English

On uniqueness of the set of k-means

Statistics Theory 2024-10-18 v1 Methodology Machine Learning Statistics Theory

Abstract

We provide necessary and sufficient conditions for the uniqueness of the k-means set of a probability distribution. This uniqueness problem is related to the choice of k: depending on the underlying distribution, some values of this parameter could lead to multiple sets of k-means, which hampers the interpretation of the results and/or the stability of the algorithms. We give a general assessment on consistency of the empirical k-means adapted to the setting of non-uniqueness and determine the asymptotic distribution of the within cluster sum of squares (WCSS). We also provide statistical characterizations of k-means uniqueness in terms of the asymptotic behavior of the empirical WCSS. As a consequence, we derive a bootstrap test for uniqueness of the set of k-means. The results are illustrated with examples of different types of non-uniqueness and we check by simulations the performance of the proposed methodology.

Keywords

Cite

@article{arxiv.2410.13495,
  title  = {On uniqueness of the set of k-means},
  author = {Javier Cárcamo and Antonio Cuevas and Luis A. Rodríguez},
  journal= {arXiv preprint arXiv:2410.13495},
  year   = {2024}
}

Comments

41 pages, 14 figures