Asymptotics of Bayesian median loss estimation

Chi Wai Yu, Bertrand Clarke

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

We establish the consistency, asymptotic normality, and efficiency for estimators derived by minimizing the median of a loss function in a Bayesian context. We contrast this procedure with the behavior of two Frequentist procedures, the least median of squares (LMS) and the least trimmed squares (LTS) estimators, in regression problems. The LMS estimator is the Frequentist version of our estimator, and the LTS estimator approaches a median-based estimator as the trimming approaches 50% on each side. We argue that the Bayesian median-based method is a good tradeoff between the two Frequentist estimators.

Original languageEnglish (US)
Pages (from-to)1950-1958
Number of pages9
JournalJournal of Multivariate Analysis
Volume101
Issue number9
DOIs
StatePublished - Oct 1 2010

Fingerprint

Trimming
Estimator
Least Trimmed Squares
Least Median of Squares
Asymptotic Efficiency
Loss Function
Median
Asymptotic Normality
Regression
Trade-offs

Keywords

  • Asymptotics
  • Least median of squares estimator
  • Least trimmed squares estimator
  • Loss function
  • Median
  • Posterior
  • Regression

ASJC Scopus subject areas

  • Statistics and Probability
  • Numerical Analysis
  • Statistics, Probability and Uncertainty

Cite this

Asymptotics of Bayesian median loss estimation. / Yu, Chi Wai; Clarke, Bertrand.

In: Journal of Multivariate Analysis, Vol. 101, No. 9, 01.10.2010, p. 1950-1958.

Research output: Contribution to journalArticle

Yu, Chi Wai ; Clarke, Bertrand. / Asymptotics of Bayesian median loss estimation. In: Journal of Multivariate Analysis. 2010 ; Vol. 101, No. 9. pp. 1950-1958.
@article{ef0f8411504c47f79cfe63459ad1f9fc,
title = "Asymptotics of Bayesian median loss estimation",
abstract = "We establish the consistency, asymptotic normality, and efficiency for estimators derived by minimizing the median of a loss function in a Bayesian context. We contrast this procedure with the behavior of two Frequentist procedures, the least median of squares (LMS) and the least trimmed squares (LTS) estimators, in regression problems. The LMS estimator is the Frequentist version of our estimator, and the LTS estimator approaches a median-based estimator as the trimming approaches 50{\%} on each side. We argue that the Bayesian median-based method is a good tradeoff between the two Frequentist estimators.",
keywords = "Asymptotics, Least median of squares estimator, Least trimmed squares estimator, Loss function, Median, Posterior, Regression",
author = "Yu, {Chi Wai} and Bertrand Clarke",
year = "2010",
month = "10",
day = "1",
doi = "10.1016/j.jmva.2010.04.013",
language = "English (US)",
volume = "101",
pages = "1950--1958",
journal = "Journal of Multivariate Analysis",
issn = "0047-259X",
publisher = "Academic Press Inc.",
number = "9",

}

TY - JOUR

T1 - Asymptotics of Bayesian median loss estimation

AU - Yu, Chi Wai

AU - Clarke, Bertrand

PY - 2010/10/1

Y1 - 2010/10/1

N2 - We establish the consistency, asymptotic normality, and efficiency for estimators derived by minimizing the median of a loss function in a Bayesian context. We contrast this procedure with the behavior of two Frequentist procedures, the least median of squares (LMS) and the least trimmed squares (LTS) estimators, in regression problems. The LMS estimator is the Frequentist version of our estimator, and the LTS estimator approaches a median-based estimator as the trimming approaches 50% on each side. We argue that the Bayesian median-based method is a good tradeoff between the two Frequentist estimators.

AB - We establish the consistency, asymptotic normality, and efficiency for estimators derived by minimizing the median of a loss function in a Bayesian context. We contrast this procedure with the behavior of two Frequentist procedures, the least median of squares (LMS) and the least trimmed squares (LTS) estimators, in regression problems. The LMS estimator is the Frequentist version of our estimator, and the LTS estimator approaches a median-based estimator as the trimming approaches 50% on each side. We argue that the Bayesian median-based method is a good tradeoff between the two Frequentist estimators.

KW - Asymptotics

KW - Least median of squares estimator

KW - Least trimmed squares estimator

KW - Loss function

KW - Median

KW - Posterior

KW - Regression

UR - http://www.scopus.com/inward/record.url?scp=77955089428&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77955089428&partnerID=8YFLogxK

U2 - 10.1016/j.jmva.2010.04.013

DO - 10.1016/j.jmva.2010.04.013

M3 - Article

AN - SCOPUS:77955089428

VL - 101

SP - 1950

EP - 1958

JO - Journal of Multivariate Analysis

JF - Journal of Multivariate Analysis

SN - 0047-259X

IS - 9

ER -