Finding the Best Fit; Comparative Analysis of Supervised Learning Models for Student Feedback Classification

This study developed a machine learning-based system to analyze student feedback on the quality of university courses, aiming to streamline the feedback process and enhance instructional delivery. Using a dataset collected from a Zimbabwean university Political Philosophy students, feedback was gathered through Google Forms, formatted as CSV files, and processed through Natural Language Processing (NLP) techniques to clean and tokenize the text. Three machine learning models—Support Vector Machine (SVM), Logistic Regression (LR), and Multinomial Naive Bayes (MNB)—alongside deep learning models like BERT and CNN, were tested to identify the most effective classifier. SVM was selected after hyperparameter tuning, achieving a reliable 80% accuracy in sentiment classification. The final system generates a PDF-based Course Evaluation Summary Report that presents positive and negative feedback, key sentiment insights, and graphical summaries, offering instructors a clear, actionable overview of student perspectives. While deep learning models struggled with the dataset size, SVM proved optimal for small data contexts typical in university settings. Future work will focus on scaling the system with advanced NLP and larger datasets to broaden its applicability.