Sentiment Classification of Livin by Mandiri App Reviews with Support Vector Machine Based on Modified Particle Swarm Optimization
Salsa Febriliana Sandita, Iut Tri Utami, Masithoh Yessi Rochayani · INTERNATIONAL JOURNAL OF MATHEMATICS AND COMPUTER RESEARCH · 2025
This research aims to evaluate the quality of the Livin by Mandiri mobile banking application based on user reviews from the Google Play Store. A Support Vector Machine (SVM) classifier is applied to distinguish between positive and negative sentiments. Modified Particle Swarm Optimization (MPSO) is used for parameter tuning because SVM performance is sensitive to parameter selection. The baseline SVM attains 90.30% accuracy, 90.64% precision, 89.86% recall, and 90.25% F1-score with a linear kernel with C = 1 and a 90:10 training-testing split. Accuracy is increased to 91.33% by the SVM-MPSO model, with precision of 90.87%, recall of 91.71%, and F1-Score of 91.29%. The Streamlit framework for interactive sentiment classification is used to deploy the model.