A Proposed Weighted k-Medoids Clustering Method and its Application on Improving Higher Education Quality

Traditional methods of improving higher education quality rely on one-size-fits-all standards or use conventional analysis, such as teachers' classroom observations, feedback, student evaluations and questionnaire surveys. In recent years, clustering methods such as PAM and fast k-Medoids are widely used in improving higher education quality. But, due to the lack of consideration for the importance of each attribute in traditional methods, the results were poor. We improved the accuracy by weighting each attribute, firstly we propose a novel weighted k-Medoids clustering method by defining coarse entropy and attribute weight. Secondly, we perform the proposed method on UCI datasets to significantly demonstrates its validity comparing with traditional clustering algorithms of PAM and fast k-Medoids. Lastly, we apply the proposed method on educational datasets to improve higher education quality, and the results show that this new method accurately and effectively categorizes students' learning situations, enabling teachers to quantitatively evaluate students' learning outcomes and improve teaching strategies and methods based on different student categories and course characteristics, thereby enhancing teaching effectiveness and ultimately improving educational quality.