Bayes and empirical Bayes estimation with errors in variables

Shunpu Zhang, Rohana J. Karunamuni

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Suppose that the random variable X is distributed according to exponential families of distributions, conditional on the parameter θ. Assume that the parameter θ has a (prior) distribution G. Because of the measurement error, we can only observe Y = X + ε, where the measurement error ε is independent of X and has a known distribution. This paper considers the squared error loss estimation problem of θ based on the contaminated observation Y. We obtain an expression for the Bayes estimator when the prior G is known. For the case G is completely unknown, an empirical Bayes estimator is proposed based on a sequence of observations Y1,Y2,...,Yn, where Yi's are i.i.d. according to the marginal distribution of Y. It is shown that the proposed empirical Bayes estimator is asymptotically optimal.

Original languageEnglish (US)
Pages (from-to)23-34
Number of pages12
JournalStatistics and Probability Letters
Volume33
Issue number1
DOIs
StatePublished - Apr 15 1997

Fingerprint

Empirical Bayes Estimation
Empirical Bayes Estimator
Errors in Variables
Bayes
Measurement Error
Squared Error Loss
Bayes Estimator
Exponential Family
Asymptotically Optimal
Marginal Distribution
Conditional Distribution
Prior distribution
Random variable
Unknown
Observation
Empirical Bayes
Estimator
Errors in variables
Measurement error

Keywords

  • Asymptotically optimal
  • Bayes
  • Empirical Bayes
  • Kernel density estimates
  • Squared error loss estimation

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Bayes and empirical Bayes estimation with errors in variables. / Zhang, Shunpu; Karunamuni, Rohana J.

In: Statistics and Probability Letters, Vol. 33, No. 1, 15.04.1997, p. 23-34.

Research output: Contribution to journalArticle

Zhang, Shunpu ; Karunamuni, Rohana J. / Bayes and empirical Bayes estimation with errors in variables. In: Statistics and Probability Letters. 1997 ; Vol. 33, No. 1. pp. 23-34.
@article{53faea9084d14d86b33b5a131ecd9521,
title = "Bayes and empirical Bayes estimation with errors in variables",
abstract = "Suppose that the random variable X is distributed according to exponential families of distributions, conditional on the parameter θ. Assume that the parameter θ has a (prior) distribution G. Because of the measurement error, we can only observe Y = X + ε, where the measurement error ε is independent of X and has a known distribution. This paper considers the squared error loss estimation problem of θ based on the contaminated observation Y. We obtain an expression for the Bayes estimator when the prior G is known. For the case G is completely unknown, an empirical Bayes estimator is proposed based on a sequence of observations Y1,Y2,...,Yn, where Yi's are i.i.d. according to the marginal distribution of Y. It is shown that the proposed empirical Bayes estimator is asymptotically optimal.",
keywords = "Asymptotically optimal, Bayes, Empirical Bayes, Kernel density estimates, Squared error loss estimation",
author = "Shunpu Zhang and Karunamuni, {Rohana J.}",
year = "1997",
month = "4",
day = "15",
doi = "10.1016/S0167-7152(96)00106-X",
language = "English (US)",
volume = "33",
pages = "23--34",
journal = "Statistics and Probability Letters",
issn = "0167-7152",
publisher = "Elsevier",
number = "1",

}

TY - JOUR

T1 - Bayes and empirical Bayes estimation with errors in variables

AU - Zhang, Shunpu

AU - Karunamuni, Rohana J.

PY - 1997/4/15

Y1 - 1997/4/15

N2 - Suppose that the random variable X is distributed according to exponential families of distributions, conditional on the parameter θ. Assume that the parameter θ has a (prior) distribution G. Because of the measurement error, we can only observe Y = X + ε, where the measurement error ε is independent of X and has a known distribution. This paper considers the squared error loss estimation problem of θ based on the contaminated observation Y. We obtain an expression for the Bayes estimator when the prior G is known. For the case G is completely unknown, an empirical Bayes estimator is proposed based on a sequence of observations Y1,Y2,...,Yn, where Yi's are i.i.d. according to the marginal distribution of Y. It is shown that the proposed empirical Bayes estimator is asymptotically optimal.

AB - Suppose that the random variable X is distributed according to exponential families of distributions, conditional on the parameter θ. Assume that the parameter θ has a (prior) distribution G. Because of the measurement error, we can only observe Y = X + ε, where the measurement error ε is independent of X and has a known distribution. This paper considers the squared error loss estimation problem of θ based on the contaminated observation Y. We obtain an expression for the Bayes estimator when the prior G is known. For the case G is completely unknown, an empirical Bayes estimator is proposed based on a sequence of observations Y1,Y2,...,Yn, where Yi's are i.i.d. according to the marginal distribution of Y. It is shown that the proposed empirical Bayes estimator is asymptotically optimal.

KW - Asymptotically optimal

KW - Bayes

KW - Empirical Bayes

KW - Kernel density estimates

KW - Squared error loss estimation

UR - http://www.scopus.com/inward/record.url?scp=0031569671&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031569671&partnerID=8YFLogxK

U2 - 10.1016/S0167-7152(96)00106-X

DO - 10.1016/S0167-7152(96)00106-X

M3 - Article

AN - SCOPUS:0031569671

VL - 33

SP - 23

EP - 34

JO - Statistics and Probability Letters

JF - Statistics and Probability Letters

SN - 0167-7152

IS - 1

ER -