Change-point model selection via AIC

Yoshiyuki Ninomiya

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Change-point problems have been studied for a long time not only because they are needed in various fields but also because change-point models contain an irregularity that requires an alternative to conventional asymptotic theory. The purpose of this study is to derive the AIC for such change-point models. The penalty term of the AIC is twice the asymptotic bias of the maximum log-likelihood, whereas it is twice the number of parameters, $$2p_0$$2p0, in regular models. In change-point models, it is not twice the number of parameters, $$2m+2p_m$$2m+2pm, because of their irregularity, where $$m$$m and $$p_m$$pm are the numbers of the change-points and the other parameters, respectively. In this study, the asymptotic bias is shown to become $$6m+2p_m$$6m+2pm, which is simple enough to conduct an easy change-point model selection. Moreover, the validity of the AIC is demonstrated using simulation studies.

Original languageEnglish
Pages (from-to)943-961
Number of pages19
JournalAnnals of the Institute of Statistical Mathematics
Volume67
Issue number5
DOIs
Publication statusPublished - Oct 26 2015

Fingerprint

Change-point Model
Model Selection
Asymptotic Bias
Irregularity
Change-point Problem
Change Point
p.m.
Asymptotic Theory
Penalty
Likelihood
Simulation Study
Alternatives
Term

All Science Journal Classification (ASJC) codes

  • Statistics and Probability

Cite this

Change-point model selection via AIC. / Ninomiya, Yoshiyuki.

In: Annals of the Institute of Statistical Mathematics, Vol. 67, No. 5, 26.10.2015, p. 943-961.

Research output: Contribution to journalArticle

Ninomiya, Yoshiyuki. / Change-point model selection via AIC. In: Annals of the Institute of Statistical Mathematics. 2015 ; Vol. 67, No. 5. pp. 943-961.
@article{60bb37c8eee0436e913e174a7d57dc7f,
title = "Change-point model selection via AIC",
abstract = "Change-point problems have been studied for a long time not only because they are needed in various fields but also because change-point models contain an irregularity that requires an alternative to conventional asymptotic theory. The purpose of this study is to derive the AIC for such change-point models. The penalty term of the AIC is twice the asymptotic bias of the maximum log-likelihood, whereas it is twice the number of parameters, $$2p_0$$2p0, in regular models. In change-point models, it is not twice the number of parameters, $$2m+2p_m$$2m+2pm, because of their irregularity, where $$m$$m and $$p_m$$pm are the numbers of the change-points and the other parameters, respectively. In this study, the asymptotic bias is shown to become $$6m+2p_m$$6m+2pm, which is simple enough to conduct an easy change-point model selection. Moreover, the validity of the AIC is demonstrated using simulation studies.",
author = "Yoshiyuki Ninomiya",
year = "2015",
month = "10",
day = "26",
doi = "10.1007/s10463-014-0481-x",
language = "English",
volume = "67",
pages = "943--961",
journal = "Annals of the Institute of Statistical Mathematics",
issn = "0020-3157",
publisher = "Springer Netherlands",
number = "5",

}

TY - JOUR

T1 - Change-point model selection via AIC

AU - Ninomiya, Yoshiyuki

PY - 2015/10/26

Y1 - 2015/10/26

N2 - Change-point problems have been studied for a long time not only because they are needed in various fields but also because change-point models contain an irregularity that requires an alternative to conventional asymptotic theory. The purpose of this study is to derive the AIC for such change-point models. The penalty term of the AIC is twice the asymptotic bias of the maximum log-likelihood, whereas it is twice the number of parameters, $$2p_0$$2p0, in regular models. In change-point models, it is not twice the number of parameters, $$2m+2p_m$$2m+2pm, because of their irregularity, where $$m$$m and $$p_m$$pm are the numbers of the change-points and the other parameters, respectively. In this study, the asymptotic bias is shown to become $$6m+2p_m$$6m+2pm, which is simple enough to conduct an easy change-point model selection. Moreover, the validity of the AIC is demonstrated using simulation studies.

AB - Change-point problems have been studied for a long time not only because they are needed in various fields but also because change-point models contain an irregularity that requires an alternative to conventional asymptotic theory. The purpose of this study is to derive the AIC for such change-point models. The penalty term of the AIC is twice the asymptotic bias of the maximum log-likelihood, whereas it is twice the number of parameters, $$2p_0$$2p0, in regular models. In change-point models, it is not twice the number of parameters, $$2m+2p_m$$2m+2pm, because of their irregularity, where $$m$$m and $$p_m$$pm are the numbers of the change-points and the other parameters, respectively. In this study, the asymptotic bias is shown to become $$6m+2p_m$$6m+2pm, which is simple enough to conduct an easy change-point model selection. Moreover, the validity of the AIC is demonstrated using simulation studies.

UR - http://www.scopus.com/inward/record.url?scp=85027916637&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027916637&partnerID=8YFLogxK

U2 - 10.1007/s10463-014-0481-x

DO - 10.1007/s10463-014-0481-x

M3 - Article

AN - SCOPUS:85027916637

VL - 67

SP - 943

EP - 961

JO - Annals of the Institute of Statistical Mathematics

JF - Annals of the Institute of Statistical Mathematics

SN - 0020-3157

IS - 5

ER -