Empirical Bayes Risk Evaluation With Type II Censored Data

Research output: Contribution to journalArticle

Abstract

Empirical Bayes estimators for the scale parameter in a Weibull, Raleigh or an exponential distribution with type II censored data are developed. These estimators are derived by matching moment method, the maximum likelihood method and by modifying the geometric mean estimators developed by Dey and Kuo (1991). The empirical Bayes risks for these estimators and the Bayes rules are evaluated by extensive simulation. Often, the moment empirical Bayes estimator has the smallest empirical Bayes risk. The cases where the modified geometric mean estimator has the smallest empirical Bayes risk are also identified. We also obtain the empirical Bayes risk comparisons for various empirical Bayes estimators when one of the parameters in the hyperprior is known.

Original languageEnglish (US)
Pages (from-to)195-206
Number of pages12
JournalJournal of Statistical Computation and Simulation
Volume48
Issue number3-4
DOIs
StatePublished - Dec 1 1993
Externally publishedYes

Fingerprint

Bayes Risk
Risk Evaluation
Empirical Bayes
Censored Data
Empirical Bayes Estimator
Estimator
Geometric mean
Bayes Rule
Moment Method
Maximum Likelihood Method
Weibull
Scale Parameter
Method of moments
Exponential distribution
Maximum likelihood
Moment
Risk evaluation
Censored data
Simulation

Keywords

  • EB risk comparison
  • Geometric mean estimator
  • Matching moment method
  • ML-II prior
  • Parametric empirical Bayes estimation
  • Type II censored data

ASJC Scopus subject areas

  • Statistics, Probability and Uncertainty
  • Modeling and Simulation
  • Statistics and Probability
  • Applied Mathematics

Cite this

Empirical Bayes Risk Evaluation With Type II Censored Data. / Kuo, Lynn; Yiannoutsos, Constantin.

In: Journal of Statistical Computation and Simulation, Vol. 48, No. 3-4, 01.12.1993, p. 195-206.

Research output: Contribution to journalArticle

@article{cdb809ce4f1d488aaa1d49c6c825ea45,
title = "Empirical Bayes Risk Evaluation With Type II Censored Data",
abstract = "Empirical Bayes estimators for the scale parameter in a Weibull, Raleigh or an exponential distribution with type II censored data are developed. These estimators are derived by matching moment method, the maximum likelihood method and by modifying the geometric mean estimators developed by Dey and Kuo (1991). The empirical Bayes risks for these estimators and the Bayes rules are evaluated by extensive simulation. Often, the moment empirical Bayes estimator has the smallest empirical Bayes risk. The cases where the modified geometric mean estimator has the smallest empirical Bayes risk are also identified. We also obtain the empirical Bayes risk comparisons for various empirical Bayes estimators when one of the parameters in the hyperprior is known.",
keywords = "EB risk comparison, Geometric mean estimator, Matching moment method, ML-II prior, Parametric empirical Bayes estimation, Type II censored data",
author = "Lynn Kuo and Constantin Yiannoutsos",
year = "1993",
month = "12",
day = "1",
doi = "10.1080/00949659308811551",
language = "English (US)",
volume = "48",
pages = "195--206",
journal = "Journal of Statistical Computation and Simulation",
issn = "0094-9655",
publisher = "Taylor and Francis Ltd.",
number = "3-4",

}

TY - JOUR

T1 - Empirical Bayes Risk Evaluation With Type II Censored Data

AU - Kuo, Lynn

AU - Yiannoutsos, Constantin

PY - 1993/12/1

Y1 - 1993/12/1

N2 - Empirical Bayes estimators for the scale parameter in a Weibull, Raleigh or an exponential distribution with type II censored data are developed. These estimators are derived by matching moment method, the maximum likelihood method and by modifying the geometric mean estimators developed by Dey and Kuo (1991). The empirical Bayes risks for these estimators and the Bayes rules are evaluated by extensive simulation. Often, the moment empirical Bayes estimator has the smallest empirical Bayes risk. The cases where the modified geometric mean estimator has the smallest empirical Bayes risk are also identified. We also obtain the empirical Bayes risk comparisons for various empirical Bayes estimators when one of the parameters in the hyperprior is known.

AB - Empirical Bayes estimators for the scale parameter in a Weibull, Raleigh or an exponential distribution with type II censored data are developed. These estimators are derived by matching moment method, the maximum likelihood method and by modifying the geometric mean estimators developed by Dey and Kuo (1991). The empirical Bayes risks for these estimators and the Bayes rules are evaluated by extensive simulation. Often, the moment empirical Bayes estimator has the smallest empirical Bayes risk. The cases where the modified geometric mean estimator has the smallest empirical Bayes risk are also identified. We also obtain the empirical Bayes risk comparisons for various empirical Bayes estimators when one of the parameters in the hyperprior is known.

KW - EB risk comparison

KW - Geometric mean estimator

KW - Matching moment method

KW - ML-II prior

KW - Parametric empirical Bayes estimation

KW - Type II censored data

UR - http://www.scopus.com/inward/record.url?scp=84949373872&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84949373872&partnerID=8YFLogxK

U2 - 10.1080/00949659308811551

DO - 10.1080/00949659308811551

M3 - Article

VL - 48

SP - 195

EP - 206

JO - Journal of Statistical Computation and Simulation

JF - Journal of Statistical Computation and Simulation

SN - 0094-9655

IS - 3-4

ER -