Models where the Least Trimmed Squares and Least Median of Squares estimators are maximum likelihood

Publikation: Working paperForskning

Dokumenter

  • rp19_15

    Forlagets udgivne version, 664 KB, PDF-dokument

  • Vanessa Berenguer-Rico, University of Oxford, Oxford, Storbritannien
  • Søren Johansen
  • Bent Nielsen, Oxford University
The Least Trimmed Squares (LTS) and Least Median of Squares (LMS) estimators are popular robust regression estimators. The idea behind the estimators is to find, for a given h, a sub-sample of h `good' observations among n observations and estimate the regression on that sub-sample. We find models, based on the normal or the uniform distribution respectively, in which these estimators are maximum likelihood. We provide an asymptotic theory for the location-scale case in those models. The LTS estimator is found to be sqrt(h) consistent and asymptotically standard normal. The LMS estimator is found to be h consistent and asymptotically Laplace.
OriginalsprogEngelsk
UdgivelsesstedAarhus
UdgiverInstitut for Økonomi, Aarhus Universitet
Antal sider41
StatusUdgivet - 19 sep. 2019
SerietitelCREATES Research Papers
Nummer2019/15

    Forskningsområder

  • Chebychev estimator, LMS, Uniform distribution, Least squares estimator, LTS, Normal distribution, Regression, Robust statistics

Se relationer på Aarhus Universitet Citationsformater

Download-statistik

Ingen data tilgængelig

ID: 165609506