Evaluación de un algoritmo de recocido simulado con superficies de respuestas

Abstract

The solution of the geographical clustering problem includes a combinatorial classification of the geographical units. The aggregation proposed in this work requires an objective function that minimizes the distance between the objects that will be clustered together, in order to achieve geo-graphical compactness (a desirable goal in problems of geographical design). Because this problem is NP hard [10], it is usually solved with heuristic methodologies that can proportionate satisfactory so-lutions in a reasonable amount of computational time, even for large problems. The main purpose of this research, it is to propose a Box-Behnken experimental design applied into the response’s surface, in order to evaluate the quality of the generated solutions. The balance and adequacy of Simulated Annealing’s parameters would help to control and direct the heuristic method to obtain good solutions for the partitioning problem. Keywords: Geographical Clustering, Experimental Design, Response’s Surface, Simulated Annealing.