Bayesian semiparametric spatial and joint spatio-temporal smoothing

مرکز و کتابخانه مطالعات اسلامی به زبان های اروپایی

منو

" Bayesian semiparametric spatial and joint spatio-temporal smoothing "
Gentry White D. Sun

Document Type	:	Latin Dissertation
Language of Document	:	English
Record Number	:	55586
Doc. No	:	TL25540
Call number	:	‭3253189‬
Main Entry	:	Gentry White
Title & Author	:	Bayesian semiparametric spatial and joint spatio-temporal smoothing\ Gentry White
College	:	University of Missouri - Columbia
Date	:	2006
Degree	:	Ph.D.
student score	:	2006
Page No	:	141
Abstract	:	Smoothing is the practice of modeling data in order to eliminate random variation from the observed data and provide estimates of the underlying process. Models are developed here beginning with an additive model that incorporates spatio-temporal smoothing of the observed mortality rates for female breast cancer in Missouri from 1969 through 2001. The next model developed uses an intrinsic auto regressive (IAR) prior to smooth the temporal trends in the data and a conditional auto-regressive (CAR) prior for the spatial effects. These two are combined in a single joint prior for spatio-temporal effects. The third model is a joint spatio-temporal model, using the IAR prior for the temporal trends and a spatial prior based on the thin-plate spline solution. These results open the door for further exploration including an alternate parameterization of the thin-plate splines prior to allow the computation of Bayes factors comparing the CAR prior and the thin-plate splines prior. This example is illustrated using a data set of responses to the Missouri Turkey Hunting Survey of 1996, conducted by the Missouri Department of Conservation. Additional strategies for dimension reduction of large scale problems are explored by reducing the number of basis functions in the thin-plate spline prior, results are compared for various degrees of dimension reduction. The example in this case involves the analysis of data for U.S. mortality due to colorectal cancer among men during the period 1999-2003.
Subject	:	Pure sciences; Bayesian semiparametric spatial smoothing; Semiparametric; Smoothing; Spatiotemporal; Statistics; 0463:Statistics
Added Entry	:	D. Sun
Added Entry	:	University of Missouri - Columbia