Comparative Analysis of Constraint Handling Techniques Based on Taguchi Design of Experiments

Abstract

In this chapter, we analyze the effect that constraint-handling techniques such as penalty functions, repairmethods, and decoders have on a steady-state genetic algorithm running on a smartphone. We examine these techniques on one particular problem: the tourist trip design problem. This problem selects a set of points of interest that matches tourist preferences to bring a personalized trip plan. Our points of interest are focused on Mexico City. In order to test the differences among the constraint-handling techniques, we apply the Taguchi design of experiments. The results support the decision that a random decoder is the best choice to handle constraints in the context of the tourist trip design problem.