A hybrid MOEA for the capacitated exam proximity problem

A hybrid MOEA is used to solve a biobjective version of the capacitated exam proximity problem. In this MOEA, the traditional genetic crossover is replaced by two local search operators. One of the search operators is designed to repair unfeasible timetables produced by the initialization procedure and the mutation operator. The other search operator implements a simplified VNS (variable neighborhood search) meta-heuristic to improve the proximity cost. The resulting nondominated timetables are compared to four other optimization methods using six enrolment datasets. The hybrid MOEA was able to produce the lowest proximity cost for two datasets and the second lowest cost for the remaining four datasets.