Sentiment Analysis of Reviews for the Green Sm Online Taxi App in Indonesia using Machine Learning Algorithms

Green SM is an electric vehicle-based online taxi service that has begun operating in Indonesia and needs to be evaluated based on user experience. This study aims to map the sentiment of user reviews of the Green SM app and compare the performance of the Support Vector Machine (SVM), Naïve Bayes Classifier (NBC), and Random Forest. The dataset consists of 1,616 Indonesian-language reviews collected from the Google Play Store for the period 2024–2026. The research process included text cleaning, case folding, tokenization, normalization, stopword removal, stemming, lexicon-based labeling, TF-IDF feature extraction, and splitting the data into training and testing sets at an 80:20 ratio. Evaluation was conducted using accuracy, precision, recall, F1-score, and the confusion matrix. Negative sentiment dominated at 55.3%, while positive sentiment accounted for 44.7%. Random Forest achieved the highest accuracy at 88.36%, with a precision of 0.927, a recall of 0.803, and an F1-score of 0.860. SVM achieved an accuracy of 87.42%, while NBC achieved 87.11%. These results indicate that Random Forest provided the best overall performance on the dataset used, while the dominance of negative sentiment can serve as input for service evaluation. Keywords: Green SM; Machine learning; Naïve Bayes; Random Forest; Sentiment analysis