Application of genetic algorithms to teacher volunteer transferring problems

Teacher volunteer transferring program (TVTP) in Taiwan provides transferring chances for the teachers serving in public elementary schools. The kernel flow of TVTP involves a special NP-hard rostering operation. That is, an exhausted search for exploring the whole transferring cases is impossible while the practical problem size is not small. Therefore, current Taiwan TVTP adopts deterministically heuristics rules to arrange the transferring plans. However, due to the limited results, this paper proposes a new approach which adopts the genetic algorithms hybridized with a greedy-search mutation operator to serve as a global solver. The positive results show that the application based on the proposed approach can obtain up-to 33% improvements compared to official published data. In additional, it also has great scalability to provide a quite good solution even in a larger problem size, for example, more than nine hundred teachers joined in a practical transferring program.