Vol. 12, No. 2, 2019

Download this article
Download this article For screen
For printing
Recent Issues

Volume 12
Issue 2, 181–360
Issue 1, 1–180

Volume 11, 5 issues

Volume 10, 5 issues

Volume 9, 5 issues

Volume 8, 5 issues

Volume 7, 6 issues

Volume 6, 4 issues

Volume 5, 4 issues

Volume 4, 4 issues

Volume 3, 4 issues

Volume 2, 5 issues

Volume 1, 2 issues

The Journal
About the Journal
Subscriptions
Editorial Board
Editors’ Interests
Scientific Advantages
Submission Guidelines
Submission Form
Ethics Statement
Editorial Login
Author Index
Coming Soon
Contacts
 
ISSN: 1944-4184 (e-only)
ISSN: 1944-4176 (print)
On the minimum of the mean-squared error in 2-means clustering

Bernhard G. Bodmann and Craig J. George

Vol. 12 (2019), No. 2, 301–319
DOI: 10.2140/involve.2019.12.301
Abstract

We study the minimum mean-squared error for 2-means clustering when the outcomes of the vector-valued random variable to be clustered are on two spheres, that is, the surface of two touching balls of unit radius in n-dimensional Euclidean space, and the underlying probability distribution is the normalized surface measure. For simplicity, we only consider the asymptotics of large sample sizes and replace empirical samples by the probability measure. The concrete question addressed here is whether a minimizer for the mean-squared error identifies the two individual spheres as clusters. Indeed, in dimensions n 3, the minimum of the mean-squared error is achieved by a partition obtained from a separating hyperplane tangent to both spheres at the point where they touch. In dimension n = 2, however, the minimizer fails to identify the individual spheres; an optimal partition is associated with a hyperplane that does not contain the intersection of the two spheres.

Keywords
$k$-means clustering, performance guarantees, mean-squared error
Mathematical Subject Classification 2010
Primary: 62H30
Milestones
Received: 6 November 2017
Revised: 9 February 2018
Accepted: 7 March 2018
Published: 8 October 2018

Communicated by John C. Wierman
Authors
Bernhard G. Bodmann
Department of Mathematics
University of Houston
Houston, TX
United States
Craig J. George
Department of Mathematics
University of Houston
Houston, TX
United States