Fast and compact smoothing on large multidimensional grids

P. H C Eilers, Iain D. Currie, Maria Durbán

Research output: Contribution to journalArticle

Abstract

A framework of penalized generalized linear models and tensor products of B-splines with roughness penalties allows effective smoothing of data in multidimensional arrays. A straightforward application of the penalized Fisher scoring algorithm quickly runs into storage and computational difficulties. A novel algorithm takes advantage of the special structure of both the data as an array and the model matrix as a tensor product; the algorithm is fast, uses only a moderate amount of memory and works for any number of dimensions. Examples are given of how the method is used to smooth life tables and image data. © 2005 Elsevier B.V. All rights reserved.

Original languageEnglish
Pages (from-to)61-76
Number of pages16
JournalComputational Statistics and Data Analysis
Volume50
Issue number1 SPEC. ISS.
DOIs
Publication statusPublished - 10 Jan 2006

Fingerprint

Smoothing
Grid
Tensor Product
Roughness Penalty
Fisher Scoring
Multidimensional Arrays
Life Table
Matrix Models
Generalized Linear Model
B-spline
Framework

Keywords

  • B-splines
  • Difference penalty
  • Multidimensional array
  • P-splines
  • Smoothing
  • Tensor product

Cite this

Eilers, P. H C ; Currie, Iain D. ; Durbán, Maria. / Fast and compact smoothing on large multidimensional grids. In: Computational Statistics and Data Analysis. 2006 ; Vol. 50, No. 1 SPEC. ISS. pp. 61-76.
@article{dfb24b04319f4d3598037541b236c322,
title = "Fast and compact smoothing on large multidimensional grids",
abstract = "A framework of penalized generalized linear models and tensor products of B-splines with roughness penalties allows effective smoothing of data in multidimensional arrays. A straightforward application of the penalized Fisher scoring algorithm quickly runs into storage and computational difficulties. A novel algorithm takes advantage of the special structure of both the data as an array and the model matrix as a tensor product; the algorithm is fast, uses only a moderate amount of memory and works for any number of dimensions. Examples are given of how the method is used to smooth life tables and image data. {\circledC} 2005 Elsevier B.V. All rights reserved.",
keywords = "B-splines, Difference penalty, Multidimensional array, P-splines, Smoothing, Tensor product",
author = "Eilers, {P. H C} and Currie, {Iain D.} and Maria Durb{\'a}n",
year = "2006",
month = "1",
day = "10",
doi = "10.1016/j.csda.2004.07.008",
language = "English",
volume = "50",
pages = "61--76",
journal = "Computational Statistics and Data Analysis",
issn = "0167-9473",
publisher = "Elsevier",
number = "1 SPEC. ISS.",

}

Fast and compact smoothing on large multidimensional grids. / Eilers, P. H C; Currie, Iain D.; Durbán, Maria.

In: Computational Statistics and Data Analysis, Vol. 50, No. 1 SPEC. ISS., 10.01.2006, p. 61-76.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Fast and compact smoothing on large multidimensional grids

AU - Eilers, P. H C

AU - Currie, Iain D.

AU - Durbán, Maria

PY - 2006/1/10

Y1 - 2006/1/10

N2 - A framework of penalized generalized linear models and tensor products of B-splines with roughness penalties allows effective smoothing of data in multidimensional arrays. A straightforward application of the penalized Fisher scoring algorithm quickly runs into storage and computational difficulties. A novel algorithm takes advantage of the special structure of both the data as an array and the model matrix as a tensor product; the algorithm is fast, uses only a moderate amount of memory and works for any number of dimensions. Examples are given of how the method is used to smooth life tables and image data. © 2005 Elsevier B.V. All rights reserved.

AB - A framework of penalized generalized linear models and tensor products of B-splines with roughness penalties allows effective smoothing of data in multidimensional arrays. A straightforward application of the penalized Fisher scoring algorithm quickly runs into storage and computational difficulties. A novel algorithm takes advantage of the special structure of both the data as an array and the model matrix as a tensor product; the algorithm is fast, uses only a moderate amount of memory and works for any number of dimensions. Examples are given of how the method is used to smooth life tables and image data. © 2005 Elsevier B.V. All rights reserved.

KW - B-splines

KW - Difference penalty

KW - Multidimensional array

KW - P-splines

KW - Smoothing

KW - Tensor product

UR - http://www.scopus.com/inward/record.url?scp=24644494711&partnerID=8YFLogxK

U2 - 10.1016/j.csda.2004.07.008

DO - 10.1016/j.csda.2004.07.008

M3 - Article

VL - 50

SP - 61

EP - 76

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

SN - 0167-9473

IS - 1 SPEC. ISS.

ER -