|
Activity Number:
|
410
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #307364 |
|
Title:
|
Reference Priors for Gaussian Processes with Spatial Correlation Structure
|
|
Author(s):
|
Mi Hyun Lee*+ and Dongchu Sun
|
|
Companies:
|
Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University/University of Missouri-Columbia
|
|
Address:
|
208 Turner Street, Apt. 220, Blacksburg, VA, 24060,
|
|
Keywords:
|
Gaussian process ; spatial correlation function ; reference prior
|
|
Abstract:
|
In many of environmental applications, Gaussian random processes---which are determined by mean and covariance functions---are commonly used to model spatial data by specifying the spatial correlation structure of the covariance functions. In this paper, Bayesian analysis on the unknown mean and covariance parameters of the spatial Gaussian random fields are proposed. The reference priors for the unknown covariance parameters are developed when the mean function is assumed to be linear and the regression parameters in the mean function are considered to be nuisance parameters. Under the restricted situations, Berger, De Oliveira, and Sanso (2001) and Paulo (2005) derived the reference priors for the unknown covariance parameters of the spatial Gaussian random processes. The reference priors are studied under more general situation in this paper.
|
