Change search
ReferencesLink to record
Permanent link

Direct link
Stochastic Models for Smoothing Splines: A Bayesian Approach
Norwegian University of Science and Technology, Faculty of Information Technology, Mathematics and Electrical Engineering, Department of Mathematical Sciences.
2011 (English)MasteroppgaveStudent thesis
Abstract [en]

Flexible data regression is an important tool for capturing complicated trends in data. One approach is penalized smoothing splines, where there are several mainstream methods. A weakness is, however, the quantification of uncertainty. We will in thesis present two mainstream smoothing spline methods, P-splines and O'Sullivan splines, and the RW2 model; a Bayesian hierarchical model based on a latent field. The Bayesian priors are specified by a stochastic Poisson equation, and spline estimates are approximated along a finite element Galerkin approach. We evaluate the three methods using integrated nested Laplace approximations (INLA) for a full Bayesian analysis supplying credible bands. The methods give fairly similar results and we investigate the theoretical motivates behind the methods. As an extension of the Bayesian models, the smoothing parameter is incorporated in latent field. This gives an adaptive smoothing method, which better estimates jumps and quick curvature changes. Further, the close relationship between O'Sullivan splines smoothing splines is discussed, revealing O'Sullivan splines to be a finite element Petrov-Galerkin approximation of smoothing splines. The main results are the possibility of credible bands, the extension to adaptive smoothing and the finite element understanding of O'Sullivan splines.

Place, publisher, year, edition, pages
Institutt for matematiske fag , 2011. , 63 p.
Keyword [no]
ntnudaim:6160, MTFYMA fysikk og matematikk, Industriell matematikk
URN: urn:nbn:no:ntnu:diva-13724Local ID: ntnudaim:6160OAI: diva2:442039
Available from: 2011-09-20 Created: 2011-09-20

Open Access in DiVA

fulltext(1305 kB)641 downloads
File information
File name FULLTEXT01.pdfFile size 1305 kBChecksum SHA-512
Type fulltextMimetype application/pdf
cover(47 kB)37 downloads
File information
File name COVER01.pdfFile size 47 kBChecksum SHA-512
Type coverMimetype application/pdf

By organisation
Department of Mathematical Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 641 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 113 hits
ReferencesLink to record
Permanent link

Direct link