|
Activity Number:
|
77
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305419 |
|
Title:
|
Bayesian Analysis of Vector Spline Smoothing
|
|
Author(s):
|
Shawn Ni*+ and Dongchu Sun
|
|
Companies:
|
University of Missouri and University of Missouri
|
|
Address:
|
118 Professional Building, Columbia, MO, 65211,
|
|
Keywords:
|
spline smoothing ; Bayesian linear mixed model ; noninformative prior ; MCMC
|
|
Abstract:
|
We show that vector cubic spline smoothing is equivalent to a Bayesian linear mixed model containing a non-stationary vector component and a stationary vector component. The Bayesian linear mixed model corresponds to a multivariate state space model where the magnitude of signal versus noise is captured by the ratio of covariance matrices of the non-stationary and stationary components. We propose a noninformative prior on the ratio of the covariances of these components. We establish posterior propriety under the noninformative prior and develop an MCMC algorithm to draw from the posteriors.
|
