Comparison of Random Forest and Support Vector Machine Methods in Sentiment Analysis of Student Satisfaction Questionnaire Comments at ITB STIKOM Bali

ITB STIKOM Bali is one of the higher education institutions in Bali that focuses on academic activities, particularly in the field of Information Technology. To maintain its educational quality, the Quality Assurance Department collaborates with the Center for Information and Communication (Puskom) to distribute a student satisfaction questionnaire at the end of each semester. In evaluating student satisfaction with campus facilities, the comment section is one of the key indicators, featuring the question: “Based on your experience, please describe which AAK services you found disappointing and in need of improvement.” This study compares the performance of the Random Forest and Support Vector Machine (SVM) methods in conducting sentiment analysis on historical student satisfaction comments. The research involved several stages, including literature review, data collection, preprocessing, transformation, data mining, evaluation, and visualization. The results demonstrate strong accuracy, precision, recall, and F1-scores for both methods using an 80:20 data split. Before applying the SMOTE technique, the best result was achieved by the Support Vector Machine method with a score of 0.90, while the Random Forest method yielded an accuracy of 0.81, precision of 0.85, recall of 0.81, and F1-score of 0.76. After applying SMOTE, both methods achieved an improved and equal score of 0.90. The study also produced an excellent classification result based on the ROC curve. It is expected that this research can serve as an additional reference for the assessment of student satisfaction at ITB STIKOM Bali at the end of each academic semester.