors such as language complexity, class imbalance, and algorithm parameter selection are found to influence the performance of each method. This study provides valuable insights for application developers to improve service quality based on user sentiment analysis. Additionally, the results are expected to contribute to the development of more advanced and targeted sentiment analysis strategies, particularly in the digital transportation domain.Keyword: Analisis Sentimen; Naïve Bayes; Support Vector Machine; Ferizy; Ulasan]]>
Copyrights © 2025
