Selecting explanatory variables with the modified version of the bayesian information criterion
(2008) In Quality and Reliability Engineering International 24(6). p.627-641- Abstract
We consider the situation in which a large database needs to be analyzed to identify a few important predictors of a given quantitative response variable. There is a lot of evidence that in this case classical model selection criteria, such as the Akaike information criterion or the Bayesian information criterion (BIC), have a strong tendency to overestimate the number of regressors. In our earlier papers, we developed the modified version of BIC (mBIC), which enables the incorporation of prior knowledge on a number of regressors and prevents overestimation. In this article, we review earlier results on mBIC and discuss the relationship of this criterion to the well-known Bonferroni correction for multiple testing and the Bayes oracle,... (More)
We consider the situation in which a large database needs to be analyzed to identify a few important predictors of a given quantitative response variable. There is a lot of evidence that in this case classical model selection criteria, such as the Akaike information criterion or the Bayesian information criterion (BIC), have a strong tendency to overestimate the number of regressors. In our earlier papers, we developed the modified version of BIC (mBIC), which enables the incorporation of prior knowledge on a number of regressors and prevents overestimation. In this article, we review earlier results on mBIC and discuss the relationship of this criterion to the well-known Bonferroni correction for multiple testing and the Bayes oracle, which minimizes the expected costs of inference. We use computer simulations and a real data analysis to illustrate the performance of the original mBIC and its rank version, which is designed to deal with data that contain some outlying observations.
(Less)
- author
- Bogdan, Malgorzata LU ; Ghosh, Jayanta K. and Zak-Szatkowska, Małgorzata
- publishing date
- 2008-10
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Bayes oracle, Data mining, Model selection, Multiple regression, Multiple testing
- in
- Quality and Reliability Engineering International
- volume
- 24
- issue
- 6
- pages
- 15 pages
- publisher
- John Wiley & Sons Inc.
- external identifiers
-
- scopus:54949098146
- ISSN
- 0748-8017
- DOI
- 10.1002/qre.936
- language
- English
- LU publication?
- no
- id
- 98d1dffc-5021-44aa-8b51-9f8826fcf038
- date added to LUP
- 2023-12-08 09:23:22
- date last changed
- 2023-12-11 10:31:46
@article{98d1dffc-5021-44aa-8b51-9f8826fcf038, abstract = {{<p>We consider the situation in which a large database needs to be analyzed to identify a few important predictors of a given quantitative response variable. There is a lot of evidence that in this case classical model selection criteria, such as the Akaike information criterion or the Bayesian information criterion (BIC), have a strong tendency to overestimate the number of regressors. In our earlier papers, we developed the modified version of BIC (mBIC), which enables the incorporation of prior knowledge on a number of regressors and prevents overestimation. In this article, we review earlier results on mBIC and discuss the relationship of this criterion to the well-known Bonferroni correction for multiple testing and the Bayes oracle, which minimizes the expected costs of inference. We use computer simulations and a real data analysis to illustrate the performance of the original mBIC and its rank version, which is designed to deal with data that contain some outlying observations.</p>}}, author = {{Bogdan, Malgorzata and Ghosh, Jayanta K. and Zak-Szatkowska, Małgorzata}}, issn = {{0748-8017}}, keywords = {{Bayes oracle; Data mining; Model selection; Multiple regression; Multiple testing}}, language = {{eng}}, number = {{6}}, pages = {{627--641}}, publisher = {{John Wiley & Sons Inc.}}, series = {{Quality and Reliability Engineering International}}, title = {{Selecting explanatory variables with the modified version of the bayesian information criterion}}, url = {{http://dx.doi.org/10.1002/qre.936}}, doi = {{10.1002/qre.936}}, volume = {{24}}, year = {{2008}}, }