A note on maximum likelihood estimation for mixture models.

Practitioners as well as some statistics students often blindly use standard software or algorithms to get maximum likelihood estimator (MLE) without checking the validity of existence of such an estimator. Even in simple situations where data comes from mixtures of Gaussians, global MLE does not exist. This note is intended as a teachers corner, highlighting existential issues related to MLE for mixture models, even when the components are not necessarily Gaussian.

This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1007/s42952-022-00180-6

Files

Metadata

Work Title A note on maximum likelihood estimation for mixture models.
Access
Open Access
Creators
  1. G. Jogesh Babu
License In Copyright (Rights Reserved)
Work Type Article
Publisher
  1. Springer Science and Business Media LLC
Publication Date July 23, 2022
Publisher Identifier (DOI)
  1. 10.1007/s42952-022-00180-6
Source
  1. Journal of the Korean Statistical Society
Deposited August 30, 2022

Versions

Analytics

Collections

This resource is currently not in any collection.

Work History

Version 1
published

  • Created
  • Added JKSS_MLE-GMM-final-1.pdf
  • Added Creator G. Jogesh Babu
  • Published
  • Updated Work Title Show Changes
    Work Title
    • A note on maximum likelihood estimation for mixture models.
    • ! A note on maximum likelihood estimation for mixture models.
  • Updated Work Title Show Changes
    Work Title
    • ! A note on maximum likelihood estimation for mixture models.
    • A note on maximum likelihood estimation for mixture models.
  • Updated