Clustering Problem Enabled via Kernel Density Estimator

In this work, we present a new clustering algorithm with capacity constraint on clusters, called WKDE-CCP. It is based on the optimization of the weighted kernel density estimator for the determination of centroids, followed by the allocation of observations to the clusters in order to maximize the dissimilarity between them, respecting the capacity constraint of each cluster. In addition, we propose a solution refinement algorithm, which uses the simple comparison and exchange of observations between the clusters in order to minimize the sum of the distances of the observations to the respective centroids, ensuring a better distribution of the elements. Numerical experiments were conducted to illustrate the performance and versatility of the proposed algorithm, indicating that the WKDE-CCP Algorithm is efficient and competitive when compared to other algorithms in the literature that use capacity constraints in their formulations. Features such as simplicity and efficiency make the proposed algorithm a promising field of research.