Existence of maximum likelihood estimation for three-parameter log-normal distribution

Yincai Tang, Xiaoling Wei

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In the literature concerning maximum likelihood estimation (MLE), likelihood is always defined as the product of corresponding density function for the observations. When the distribution function is log-normal, the unboundedness problem will lead to nonexistence of the estimation. In this article, we introduce an observed error term to modify the traditional likelihood function and consider the condition for the existence of the MLEs of three parameters in the corrected likelihood function. It is shown through an example that the MLEs of three parameters exists if the condition is satisfied and the estimation is not sensitive to the choice of the observation error.

Original languageEnglish
Title of host publicationProceedings of 2009 8th International Conference on Reliability, Maintainability and Safety, ICRMS 2009
Pages305-307
Number of pages3
DOIs
StatePublished - 2009
Event2009 8th International Conference on Reliability, Maintainability and Safety, ICRMS 2009 - Chengdu, China
Duration: 20 Jul 200924 Jul 2009

Publication series

NameProceedings of 2009 8th International Conference on Reliability, Maintainability and Safety, ICRMS 2009

Conference

Conference2009 8th International Conference on Reliability, Maintainability and Safety, ICRMS 2009
Country/TerritoryChina
CityChengdu
Period20/07/0924/07/09

Keywords

  • Existence
  • Maximum likelihood estimate
  • Observation error
  • Three-parameter log-normal distribution

Fingerprint

Dive into the research topics of 'Existence of maximum likelihood estimation for three-parameter log-normal distribution'. Together they form a unique fingerprint.

Cite this