A Clustering-Based Approach for Ranking Universities
George Matlis, Nikos Dimokas, Petros S. Karvelis · 2023
In recent years, there has been an increasing interest among international organizations and academic communities in the evaluation and ranking of educational institutions. Such rankings are of great significance to various stakeholders, including students and their families, academic staff, funding agencies, and universities themselves. Ranking methodologies typically involve the use of diverse indicator categories, aimed at ensuring objec-tivity and reflecting the relative importance of each university. These categories may include, but are not limited to, research output and impact, teaching quality and learning environment, international diversity, and reputation in the employment market. Apart from conventional ranking methods, clustering-based techniques can also be used to group universities according to their common characteristics and to evaluate their performance within each cluster. In this paper, we seek to explore how universities from various regions can be clustered into distinct groups and ranked based on their unique features. For this purpose, we employ various clustering algorithms, each possessing distinct characteristics, and we present the similarities between them to obtain the most effective evaluation. We also analyze how university features can impact the evaluation and classification process into a cluster. We present some statistical results of university rankings and finally, we visualize the clustering process to facilitate a deeper understanding of the findings.