Sentiment Analysis of GOJEK User Reviews Using SVM, Naïve Bayes, and Random Forest Models

Sentiment analysis has become a crucial technique for understanding user feedback and improving the quality of applications. This study focuses on the sentiment analysis of user reviews for the GOJEK application, a popular ride-hailing and on-demand service platform, available on the Google Play Store. The primary aim is to compare the effectiveness of three machine learning algorithms Support Vector Machine (SVM), Naïve Bayes, and Random Forest in accurately classifying user sentiments. The research employs a data driven approach, using web scraping to gather user reviews, followed by data preprocessing and feature extraction using the Term Frequency Inverse Document Frequency (TF-IDF) method with n-grams for vectorization. Model performance is evaluated using a confusion matrix, with an 80:20 data split for training and testing. The results show that SVM achieved the highest accuracy of 92.6%, followed by Naïve Bayes at 91.4% and Random Forest at 90.4%. Positive sentiment reviews highlighted the helpfulness and overall satisfaction with the app, while negative reviews pointed to concerns regarding slow performance. The findings demonstrate that sentiment analysis can effectively capture user satisfaction and pinpoint areas for app improvement. This study contributes to the optimization of GOJEK’s service by providing insights into user perceptions and guiding future developments, thus offering valuable implications for enhancing customer experience in similar mobile applications.