Minimum Message Length Autoregressive Model Order Selection

Leigh J. Fitzgibbon; David L. Dowe; Farshid Vahid

Minimum Message Length Autoregressive Model Order Selection

Leigh J. Fitzgibbon^*, David L. Dowe, Farshid Vahid

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

21 Citations (Scopus)

Abstract

We derive a Minimum Message Length (MML) estimator for stationary and nonstationary autoregressive models using the Wallace and Freeman (1987) approximation. The MML estimator's model selection performance is empirically compared with AIC, AIC_c, BIC and HQ in a Monte Carlo experiment by uniformly sampling from the autoregressive stationarity region. Generally applicable, uniform priors are used on the coefficients, model order and log σ² for the MML estimator. The experimental results show the MML estimator to have the best overall average mean squared prediction error and best ability to choose the true model order.

Original language	English
Title of host publication	Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004
Editors	M. Palaniswami, C. Chandra Sekhar, G.K. Venayagamoorthy, S. Mohan, M.K. Ghantasala
Pages	439-444
Number of pages	6
Publication status	Published - 2004
Externally published	Yes
Event	Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004 - Chennai, India Duration: 4 Jan 2004 → 7 Jan 2004

Publication series

Name	Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004

Conference

Conference	Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004
Country/Territory	India
City	Chennai
Period	4/01/04 → 7/01/04

Cite this

Fitzgibbon, L. J., Dowe, D. L., & Vahid, F. (2004). Minimum Message Length Autoregressive Model Order Selection. In M. Palaniswami, C. Chandra Sekhar, G. K. Venayagamoorthy, S. Mohan, & M. K. Ghantasala (Eds.), Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004 (pp. 439-444). (Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004).

Fitzgibbon, Leigh J. ; Dowe, David L. ; Vahid, Farshid. / Minimum Message Length Autoregressive Model Order Selection. Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004. editor / M. Palaniswami ; C. Chandra Sekhar ; G.K. Venayagamoorthy ; S. Mohan ; M.K. Ghantasala. 2004. pp. 439-444 (Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004).

@inproceedings{73dffc4029b04e7387d0e61e0ee11b51,

title = "Minimum Message Length Autoregressive Model Order Selection",

abstract = "We derive a Minimum Message Length (MML) estimator for stationary and nonstationary autoregressive models using the Wallace and Freeman (1987) approximation. The MML estimator's model selection performance is empirically compared with AIC, AICc, BIC and HQ in a Monte Carlo experiment by uniformly sampling from the autoregressive stationarity region. Generally applicable, uniform priors are used on the coefficients, model order and log σ2 for the MML estimator. The experimental results show the MML estimator to have the best overall average mean squared prediction error and best ability to choose the true model order.",

keywords = "AR, Autoregression, Bayesian, Information, MML, Minimum Message Length, Order Selection, Time Series",

author = "Fitzgibbon, {Leigh J.} and Dowe, {David L.} and Farshid Vahid",

year = "2004",

language = "English",

isbn = "0780382439",

series = "Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004",

pages = "439--444",

editor = "M. Palaniswami and {Chandra Sekhar}, C. and G.K. Venayagamoorthy and S. Mohan and M.K. Ghantasala",

booktitle = "Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004",

note = "Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004 ; Conference date: 04-01-2004 Through 07-01-2004",

}

Fitzgibbon, LJ, Dowe, DL & Vahid, F 2004, Minimum Message Length Autoregressive Model Order Selection. in M Palaniswami, C Chandra Sekhar, GK Venayagamoorthy, S Mohan & MK Ghantasala (eds), Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004. Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004, pp. 439-444, Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004, Chennai, India, 4/01/04.

Minimum Message Length Autoregressive Model Order Selection. / Fitzgibbon, Leigh J.; Dowe, David L.; Vahid, Farshid.
Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004. ed. / M. Palaniswami; C. Chandra Sekhar; G.K. Venayagamoorthy; S. Mohan; M.K. Ghantasala. 2004. p. 439-444 (Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Minimum Message Length Autoregressive Model Order Selection

AU - Fitzgibbon, Leigh J.

AU - Dowe, David L.

AU - Vahid, Farshid

PY - 2004

Y1 - 2004

N2 - We derive a Minimum Message Length (MML) estimator for stationary and nonstationary autoregressive models using the Wallace and Freeman (1987) approximation. The MML estimator's model selection performance is empirically compared with AIC, AICc, BIC and HQ in a Monte Carlo experiment by uniformly sampling from the autoregressive stationarity region. Generally applicable, uniform priors are used on the coefficients, model order and log σ2 for the MML estimator. The experimental results show the MML estimator to have the best overall average mean squared prediction error and best ability to choose the true model order.

AB - We derive a Minimum Message Length (MML) estimator for stationary and nonstationary autoregressive models using the Wallace and Freeman (1987) approximation. The MML estimator's model selection performance is empirically compared with AIC, AICc, BIC and HQ in a Monte Carlo experiment by uniformly sampling from the autoregressive stationarity region. Generally applicable, uniform priors are used on the coefficients, model order and log σ2 for the MML estimator. The experimental results show the MML estimator to have the best overall average mean squared prediction error and best ability to choose the true model order.

KW - AR

KW - Autoregression

KW - Bayesian

KW - Information

KW - MML

KW - Minimum Message Length

KW - Order Selection

KW - Time Series

UR - http://www.scopus.com/inward/record.url?scp=1942503940&partnerID=8YFLogxK

M3 - Conference contribution

SN - 0780382439

T3 - Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004

SP - 439

EP - 444

BT - Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004

A2 - Palaniswami, M.

A2 - Chandra Sekhar, C.

A2 - Venayagamoorthy, G.K.

A2 - Mohan, S.

A2 - Ghantasala, M.K.

T2 - Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004

Y2 - 4 January 2004 through 7 January 2004

ER -

Fitzgibbon LJ, Dowe DL, Vahid F. Minimum Message Length Autoregressive Model Order Selection. In Palaniswami M, Chandra Sekhar C, Venayagamoorthy GK, Mohan S, Ghantasala MK, editors, Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004. 2004. p. 439-444. (Proceedings of International Conference on Intelligent Sensing and Information Processing, ICISIP 2004).

Minimum Message Length Autoregressive Model Order Selection

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this