Course recommendation as graphical analysis

Connor Bridges, James Jared, Joshua Weissmann, Astrid Montanez-Garay, Jonathan C. Spencer, Christopher G. Brinton

Abstract: This work proposes a method for course recommendation using grade and enrollment data. We analyze the per-semester sequence in which courses are taken in order to create a personalized course transition graph that balances the student's current grades, their expected improvement, and course popularity. Using a dataset of 6000 students and 1500 courses, we compare the recommended trajectories of top performing and low performing students to show that popularity alone is a strong heuristic for recommending successful trajectories.