Comparison of selection operators for a genetic algorithm-based university course timetabling for the College of Science and Mathematics /
Jelin Celeridad Garsuta.
- 2006
- 78 leaves
Thesis (BS Applied Mathematics) -- University of the Philippines Mindanao, 2006
A comparison on selection operators of genetic algorithm (GA) applied in university course timetabling was conducted. GA is one of the meta-heuristic techniques used for combinatorial- and optimization-formulated problems such as timetabling. Based on Darwin's evolutionary theory, the natural selection or survival of the fittest, GA uses selection operators to identify good parents for producing better solution for the next generation. Four selection operators were compared to determine which selection operators is suitable for the GA- based College of Science and Mathematics, University of the Philippines in Mindanao course timetabling. The operators include the three ranking selections (rank-based, roulette wheel, and stochastic universal sampling) and the local selections (nine combinations of the ranking elections). For each operator, 30 trials at 1000 iterations were tested and comparisons were based on the obtained population fitness and terminating points. Among the ranking selections, the rank-based selection performed the least compared to roulette wheel selection and stochastic universal sampling. From the nine local selection operators, combinations with the roulette wheel selection as the pool selection operator performed better than the combinations with the rank-based selection as the pool selection operator. Overall, ranking selection operators performed better than local selection operators. Results were confirmed through statistical analysis. However, further experiments on other selection operators should be considered to test for better results