ANALISIS SENTIMEN MAHASISWA TERHADAP SIASAT UKSW BERDASARKAN KUESIONER MENGGUNAKAN METODE LOGISTIC REGRESSION

Abstract

SIASAT is the primary academic information system at Satya Wacana Christian University (UKSW) used by students for academic administrative activities. This study aims to analyze students’ sentiment toward SIASAT using a Logistic Regression classifier based on open-ended responses collected through an online questionnaire. A total of 58 textual responses were manually labeled into two classes, namely positive and negative. The research stages include text preprocessing (lowercasing, cleaning, tokenization, stopword removal, and stemming), feature weighting using TF-IDF, handling class imbalance through Random Oversampling, and classification using Logistic Regression. Evaluation was performed using Stratified 5-Fold Cross-Validation, with oversampling applied exclusively to the training data within each fold to prevent data leakage. The model achieved an average accuracy of 70% and a weighted F1-score of 70%. These results suggest that Logistic Regression provides a promising baseline performance for Indonesian text sentiment classification on a small-scale dataset, although its generalizability remains limited due to the small sample size and the manual labeling process.