Sentiment Classification of MyTelkomsel Reviews Using SVM and Logistic Regression

Abstract

The development of digital technology has encouraged increased user participation in expressing opinions through review platforms, such as the Google Play Store. MyTelkomsel's application, a digital service from Indonesia's leading telecommunications provider, has received various responses, from appreciation to complaints related to app performance and customer service. This study aims to evaluate sentiment in user reviews using Support Vector Machine (SVM) and Logistic Regression algorithms. Data was collected from the Google Play Store and underwent a series of pre-processing stages, including data cleaning, case folding, normalization, tokenization, stopword removal, and stemming. The feature extraction process uses the TF-IDF approach, while model performance evaluation is based on accuracy, precision, recall, F1-score, and Area Under Curve (AUC) metrics. The results showed that the performance of both models was relatively balanced, but SVM exhibited an advantage in recall for positive sentiment (82%), accuracy (93.36%), and AUC (0.9680). Logistic Regression excels in precision (99%) in the positive class. WordCloud visualization illustrates consistency of dominant words in each sentiment class, reflecting the model's ability to identify patterns in user opinion. These findings are expected to contribute to the improvement of digital services based on user input.