Enhancing Academic Pathways: A Data-Driven Approach to Reducing Curriculum Complexity and Improving Graduation Rates in Higher Education

Curriculum structure and prerequisite complexity significantly influence student progression and graduation rates. Thus, efforts to find suitable measures to reduce curriculum complexity have recently been employed to the utmost. Most of these efforts use the services of domain experts, such as faculty and student affairs staff. However, it is tedious for a domain expert to study and analyze a full curriculum in an attempt to reform its structure, given all the complexities associated with its prerequisite dependencies and learning outcomes. Things can become even more complicated when a set of curricula is examined. Therefore, efforts to automate the process of restructuring curricula are beneficial to helping the university community find the best available practices to reduce the complexity of their institutional curricula. This study introduces an innovative framework for automating curriculum restructuring, employing a combination of graphical models and machine learning techniques. In particular, we use latent tree graphical models and collaborative filtering to induce curriculum reforms without needing a domain expert. The approach used in this paper is data-driven, where actual student data and actual university curricula are utilized. Five thousand seventy-three student records from the University of New Mexico (UNM) are used for this purpose. Results demonstrate the restructuring impact on an engineering curriculum, particularly the computer engineering program at UNM. The effect is an improvement in the graduation rates of the students attending the revised engineering programs. These results are validated using a Markov Decision Processes (MDP) model. Furthermore, the findings of this paper showcase the practical benefits of our approach and offer valuable insight for future advancements in curriculum restructuring methodologies.