Optimization support for senior design project assignments

Each year, the College of Engineering at the University of Arizona offers senior design projects to approximately 300 seniors. Sponsors provide the university with potential student projects; three to six students from various disciplines must collaborate on each project. Students give the university a rank ordering of their preferred projects. The program manager then selects a subset of the projects, assigns students to projects according to their preferences, and manages other requirements, such as matching sponsor-requested skills and minimizing deviations in team average GPAs. The task of assigning students to projects used to consume untold hours and Post-it notes as the program manager tried to solve the problem manually. Invariably, some students were unhappy because they did not understand why they had been assigned to less-desirable projects. In this paper, we discuss a pragmatic mixed-integer program (MIP) to optimize the solution to this problem. We cover both the modeling challenges and all aspects from design to implementation that were necessary to ensure the system's successful deployment; these include testing, interaction with business processes, user interfaces, and integration. We report the practical benefits the university derived from using our system, and summarize some of the lessons, which we learned as part of our development life cycle, that may be applicable to the development of MIP-based decision-support systems.