Anita Qoiriah, Yuni Yamasari, Naim Rochmawati
Each course subject has a specific learning level that the students grasp, and sometimes they are shy to ask when they don't understand anything. To ensure that the quality of learning is improved, it is critical to assess the teaching and learning process using feedback from the students. Qualitative feedback allows students to provide answers freely. According to research, sentiment analysis with machine learning is the best way for performing sentiment analysis on student comments. The idea of this research is to use an appropriate machine learning method to do sentiment analysis of student feedback on lecture materials using a case study on Basic Programming courses, so that lecturers can build better strategies for lectures. The system for sentiment analysis of student feedback uses a support vector machine algorithm with TF-IDF and bi-grams for feature extraction. The best performance based on hyper-parameter tuning is obtained by the value of C = 1, using Kernel = linear and Gamma = 0.1. For the training process, using k-fold cross-validation with a value of K = 5 gives the best results. The evaluation results obtained the average value of precision: 90%, recall: 90%, f1-score: 90%, and accuracy: 90%. In the Basic Programming participant class, model trials produced findings where students struggled with the beginning material, notably flowcharts. Students struggled with input-output, conditional, string, and function materials as well. Students’ comprehension of looping, arrays, and files materials was excellent. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.
Informatics Department, Universitas Negeri Surabaya, Jln. Ketintang, Surabaya, Indonesia
