Exam Timetabling via Integer Programming: A case study at UTDT

Abstract

This work tackles the scheduling of midterm exams at Universidad Torcuato Di Tella (UTDT). Given a set of exams as well as the students enrolled therein, the problem involves determining the date, time and classrooms for each exam. Operational constraints such as conflicts of time, room capacities and availability, as well spacing (i.e., the number of days) between exams for a student make this problem very difficult from an optimization standpoint, in particular when considering real-world instances. We explore three Integer Linear Programming based algorithms for scheduling midterm exams within this context. Based on instances obtained from real data, the methods showed to provide good quality solutions, providing the students a balanced schedule for the exams over the planning horizon.