Models Where the Least Trimmed Squares and Least Median of Squares Estimators Are Maximum Likelihood
Publikation: Working paper › Forskning
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 <i>h</i>, a sub-sample of <i>h</i> 'good' observations among <i>n</i> 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 <i>h</i><sup>1/2</sup> consistent and asymptotically standard normal. The LMS estimator is found to be <i>h</i> consistent and asymptotically Laplace.
Originalsprog | Engelsk |
---|---|
Antal sider | 39 |
DOI | |
Status | Udgivet - 27 sep. 2019 |
Navn | University of Copenhagen. Institute of Economics. Discussion Papers (Online) |
---|---|
Nummer | 19-11 |
ISSN | 1601-2461 |
- Chebychev estimator, LMS, Uniform distribution, Least squares estimator, LTS, Normal distribution, Regression, Robust statistics
Forskningsområder
Links
- https://www.economics.ku.dk/research/publications/wp/dp_2019/1911.pdf
Indsendt manuskript
ID: 248551490