Interpretation of Student Responses in Teacher Evaluations: A Comparative Cluster Analysis Approach

Abstract

This study interprets student responses regarding teacher evaluations using advanced cluster analysis techniques. The responses were clustered using the K-Means and HDBSCAN algorithm from the Data Science GPT [Large language model]. Fifteen main features influencing teacher evaluations were identified, and their relationships were visualized using bar charts and heatmaps to illustrate cluster overlaps. The analysis compares traditional K‑Means clustering with Hierarchical Density-Based Spatial Clustering (HDBSCAN), highlighting the benefits of density-based clustering in capturing nuanced insights. These findings provide actionable recommendations for enhancing teaching quality and student satisfaction in higher education.