Extending the Hybridization of Metaheuristics with Data Mining to a Broader Domain
Marcos A. Guerine, Isabel Rosseti, Alexandre Plastino · 2014
The incorporation of data mining techniques into metaheuristics has been efficiently adopted to solve several optimization problems. Nevertheless, we observe in the literature that this hybridization has been limited to problems for which the solutions are characterized for sets of elements. In this work, we develop a hybrid data mining metaheuristic based on GRASP and VND to solve the one-commodity pickup-and-delivery traveling salesman, a problem for which solutions are defined by sequences of elements. This way, we extend the domain of combinatorial optimization problems which can benefit from the combination of data mining and metaheuristic. Computational experiments showed that this hybridization with data mining process improves the pure algorithm both in average quality of solution and execution time.