Bayesian Inference for Logistic Regression Models Using Sequential Posterior Simulation

Geweke, J; Durham, G; Xu, H

Bayesian Inference for Logistic Regression Models Using Sequential Posterior Simulation

Geweke, J Durham, G Xu, H

Permalink

Publisher:: CRC Press
Publication Type:: Chapter
Citation:: Current Trends in Bayesian Methodology with Applications, 2015, pp. 290 - 310
Issue Date:: 2015

Closed Access

	Filename	Description	Size
	Bayesian_Inference.pdf	Published version	243.62 kB		View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Geweke, J	en_US
dc.contributor.author	Durham, G	en_US
dc.contributor.author	Xu, H	en_US
dc.contributor.editor	Upadhyay, S	en_US
dc.contributor.editor	Singh, U	en_US
dc.contributor.editor	Dey, D	en_US
dc.contributor.editor	Loganathan, A	en_US
dc.date.issued	2015	en_US
dc.identifier.citation	Current Trends in Bayesian Methodology with Applications, 2015, pp. 290 - 310	en_US
dc.identifier.isbn	978-1482235111	en_US
dc.identifier.uri	http://hdl.handle.net/10453/32945
dc.description.abstract	The logistic specification has been used extensively in non-Bayesian statistics to model the dependence of discrete outcomes on the values of specified covariates. Because the likelihood function is globally weakly concave estimation by maximum likelihood is generally straightforward even in commonly arising applications with scores or hundreds of parameters. In contrast Bayesian inference has proven awkward, requiring normal approximations to the likelihood or specialized adaptations of existing Markov chain Monte Carlo and data augmentation methods. This paper approaches Bayesian inference in logistic models using recently developed generic sequential posterior simulaton (SPS) methods that require little more than the ability to evaluate the likelihood function. Compared with existing alternatives SPS is much simpler, and provides numerical standard errors and accurate approximations of marginal likelihoods as by-products. The SPS algorithm for Bayesian inference is amenable to massively parallel implementation, and when implemented using graphical processing units it is more efficient than existing alternatives. The paper demonstrates these points by means of several examples.	en_US
dc.publisher	CRC Press	en_US
dc.relation.ispartof	Current Trends in Bayesian Methodology with Applications	en_US
dc.title	Bayesian Inference for Logistic Regression Models Using Sequential Posterior Simulation	en_US
dc.type	Chapter
utslib.location	USA	en_US
utslib.for	140303 Economic Models and Forecasting	en_US
utslib.for	140305 Time-Series Analysis	en_US
utslib.for	140302 Econometric and Statistical Methods	en_US
pubs.embargo.period	Not known	en_US
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Business
pubs.organisational-group	/University of Technology Sydney/Faculty of Business/Economics Discipline
utslib.copyright.status	closed_access
pubs.consider-herdc	true	en_US
pubs.place-of-publication	USA	en_US
pubs.publication-status	Published	en_US

Abstract:

The logistic specification has been used extensively in non-Bayesian statistics to model the dependence of discrete outcomes on the values of specified covariates. Because the likelihood function is globally weakly concave estimation by maximum likelihood is generally straightforward even in commonly arising applications with scores or hundreds of parameters. In contrast Bayesian inference has proven awkward, requiring normal approximations to the likelihood or specialized adaptations of existing Markov chain Monte Carlo and data augmentation methods. This paper approaches Bayesian inference in logistic models using recently developed generic sequential posterior simulaton (SPS) methods that require little more than the ability to evaluate the likelihood function. Compared with existing alternatives SPS is much simpler, and provides numerical standard errors and accurate approximations of marginal likelihoods as by-products. The SPS algorithm for Bayesian inference is amenable to massively parallel implementation, and when implemented using graphical processing units it is more efficient than existing alternatives. The paper demonstrates these points by means of several examples.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/32945