A metric to discriminate the selection of algorithms for the general ATSP problem

Abstract

In this paper we propose: (1) the use of discriminant analysis as a means for predictive learning (data-mining techniques) aiming at selecting metaheuristic algorithms and (2) the use of a metric for improving the selection of the algorithms that best solve a given instance of the Asymmetric Traveling Salesman Problem (ATSP). The only metric that had existed so far to determine the best algorithm for solving an ATSP instance is based on the number of cities; nevertheless, it is not sufficiently adequate for discriminating the best algorithm for solving an ATSP instance, thus the necessity for devising a new metric through the use of data-mining techniques. © 2008 Springer-Verlag Berlin Heidelberg.