This paper presents an investigation of a combination of graph coloring heuristics in construction approach in University course timetabling problem (UCTP) to produce a population of initial solutions. The graph coloring heuristics were set as individual, combination of two heuristics and combination of three heuristics. In addition, several steps of courses assignment were applied to all the settings. All settings of heuristics are then tested on the same curriculum-based problem instances and are compared with each other in terms of number of population produced. The results can be used for improvement phase which are the second phase of UCTP. This approach allows generalization over a set of problems instead of producing feasible timetables for some of the problems only. Future work will use the best settings of heuristic to the improvement phase in population-based improvement algorithm.
展开▼
