An Akaike-type information criterion for model selection under inequality constraints

R Kuiper, Herbert Hoijtink, Mervyn Silvapulle

Research output: Contribution to journalArticleResearchpeer-review

Abstract

The Akaike information criterion for model selection presupposes that the parameter space is not subject to order restrictions or inequality constraints. Anraku (1999) proposed a modified version of this criterion, called the order-restricted information criterion, for model selection in the one-way analysis of variance model when the population means are monotonic. We propose a generalization of this to the case when the population means may be restricted by a mixture of linear equality and inequality constraints. If the model has no inequality constraints, then the generalized order-restricted information criterion coincides with the Akaike information criterion. Thus, the former extends the applicability of the latter to model selection in multi-way analysis of variance models when some models may have inequality constraints while others may not. Simulation shows that the information criterion proposed in this paper performs well in selecting the correct model.
Original languageEnglish
Pages (from-to)495 - 501
Number of pages7
JournalBiometrika
Volume98
Issue number2
DOIs
Publication statusPublished - 2011

Cite this

Kuiper, R ; Hoijtink, Herbert ; Silvapulle, Mervyn. / An Akaike-type information criterion for model selection under inequality constraints. In: Biometrika. 2011 ; Vol. 98, No. 2. pp. 495 - 501.
@article{f1c12fb6f98c4d0ba3adcc47d4e4595c,
title = "An Akaike-type information criterion for model selection under inequality constraints",
abstract = "The Akaike information criterion for model selection presupposes that the parameter space is not subject to order restrictions or inequality constraints. Anraku (1999) proposed a modified version of this criterion, called the order-restricted information criterion, for model selection in the one-way analysis of variance model when the population means are monotonic. We propose a generalization of this to the case when the population means may be restricted by a mixture of linear equality and inequality constraints. If the model has no inequality constraints, then the generalized order-restricted information criterion coincides with the Akaike information criterion. Thus, the former extends the applicability of the latter to model selection in multi-way analysis of variance models when some models may have inequality constraints while others may not. Simulation shows that the information criterion proposed in this paper performs well in selecting the correct model.",
author = "R Kuiper and Herbert Hoijtink and Mervyn Silvapulle",
year = "2011",
doi = "10.1093/biomet/asr002",
language = "English",
volume = "98",
pages = "495 -- 501",
journal = "Biometrika",
issn = "0006-3444",
publisher = "Oxford University Press",
number = "2",

}

An Akaike-type information criterion for model selection under inequality constraints. / Kuiper, R; Hoijtink, Herbert; Silvapulle, Mervyn.

In: Biometrika, Vol. 98, No. 2, 2011, p. 495 - 501.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - An Akaike-type information criterion for model selection under inequality constraints

AU - Kuiper, R

AU - Hoijtink, Herbert

AU - Silvapulle, Mervyn

PY - 2011

Y1 - 2011

N2 - The Akaike information criterion for model selection presupposes that the parameter space is not subject to order restrictions or inequality constraints. Anraku (1999) proposed a modified version of this criterion, called the order-restricted information criterion, for model selection in the one-way analysis of variance model when the population means are monotonic. We propose a generalization of this to the case when the population means may be restricted by a mixture of linear equality and inequality constraints. If the model has no inequality constraints, then the generalized order-restricted information criterion coincides with the Akaike information criterion. Thus, the former extends the applicability of the latter to model selection in multi-way analysis of variance models when some models may have inequality constraints while others may not. Simulation shows that the information criterion proposed in this paper performs well in selecting the correct model.

AB - The Akaike information criterion for model selection presupposes that the parameter space is not subject to order restrictions or inequality constraints. Anraku (1999) proposed a modified version of this criterion, called the order-restricted information criterion, for model selection in the one-way analysis of variance model when the population means are monotonic. We propose a generalization of this to the case when the population means may be restricted by a mixture of linear equality and inequality constraints. If the model has no inequality constraints, then the generalized order-restricted information criterion coincides with the Akaike information criterion. Thus, the former extends the applicability of the latter to model selection in multi-way analysis of variance models when some models may have inequality constraints while others may not. Simulation shows that the information criterion proposed in this paper performs well in selecting the correct model.

U2 - 10.1093/biomet/asr002

DO - 10.1093/biomet/asr002

M3 - Article

VL - 98

SP - 495

EP - 501

JO - Biometrika

JF - Biometrika

SN - 0006-3444

IS - 2

ER -