Simulated annealing and genetic algorithms for exam scheduling. (c1997)

Abstract

Scheduling final exams for large numbers of courses and students in Liberal Arts Universities is an intractable problem. If scheduling is done manually by the University Registrar's Office, a huge number of complaints will be raised by students about the unfairness of the schedule. Unfairness refers to conflicting simultaneous exams, consecutive exams, or giving more than two exams on the same day. In this work, we propose a simulated annealing algorithm and a classical genetic algorithm for exam scheduling. The experimental results, using realistic data, show that simulated annealing and genetic algorithms give better exam schedules than both manual scheduling and another heuristic method, in a reasonable time. Further, these algorithms are flexible and allow giving multi-section exams at the same time and pre-selecting certain exams.