Feature-based tuning of simulated annealing applied to the curriculum-based course timetabling problem

We consider the university course timetabling problem, which is one of the most studied problems in educational timetabling. In particular, we focus our attention on the formulation known as curriculum-based course timetabling problem (CB-CTT), which has been tackled by many researchers and has many available benchmarks. The contributions of this paper are twofold. On the one side, we propose an effective and robust single-stage simulated annealing search method for solving the problem. On the other side, we design and apply an extensive and statistically-principled analysis methodology for the algorithm parameter tuning procedure. The outcome of this analysis is a linear regression model between instance features and search method parameters, that allows us to set the parameters for unseen instances on the basis of a simple inspection of the instance itself. Using this method, our algorithm, despite its apparent simplicity, has been able to achieve very good experimental results on a set of popular benchmark testbeds. Finally, we propose many new, real-world instances, that could be used as ground for future comparisons.