GradPlanner: A Deployed Multi-Objective Optimization-Based Decision Support Tool for Student Degree Planning and Timely Graduation

Timely graduation is a central goal for students and universities, yet many undergraduates still take longer than four years to complete their degrees. This paper focuses on three major contributors to delayed graduation in four-year programs: (1) unoptimized and poorly balanced multi-semester degree plans, (2) registration bottlenecks in required courses because of limited capacity, and (3) course failures that disrupt prerequisite progress. We formalize these challenges as the Student Graduation Planning Problem (SGPP) and develop the GradPlanner, a multi-objective optimization tool that generates personalized, feasible semester-by-semester plans. The SGPP is modeled as a staged mixed-integer linear program that first enforces feasibility and minimum-load intent, then minimizes time to graduation, and finally applies a Chebyshev-based refinement that balances similarity to the departmental four-year reference plan and difficulty load across semesters. The method is implemented as a web-based system integrated with a shared course and policy database and is deployed for five undergraduate programs in the Bellini College at the University of South Florida. Numerical experiments indicate that the lexicographic-Chebyshev pipeline is fast and stable enough for interactive advising. Applied to real student records, the planner both reduces graduation time and yields plans that align with the departmental four-year reference plan more closely while spreading difficult courses more evenly. Analysis of real student records indicates that the proposed optimization-based planning approach improves all three performance metrics, graduation timing by approximately 6%, curriculum alignment by about 54%, and workload balance by nearly 55%, on average, without requiring changes to existing advising workflows. The results further show that among on-track students (those expected to graduate within four years), the proposed method improves at least two of the three metrics for 56% of students. For delayed students, the method improves all three metrics for 72% of students.

History: This paper was refereed.

Supplemental Material: The online appendix is available at https://doi.org/10.1287/inte.2026.0325.