Singular prior distributions in Bayesian D-optimal design for nonlinear models
Singular prior distributions in Bayesian D-optimal design for nonlinear models
For Bayesian D-optimal design, we define a singular prior distribution to be a prior distribution such that the determinant of the Fisher information matrix has a prior geometric mean of zero for all designs. For such a prior distribution, the Bayesian D-optimality criterion fails to select a design. For the exponential decay model, we characterize singularity of the prior distribution in terms of the expectations of a few elementary transformations of the parameter. For a compartmental model and multi-parameter logistic regression we establish sufficient conditions for singularity of a prior distribution. For logistic regression we also obtain sufficient conditions for non-singularity. The results are applied to show that the weakly informative prior distribution proposed as a default for inference by Gelman, Jakulin, Pittau and Su (2008) should not be used for Bayesian D-optimal design. Additionally, we develop methods to derive and assess Bayesian D-efficient designs for logistic regression when numerical evaluation of the objective function fails due to ill-conditioning.
compartmental model, exponential decay model, ill-conditioning, logistic regression
University of Southampton
Waite, Timothy W.
67ff61af-f85f-4cc7-a7b8-4f188b624266
9 June 2015
Waite, Timothy W.
67ff61af-f85f-4cc7-a7b8-4f188b624266
Waite, Timothy W.
(2015)
Singular prior distributions in Bayesian D-optimal design for nonlinear models
Southampton, GB.
University of Southampton
14pp.
Record type:
Monograph
(Project Report)
Abstract
For Bayesian D-optimal design, we define a singular prior distribution to be a prior distribution such that the determinant of the Fisher information matrix has a prior geometric mean of zero for all designs. For such a prior distribution, the Bayesian D-optimality criterion fails to select a design. For the exponential decay model, we characterize singularity of the prior distribution in terms of the expectations of a few elementary transformations of the parameter. For a compartmental model and multi-parameter logistic regression we establish sufficient conditions for singularity of a prior distribution. For logistic regression we also obtain sufficient conditions for non-singularity. The results are applied to show that the weakly informative prior distribution proposed as a default for inference by Gelman, Jakulin, Pittau and Su (2008) should not be used for Bayesian D-optimal design. Additionally, we develop methods to derive and assess Bayesian D-efficient designs for logistic regression when numerical evaluation of the objective function fails due to ill-conditioning.
Text
sing_revision_unformatted.pdf
- Other
More information
Published date: 9 June 2015
Keywords:
compartmental model, exponential decay model, ill-conditioning, logistic regression
Organisations:
Statistical Sciences Research Institute
Identifiers
Local EPrints ID: 377287
URI: http://eprints.soton.ac.uk/id/eprint/377287
PURE UUID: 0cc1dfb5-2b34-47b8-9c3f-d11e818b06f6
Catalogue record
Date deposited: 23 Jun 2015 13:59
Last modified: 14 Mar 2024 20:00
Export record
Contributors
Author:
Timothy W. Waite
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics