Mining Course Groupings from Student Performance
Abstract
Educational institutions generate and maintain detailed electronic records that describe academic coursework and associated student performance. This study mines such data from a major university in order to identify relationships between courses based on patterns in student performance. Two courses are considered similar if students taking both courses perform similarly in each. Once the course similarities are computed, graphs are generated where the nodes represent the courses and the edges represent the degree of similarity between the courses. Graph mining algorithms are then applied to identify cliques (groups of courses that behave similarly with respect to student performance) and courses that exhibit a high degree of centrality (i.e., are connected to many other courses). Visualizations of the graphs also show more specific information about the relationships between courses. At a more granular level, course pairs with high degrees of grade correlation are presented and discussed. Aggregate behavior for groups of courses, such as all courses in a given discipline/major, is also examined, and the differences are discussed. As far as we know, no prior study using this notion of course similarity has been conducted.
Subject Area
Computer science|Higher education
Recommended Citation
Zhang, Tianyi, "Mining Course Groupings from Student Performance" (2020). ETD Collection for Fordham University. AAI27964975.
https://research.library.fordham.edu/dissertations/AAI27964975
