Data Driven University Capacity Analysis via Mathematical Optimization

Aygul, Ozge

Etd

Data Driven University Capacity Analysis via Mathematical Optimization

Public Deposited

We propose a framework that uses advanced analytics to inform university course scheduling decisions. The analytics include mathematical optimization, data analysis, prediction, and algorithms to transform data into decisions that optimize the use of scarce capacity. We begin with a baseline formulation that addresses the scheduling of course sections with time patterns into classroom spaces while considering instructor preferences, as well as constraints related to conflicts and capacity. We then demonstrate some important extensions to this feasibility system that can split courses to address overflows while prioritizing classroom utilization and honoring instructor preferences. This dynamic splitting model enables determining optimal room allocations for a semester, providing valuable insights into the efficiency of current and future schedules. Additionally, we emphasize the importance of incorporating instructor preferences into the objective function for real-world timetabling applications. While individual objective functions can be optimized separately, we recognize the significance of considering combinations of objectives that align with the priorities of multiple stakeholders. To this end, we propose a hierarchical multi-objective framework that integrates the priorities of multiple decision makers, facilitating informed decision making in academic scheduling. To validate our framework, we conduct various experiments with different university expansion scenarios. Through these experiments, we examine the tradeoffs between objectives and discuss the implications of our findings. Our results demonstrate the effectiveness of utilizing advanced analytics to support strategic long-term decision-making within the context of university scheduling.

Creator