A Strategy for Finding the Optimal Number of Clusters Based on the Grey Wolf Algorithm

Abstract

In computational sciences literature, we can find different problems that can be solved by Optimization Algorithms. In this case, we are using an Optimization Algorithm based on wolves’ behavior for a clustering problem. This study proposes an optimization approach based on the Grey Wolf Optimizer (GWO), designed to determine the optimal number of clusters by leveraging centroid data. We consider that the mathematic model used by the authors into the mentioned algorithm, it is the ideal to solve a clustering problem. Therefore, we tested the proposed algorithm with three different datasets: Iris, Wine and Diagnostic Wisconsin Breast Cancer Database, respectively, and we called the method Clustering Grey Wolf Optimizer, and we denoted as CGWO. Besides, to test the proposed method, we show a comparison of results versus the Firework Algorithm. We presented a statistical comparison between both mentioned methods, GWO and FWO, for clustering with the main objective of complementing the conclusions of this work.