A method of two-stage clustering with constraints using agglomerative hierarchical algorithm and one-pass k-means++

Abstract

The aim of this paper is to propose a two-stage method of clustering in which the first stage uses one-pass k-means++ and the second stage uses an agglomerative hierarchical algorithm. This method outperforms a foregoing two-stage algorithm by replacing the ordinary one-pass k-means by one-pass k-means++ in the first stage. Pairwise constraints are also taken into consideration in order to improve its performance. Effectiveness of the proposed method is shown by numerical examples.