An Ant Colony Optimization for Automatic Data Clustering Problem

Abstract

Approximative methods are widely used in the resolution of complex computational problems, since they are able to present significant results regarding the quality of the solution in a satisfactory time in general. This paper addresses the problem of automatic grouping, admittedly NP-difficult. In the context of this problem, non-exact methods, whose complexity is polynomial, are desirable. To solve the problem, the work presents a metaheuristic based on collective intelligence, inspired by the behavior of the ants. The silhouette index was used to measure the quality of the generated clusters. The results obtained by the proposed algorithm were compared with the literature and indicated that the algorithm is able to find clusters that represent well the original distribution of the data.