Exact confidence intervals for Weibull parameters and percentiles

L. D. Phan, John I. McCool

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Simple estimators of the Weibull shape parameter and any quantile in uncensored samples are shown to share the pivotal function properties of maximum likelihood estimators. This allows interval estimation of the Weibull parameters and quantiles once the distributions of the necessary pivotal functions are determined by Monte Carlo sampling for a given sample size and quantile of interest. Free software for this purpose is made available for download. It is shown that the quantile and shape parameter estimates may be readily corrected so as to be median unbiased. The shape parameter estimator is shown to be less precise than the corresponding maximum likelihood estimator. However, the precision of the estimator of the pth quantile is quite comparable to that of the maximum likelihood method over the range from 0.4≤p≤0.7.

Original languageEnglish (US)
Pages (from-to)387-394
Number of pages8
JournalProceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability
Volume223
Issue number4
DOIs
StatePublished - Dec 1 2009

Fingerprint

Maximum likelihood
Sampling

All Science Journal Classification (ASJC) codes

  • Safety, Risk, Reliability and Quality

Cite this

@article{2d84152171f84cc8864dafba8de00404,
title = "Exact confidence intervals for Weibull parameters and percentiles",
abstract = "Simple estimators of the Weibull shape parameter and any quantile in uncensored samples are shown to share the pivotal function properties of maximum likelihood estimators. This allows interval estimation of the Weibull parameters and quantiles once the distributions of the necessary pivotal functions are determined by Monte Carlo sampling for a given sample size and quantile of interest. Free software for this purpose is made available for download. It is shown that the quantile and shape parameter estimates may be readily corrected so as to be median unbiased. The shape parameter estimator is shown to be less precise than the corresponding maximum likelihood estimator. However, the precision of the estimator of the pth quantile is quite comparable to that of the maximum likelihood method over the range from 0.4≤p≤0.7.",
author = "Phan, {L. D.} and McCool, {John I.}",
year = "2009",
month = "12",
day = "1",
doi = "10.1243/1748006XJRR247",
language = "English (US)",
volume = "223",
pages = "387--394",
journal = "Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability",
issn = "1748-006X",
publisher = "SAGE Publications Ltd",
number = "4",

}

Exact confidence intervals for Weibull parameters and percentiles. / Phan, L. D.; McCool, John I.

In: Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, Vol. 223, No. 4, 01.12.2009, p. 387-394.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Exact confidence intervals for Weibull parameters and percentiles

AU - Phan, L. D.

AU - McCool, John I.

PY - 2009/12/1

Y1 - 2009/12/1

N2 - Simple estimators of the Weibull shape parameter and any quantile in uncensored samples are shown to share the pivotal function properties of maximum likelihood estimators. This allows interval estimation of the Weibull parameters and quantiles once the distributions of the necessary pivotal functions are determined by Monte Carlo sampling for a given sample size and quantile of interest. Free software for this purpose is made available for download. It is shown that the quantile and shape parameter estimates may be readily corrected so as to be median unbiased. The shape parameter estimator is shown to be less precise than the corresponding maximum likelihood estimator. However, the precision of the estimator of the pth quantile is quite comparable to that of the maximum likelihood method over the range from 0.4≤p≤0.7.

AB - Simple estimators of the Weibull shape parameter and any quantile in uncensored samples are shown to share the pivotal function properties of maximum likelihood estimators. This allows interval estimation of the Weibull parameters and quantiles once the distributions of the necessary pivotal functions are determined by Monte Carlo sampling for a given sample size and quantile of interest. Free software for this purpose is made available for download. It is shown that the quantile and shape parameter estimates may be readily corrected so as to be median unbiased. The shape parameter estimator is shown to be less precise than the corresponding maximum likelihood estimator. However, the precision of the estimator of the pth quantile is quite comparable to that of the maximum likelihood method over the range from 0.4≤p≤0.7.

UR - http://www.scopus.com/inward/record.url?scp=77149143830&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77149143830&partnerID=8YFLogxK

U2 - 10.1243/1748006XJRR247

DO - 10.1243/1748006XJRR247

M3 - Article

VL - 223

SP - 387

EP - 394

JO - Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability

JF - Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability

SN - 1748-006X

IS - 4

ER -