Cargando…

Study of Bayesian variable selection method on mixed linear regression models

Variable selection has always been an important issue in statistics. When a linear regression model is used to fit data, selecting appropriate explanatory variables that strongly impact the response variables has a significant effect on the model prediction accuracy and interpretation effect. redThi...

Descripción completa

Detalles Bibliográficos
Autores principales: Li, Yong, Liu, Hefei, Li, Rubing
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Public Library of Science 2023
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10022788/
https://www.ncbi.nlm.nih.gov/pubmed/36930589
http://dx.doi.org/10.1371/journal.pone.0283100
_version_ 1784908793985892352
author Li, Yong
Liu, Hefei
Li, Rubing
author_facet Li, Yong
Liu, Hefei
Li, Rubing
author_sort Li, Yong
collection PubMed
description Variable selection has always been an important issue in statistics. When a linear regression model is used to fit data, selecting appropriate explanatory variables that strongly impact the response variables has a significant effect on the model prediction accuracy and interpretation effect. redThis study introduces the Bayesian adaptive group Lasso method to solve the variable selection problem under a mixed linear regression model with a hidden state and explanatory variables with a grouping structure. First, the definition of the implicit state mixed linear regression model is presented. Thereafter, the Bayesian adaptive group Lasso method is used to determine the penalty function and parameters, after which each parameter’s specific form of the fully conditional posterior distribution is calculated. Moreover, the Gibbs algorithm design is outlined. Simulation experiments are conducted to compare the variable selection and parameter estimation effects in different states. Finally, a dataset of Alzheimer’s Disease is used for application analysis. The results demonstrate that the proposed method can identify the observation from different hidden states, but the results of the variable selection in different states are obviously different.
format Online
Article
Text
id pubmed-10022788
institution National Center for Biotechnology Information
language English
publishDate 2023
publisher Public Library of Science
record_format MEDLINE/PubMed
spelling pubmed-100227882023-03-18 Study of Bayesian variable selection method on mixed linear regression models Li, Yong Liu, Hefei Li, Rubing PLoS One Research Article Variable selection has always been an important issue in statistics. When a linear regression model is used to fit data, selecting appropriate explanatory variables that strongly impact the response variables has a significant effect on the model prediction accuracy and interpretation effect. redThis study introduces the Bayesian adaptive group Lasso method to solve the variable selection problem under a mixed linear regression model with a hidden state and explanatory variables with a grouping structure. First, the definition of the implicit state mixed linear regression model is presented. Thereafter, the Bayesian adaptive group Lasso method is used to determine the penalty function and parameters, after which each parameter’s specific form of the fully conditional posterior distribution is calculated. Moreover, the Gibbs algorithm design is outlined. Simulation experiments are conducted to compare the variable selection and parameter estimation effects in different states. Finally, a dataset of Alzheimer’s Disease is used for application analysis. The results demonstrate that the proposed method can identify the observation from different hidden states, but the results of the variable selection in different states are obviously different. Public Library of Science 2023-03-17 /pmc/articles/PMC10022788/ /pubmed/36930589 http://dx.doi.org/10.1371/journal.pone.0283100 Text en © 2023 Li et al https://creativecommons.org/licenses/by/4.0/This is an open access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/) , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
spellingShingle Research Article
Li, Yong
Liu, Hefei
Li, Rubing
Study of Bayesian variable selection method on mixed linear regression models
title Study of Bayesian variable selection method on mixed linear regression models
title_full Study of Bayesian variable selection method on mixed linear regression models
title_fullStr Study of Bayesian variable selection method on mixed linear regression models
title_full_unstemmed Study of Bayesian variable selection method on mixed linear regression models
title_short Study of Bayesian variable selection method on mixed linear regression models
title_sort study of bayesian variable selection method on mixed linear regression models
topic Research Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10022788/
https://www.ncbi.nlm.nih.gov/pubmed/36930589
http://dx.doi.org/10.1371/journal.pone.0283100
work_keys_str_mv AT liyong studyofbayesianvariableselectionmethodonmixedlinearregressionmodels
AT liuhefei studyofbayesianvariableselectionmethodonmixedlinearregressionmodels
AT lirubing studyofbayesianvariableselectionmethodonmixedlinearregressionmodels