ON SPLINE SMOOTHING WITH AUTOCORRELATED ERRORS

Peter J. Diggle, Michael F. Hutchinson

Research output: Contribution to journalArticle

Abstract

The generalised cross‐validation criterion for choosing the degree of smoothing in spline regression is extended to accommodate an autocorrelated error sequence. It is demonstrated via simulation that the minimum generalised cross‐validation smoothing spline is an inconsistent estimator in the presence of autocorrelated errors and that ignoring even moderate autocorrelation structure can seriously affect the performance of the cross‐validated smoothing spline. The method of penalised maximum likelihood is used to develop an efficient algorithm for the case in which the autocorrelation decays exponentially. An application of the method to a published data‐set is described. The method does not require the data to be equally spaced in time.

Original languageEnglish (US)
Pages (from-to)166-182
Number of pages17
JournalAustralian Journal of Statistics
Volume31
Issue number1
DOIs
StatePublished - 1989
Externally publishedYes

Fingerprint

Spline Smoothing
Generalized Cross-validation
Smoothing Splines
Autocorrelation
Penalized Maximum Likelihood
Regression Splines
Inconsistent
Smoothing
Efficient Algorithms
Decay
Estimator
Simulation

Keywords

  • cross‐validation
  • non‐parametric regression
  • Splinesy
  • time series

ASJC Scopus subject areas

  • Statistics and Probability

Cite this

ON SPLINE SMOOTHING WITH AUTOCORRELATED ERRORS. / Diggle, Peter J.; Hutchinson, Michael F.

In: Australian Journal of Statistics, Vol. 31, No. 1, 1989, p. 166-182.

Research output: Contribution to journalArticle

Diggle, Peter J. ; Hutchinson, Michael F. / ON SPLINE SMOOTHING WITH AUTOCORRELATED ERRORS. In: Australian Journal of Statistics. 1989 ; Vol. 31, No. 1. pp. 166-182.
@article{84853c432c834e098489502affea5bbf,
title = "ON SPLINE SMOOTHING WITH AUTOCORRELATED ERRORS",
abstract = "The generalised cross‐validation criterion for choosing the degree of smoothing in spline regression is extended to accommodate an autocorrelated error sequence. It is demonstrated via simulation that the minimum generalised cross‐validation smoothing spline is an inconsistent estimator in the presence of autocorrelated errors and that ignoring even moderate autocorrelation structure can seriously affect the performance of the cross‐validated smoothing spline. The method of penalised maximum likelihood is used to develop an efficient algorithm for the case in which the autocorrelation decays exponentially. An application of the method to a published data‐set is described. The method does not require the data to be equally spaced in time.",
keywords = "cross‐validation, non‐parametric regression, Splinesy, time series",
author = "Diggle, {Peter J.} and Hutchinson, {Michael F.}",
year = "1989",
doi = "10.1111/j.1467-842X.1989.tb00510.x",
language = "English (US)",
volume = "31",
pages = "166--182",
journal = "Australian and New Zealand Journal of Statistics",
issn = "1369-1473",
publisher = "Wiley-Blackwell",
number = "1",

}

TY - JOUR

T1 - ON SPLINE SMOOTHING WITH AUTOCORRELATED ERRORS

AU - Diggle, Peter J.

AU - Hutchinson, Michael F.

PY - 1989

Y1 - 1989

N2 - The generalised cross‐validation criterion for choosing the degree of smoothing in spline regression is extended to accommodate an autocorrelated error sequence. It is demonstrated via simulation that the minimum generalised cross‐validation smoothing spline is an inconsistent estimator in the presence of autocorrelated errors and that ignoring even moderate autocorrelation structure can seriously affect the performance of the cross‐validated smoothing spline. The method of penalised maximum likelihood is used to develop an efficient algorithm for the case in which the autocorrelation decays exponentially. An application of the method to a published data‐set is described. The method does not require the data to be equally spaced in time.

AB - The generalised cross‐validation criterion for choosing the degree of smoothing in spline regression is extended to accommodate an autocorrelated error sequence. It is demonstrated via simulation that the minimum generalised cross‐validation smoothing spline is an inconsistent estimator in the presence of autocorrelated errors and that ignoring even moderate autocorrelation structure can seriously affect the performance of the cross‐validated smoothing spline. The method of penalised maximum likelihood is used to develop an efficient algorithm for the case in which the autocorrelation decays exponentially. An application of the method to a published data‐set is described. The method does not require the data to be equally spaced in time.

KW - cross‐validation

KW - non‐parametric regression

KW - Splinesy

KW - time series

UR - http://www.scopus.com/inward/record.url?scp=84985512932&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84985512932&partnerID=8YFLogxK

U2 - 10.1111/j.1467-842X.1989.tb00510.x

DO - 10.1111/j.1467-842X.1989.tb00510.x

M3 - Article

AN - SCOPUS:84985512932

VL - 31

SP - 166

EP - 182

JO - Australian and New Zealand Journal of Statistics

JF - Australian and New Zealand Journal of Statistics

SN - 1369-1473

IS - 1

ER -