Exploring Learning Patterns: A Review of Clustering in Data-Driven Pedagogy

Abstract

Educational institutes amass and retain extensive amounts of data including records of student attendance, test scores, exam results, and performance statistics. Extracting insights from this data can provide valuable information to educators and policymakers. The rapid expansion of educational data underscores the need for sophisticated algorithms to process such vast quantities of information. This challenge led to the emergence of the field of educational data mining (EDM). Clustering is a popular approach within EDM that can find hidden patterns in data. Numerous studies in EDM have concentrated on applying diverse clustering algorithms to educational attributes. This paper presents a comprehensive literature review focusing on 43 papers spanning between 2013 to 2023 on the use of clustering algorithms and their effectiveness within the realm of EDM. The review indicates that K-means clustering has been utilized extensively in the reviewed literature with 29 of the 43 reviewed papers using K-means clustering in their analysis. It was also uncovered that cluster-based analysis majorly focuses on analyzing student performance in a course or in a degree program closely followed by clustering students based on class of learners. Insights are deduced from the reviewed literature highlighting the focus of current research and potential directions for the future.