• 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
https://doi.org/10.11606/D.45.2023.tde-18042023-090238
Document
Author
Full name
Zheng Zhangzhe
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2023
Supervisor
Committee
Esteves, Luís Gustavo (President)
García, Jesús Enrique
González-lópez, Verónica Andrea
Title in Portuguese
O método de ponderação bayesiana de modelos para seleção de modelos
Keywords in Portuguese
Bayesian model averaging
Fator de Bayes
Incerteza do modelo
Predição
Validação cruzada
Abstract in Portuguese
Nas pesquisas em geral, as pessoas comumente propõem um único modelo na seleção de variáveis explicativas e assumem que é o modelo final, mas isso ignora tanto a incerteza do modelo quanto em esti- mativas de coeficientes. Todos os modelos estatísticos tradicionais têm esse tipo de problema de "incerteza". O Bayesian Model Averaging (BMA) é um método que tem uma longa história de desenvolvimento teórico e aplicação que visa explicar diretamente a incerteza de seleção do modelo. O BMA não seleciona diretamente um único modelo final dentre os disponíveis, mas calcula uma média ponderada dos modelos possíveis ba- seada nas probabilidades a posteriori de tais modelos. O objetivo deste estudo é revisar o BMA e algumas de suas propriedades e aplicá-lo em alguns exemplos reais. Os resultados mostram que o BMA tem um efeito melhor do que o modelo tradicional de seleção de variáveis e melhores resultados de previsão.
Title in English
The bayesian model averaging method for model selection
Keywords in English
Bayes factor
Bayesian model averaging
Cross validation
Model uncertainty
Prediction
Abstract in English
In general research, people commonly propose a single model in selecting explanatory variables and assume it is the final model, but this ignores both model uncertainty and coefficient estimates. All traditional statistical models have this kind of "uncertainty" problem. Bayesian model averaging (BMA) is a method that has a long history of theoretical development and application that aims to directly deals with model selection uncertainty. BMA does not directly select a single final model from those available, but calculates a weighted average of the possible models based on the posteriori probabilities of that models. The aim of this study is to review the BMA and some of its properties and apply it to some real examples. The results show that the BMA has a better performance than the traditional variable selection methods and better forecasting results.
 
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.
dissertacao_final.pdf (1.07 Mbytes)
Publishing Date
2023-04-19
 
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 © 2001-2024. All rights reserved.