Optimal compromise between incompatible conditional probability distributions, with application to objective Bayesian kriging

DOI10.1051/PS/2018023MaRDI QIDQ4967800zbMATH OpenOpenAlexWikidataFDO

Authors Joseph Muré

Publication date 11 July 2019

Published in ESAIM: Probability and Statistics (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/1703.07233

zbMATH Keywords

conditional distributionfrequentist coverageGibbs samplingincompatibilityintegrated likelihoodkrigingMarkov kerneloptimal compromiseposterior proprietyreference prior

Mathematics Subject Classification ID

Gaussian processes (60G15) Bayesian inference (62F15) Inference from spatial processes (62M30)

Abstract: Models are often defined through conditional rather than joint distributions, but it can be difficult to check whether the conditional distributions are compatible, i.e. whether there exists a joint probability distribution which generates them. When they are compatible, a Gibbs sampler can be used to sample from this joint distribution. When they are not, the Gibbs sampling algorithm may still be applied, resulting in a "pseudo-Gibbs sampler". We show its stationary probability distribution to be the optimal compromise between the conditional distributions, in the sense that it minimizes a mean squared misfit between them and its own conditional distributions. This allows us to perform Objective Bayesian analysis of correlation parameters in Kriging models by using univariate conditional Jeffreys-rule posterior distributions instead of the widely used multivariate Jeffreys-rule posterior. This strategy makes the full-Bayesian procedure tractable. Numerical examples show it has near-optimal frequentist performance in terms of prediction interval coverage.

Recommendations

• Gibbs ensembles for nearly compatible and incompatible conditional models
• Conditional problem for objective probability.
• A note on compatibility of conditional autoregressive models
• Pseudo-Gibbs sampler for discrete conditional distributions
• Compatibility of conditionally specified models

Cites work

• Bayesian calibration of computer models. (With discussion)
• Bayesian variable selection in linear regression
• Compatible Prior Distributions for Directed Acyclic Graph Models
• Conditionally specified distributions: An introduction. (With comments and a rejoinder).
• Default priors for Gaussioan processes
• Dependency networks for inference, collaborative filtering, and data visualization
• Exactly and almost compatible joint distributions for high-dimensional discrete conditional distributions
• Gaussian processes for machine learning.
• scientific article; zbMATH DE number 3273551 (Why is no real title available?)
• scientific article; zbMATH DE number 3083049 (Why is no real title available?)
• Inconsistent Estimation and Asymptotically Equal Interpolations in Model-Based Geostatistics
• Interpolation of spatial data. Some theory for kriging
• Jeffreys' prior is asymptotically least favorable under entropy risk
• Objective Bayesian analysis for a spatial model with nugget effects
• Objective Bayesian analysis of spatial data with uncertain nugget and range parameters
• Objective Bayesian analysis of spatial models with separable correlation functions
• Objective Bayesian Analysis of Spatially Correlated Data
• On the consistent separation of scale and variance for Gaussian random fields
• Overall objective priors
• Robust Gaussian stochastic process emulation
• The Bayesian Choice
• The case for objective Bayesian analysis
• The design and analysis of computer experiments.

Cited in

(2)

• Propriety of the reference posterior distribution in Gaussian process modeling
• Discussion of: ``Specifying prior distributions in reliability applications: towards new formal rules for informative prior elicitation?

This page was built for publication: Optimal compromise between incompatible conditional probability distributions, with application to objective Bayesian kriging

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4967800)