• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Master's Dissertation
DOI
10.11606/D.55.2018.tde-27082018-101228
Document
Author
Full name
Margareth Moreira Cordeiro
Institute/School/College
Knowledge Area
Date of Defense
Published
São Carlos, 1993
Supervisor
Committee
Rodrigues, Josemar (President)
Bolfarine, Heleno
Milan, Luis Aparecido
Title in Portuguese
FATOR DE BAYES A POSTERIORI PARA COMPARAR OS COEFICIENTES DE MODELOS DE REGRESSÂO
Keywords in Portuguese
Não disponível
Abstract in Portuguese
Nesta dissertação, abordamos uma análise Bayesiana para comparar os coeficientes de modelos de regressão linear. Esta análise foi baseada no fator Bayes a posteriori introduzido por Aitkin (1991), considerando-se diferentes restrições sobre o modelo adotado e distribuições a priori não informativas. Em todos os modelos adotados observou-se que o fator de Bayes a posteriori não é identificado pela distribuição a priori e pelo tamanho da amostra (paradoxo de Lindley). Fez-se um estudo numérico para comparar o fator de Bayes a posteriori e o teste da razão de verossimilhança, concluindo-se que o fator de Bayes a posteriori é mais eficiente. Também foi proposto um novo critério denominado de teste da razão de entropia a posteriori. Os resultados obtidos através de simulações, quando comparados com o fator de Bayes a posteriori, indicaram que dependendo da escala utilizada o critério proposto é mais eficiente. Desenvolveu-se ainda uma aplicação do fator de Bayes a posteriori para dados relacionados ao desenvolvimento de um sensor de corrente elétrica utilizando fibras ópticas (Vieira, 1992).
Title in English
Posterior Bayes factor to compare coefficient regression models
Keywords in English
Not available
Abstract in English
The Bayesian analysis is covered in this dissertation to compare the coefficients of linear regression models. This analysis was based on the posterior Bayes factor introduced by Aitkin (1991), taking into account different restrictions on the adopted model and non informative priors distributions. On all the adoped models it was noted that the posterior Bayes factor is not influenced by the a prior distribution and by the sample size (Lindley paradox). A numerical study was made to compare the posterior Bayes factor and the likelihood ratio test. The conclusion was that the posteriori Bayes factor is more efficient. It was also proposed a new criterium called posterior entropy ratio test. The results achieved by means of simulations when compared with the posterior Bayes factor showed that depending on the scale used the proposed criterium is more efficient. A posterior Bayes factor application was also development for data related to the development of an electrical current sensor by the use of optical fibers (Vieira, 1992).
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2018-08-27
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2020. All rights reserved.