Sentiment Analysis and Classification of User Reviews on the Redbus Application Using Logistic Regression And SVM

Abstract

The increasing number of RedBus users in Indonesia has led to a growing volume of user reviews on digital platforms, especially the Google Play Store. These reviews reflect user perceptions and are valuable for sentiment analysis. This study aims to classify sentiments in RedBus user reviews using Logistic Regression and Support Vector Machine (SVM) algorithms. A total of 2,000 reviews were collected through automated web scraping and labelled using a lexicon-based approach. The data underwent preprocessing steps including normalisation, tokenisation, filtering, stemming, and labelling. Features were transformed using the TF-IDF method and split into 90% training and 10% testing sets. Evaluation results showed that SVM with a linear kernel outperformed Logistic Regression, achieving 91.10% accuracy and more balanced F1-scores across sentiment classes. Logistic Regression reached 86.39% accuracy but performed lower on positive sentiment. A paired t-test confirmed the statistical significance of the performance difference (p = 0.0005). These findings suggest that SVM is more effective in handling high-dimensional text data and can be recommended for real-world sentiment classification tasks, such as filtering negative reviews and improving customer service.