A Multi-Staged Algorithmic Process for the Solution of the Examination Timetabling Problem

We present an approach for the examination timetabling problem as defined in the second International Timetabling Competition (http://www.cs.qub.ac.uk/itc2007). The solution approach can be considered as an implementation of the GRASP (Greedy Randomized Adaptive Search Procedure) method with the combination of several other metaheuristics. Three stages are employed. The first stage is responsible for the construction of a relatively high quality feasible solution while the second stage improves it using simulated annealing local search. The final stage uses mathematical programming and analyzes each examination period in isolation proposing movements of exams to other rooms resulting in further improvement of the solution quality. The procedure produces feasible solutions for each dataset provided under the runtime limit imposed by the competition’s rules. Results are presented and analyzed.