M-estimates of regression when the scale is unknown and the error distribution is possibly asymmetric: A minimax result

Bing Li, Ruben H. Zamar

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Huber (1964) found the minimax-variance M-estimate of location under the assumption that the scale parameter is known; Li and Zamar (1991) extended this result to the case when the scale is unknown. We consider the robust estimation of the regression coefficients (β1 , . . . , βp) when the scale and the intercept parameters are unknown. The minimax-variance estimates of (β1 , . . . , βp) with respect to the trace of their asymptotic covariance matrix are derived. The maximum is taken over ∈-contamination neighbourhoods of a central regression model with Gaussian errors (asymmetric contamination is allowed), and the minimum is taken over a large class of generalized M-estimates of regression of the Mallow type. The optimal choice of estimates for the nuisance parameters (scale and intercept) is also considered.

Original languageEnglish (US)
Pages (from-to)193-206
Number of pages14
JournalCanadian Journal of Statistics
Volume24
Issue number2
DOIs
StatePublished - Jan 1 1996

Fingerprint

M-estimates
Minimax
Regression
Intercept
Contamination
Unknown
Asymptotic Covariance Matrix
Nuisance Parameter
Robust Estimation
Scale Parameter
Regression Coefficient
Estimate
Regression Model
Trace

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

@article{c24ef143b46b4ed492d5d9aa2b84f9de,
title = "M-estimates of regression when the scale is unknown and the error distribution is possibly asymmetric: A minimax result",
abstract = "Huber (1964) found the minimax-variance M-estimate of location under the assumption that the scale parameter is known; Li and Zamar (1991) extended this result to the case when the scale is unknown. We consider the robust estimation of the regression coefficients (β1 , . . . , βp) when the scale and the intercept parameters are unknown. The minimax-variance estimates of (β1 , . . . , βp) with respect to the trace of their asymptotic covariance matrix are derived. The maximum is taken over ∈-contamination neighbourhoods of a central regression model with Gaussian errors (asymmetric contamination is allowed), and the minimum is taken over a large class of generalized M-estimates of regression of the Mallow type. The optimal choice of estimates for the nuisance parameters (scale and intercept) is also considered.",
author = "Bing Li and Zamar, {Ruben H.}",
year = "1996",
month = "1",
day = "1",
doi = "10.2307/3315625",
language = "English (US)",
volume = "24",
pages = "193--206",
journal = "Canadian Journal of Statistics",
issn = "0319-5724",
publisher = "Statistical Society of Canada",
number = "2",

}

M-estimates of regression when the scale is unknown and the error distribution is possibly asymmetric : A minimax result. / Li, Bing; Zamar, Ruben H.

In: Canadian Journal of Statistics, Vol. 24, No. 2, 01.01.1996, p. 193-206.

Research output: Contribution to journalArticle

TY - JOUR

T1 - M-estimates of regression when the scale is unknown and the error distribution is possibly asymmetric

T2 - A minimax result

AU - Li, Bing

AU - Zamar, Ruben H.

PY - 1996/1/1

Y1 - 1996/1/1

N2 - Huber (1964) found the minimax-variance M-estimate of location under the assumption that the scale parameter is known; Li and Zamar (1991) extended this result to the case when the scale is unknown. We consider the robust estimation of the regression coefficients (β1 , . . . , βp) when the scale and the intercept parameters are unknown. The minimax-variance estimates of (β1 , . . . , βp) with respect to the trace of their asymptotic covariance matrix are derived. The maximum is taken over ∈-contamination neighbourhoods of a central regression model with Gaussian errors (asymmetric contamination is allowed), and the minimum is taken over a large class of generalized M-estimates of regression of the Mallow type. The optimal choice of estimates for the nuisance parameters (scale and intercept) is also considered.

AB - Huber (1964) found the minimax-variance M-estimate of location under the assumption that the scale parameter is known; Li and Zamar (1991) extended this result to the case when the scale is unknown. We consider the robust estimation of the regression coefficients (β1 , . . . , βp) when the scale and the intercept parameters are unknown. The minimax-variance estimates of (β1 , . . . , βp) with respect to the trace of their asymptotic covariance matrix are derived. The maximum is taken over ∈-contamination neighbourhoods of a central regression model with Gaussian errors (asymmetric contamination is allowed), and the minimum is taken over a large class of generalized M-estimates of regression of the Mallow type. The optimal choice of estimates for the nuisance parameters (scale and intercept) is also considered.

UR - http://www.scopus.com/inward/record.url?scp=0030304453&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030304453&partnerID=8YFLogxK

U2 - 10.2307/3315625

DO - 10.2307/3315625

M3 - Article

AN - SCOPUS:0030304453

VL - 24

SP - 193

EP - 206

JO - Canadian Journal of Statistics

JF - Canadian Journal of Statistics

SN - 0319-5724

IS - 2

ER -