Autoregressive model selection based on a prediction perspective

Yun Huan Lee, Chun Shu Chen

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

The autoregressive (AR) model is a popular method for fitting and prediction in analyzing time-dependent data, where selecting an accurate model among considered orders is a crucial issue. Two commonly used selection criteria are the Akaike information criterion and the Bayesian information criterion. However, the two criteria are known to suffer potential problems regarding overfit and underfit, respectively. Therefore, using them would perform well in some situations, but poorly in others. In this paper, we propose a new criterion in terms of the prediction perspective based on the concept of generalized degrees of freedom for AR model selection. We derive an approximately unbiased estimator of mean-squared prediction errors based on a data perturbation technique for selecting the order parameter, where the estimation uncertainty involved in a modeling procedure is considered. Some numerical experiments are performed to illustrate the superiority of the proposed method over some commonly used order selection criteria. Finally, the methodology is applied to a real data example to predict the weekly rate of return on the stock price of Taiwan Semiconductor Manufacturing Company and the results indicate that the proposed method is satisfactory.

Original languageEnglish
Pages (from-to)913-922
Number of pages10
JournalJournal of Applied Statistics
Volume39
Issue number4
DOIs
Publication statusPublished - 2012 Apr 1

Fingerprint

Autoregressive Model
Model Selection
Prediction
Data Perturbation
Uncertainty Estimation
Order Selection
Bayesian Information Criterion
Semiconductor Manufacturing
Akaike Information Criterion
Potential Problems
Unbiased estimator
Perturbation Technique
Dependent Data
Stock Prices
Taiwan
Prediction Error
Mean Squared Error
Order Parameter
Degree of freedom
Numerical Experiment

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

@article{ceeb347015f640499e75eb6dc73fc6ee,
title = "Autoregressive model selection based on a prediction perspective",
abstract = "The autoregressive (AR) model is a popular method for fitting and prediction in analyzing time-dependent data, where selecting an accurate model among considered orders is a crucial issue. Two commonly used selection criteria are the Akaike information criterion and the Bayesian information criterion. However, the two criteria are known to suffer potential problems regarding overfit and underfit, respectively. Therefore, using them would perform well in some situations, but poorly in others. In this paper, we propose a new criterion in terms of the prediction perspective based on the concept of generalized degrees of freedom for AR model selection. We derive an approximately unbiased estimator of mean-squared prediction errors based on a data perturbation technique for selecting the order parameter, where the estimation uncertainty involved in a modeling procedure is considered. Some numerical experiments are performed to illustrate the superiority of the proposed method over some commonly used order selection criteria. Finally, the methodology is applied to a real data example to predict the weekly rate of return on the stock price of Taiwan Semiconductor Manufacturing Company and the results indicate that the proposed method is satisfactory.",
author = "Lee, {Yun Huan} and Chen, {Chun Shu}",
year = "2012",
month = "4",
day = "1",
doi = "10.1080/02664763.2011.636418",
language = "English",
volume = "39",
pages = "913--922",
journal = "Journal of Applied Statistics",
issn = "0266-4763",
publisher = "Routledge",
number = "4",

}

Autoregressive model selection based on a prediction perspective. / Lee, Yun Huan; Chen, Chun Shu.

In: Journal of Applied Statistics, Vol. 39, No. 4, 01.04.2012, p. 913-922.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Autoregressive model selection based on a prediction perspective

AU - Lee, Yun Huan

AU - Chen, Chun Shu

PY - 2012/4/1

Y1 - 2012/4/1

N2 - The autoregressive (AR) model is a popular method for fitting and prediction in analyzing time-dependent data, where selecting an accurate model among considered orders is a crucial issue. Two commonly used selection criteria are the Akaike information criterion and the Bayesian information criterion. However, the two criteria are known to suffer potential problems regarding overfit and underfit, respectively. Therefore, using them would perform well in some situations, but poorly in others. In this paper, we propose a new criterion in terms of the prediction perspective based on the concept of generalized degrees of freedom for AR model selection. We derive an approximately unbiased estimator of mean-squared prediction errors based on a data perturbation technique for selecting the order parameter, where the estimation uncertainty involved in a modeling procedure is considered. Some numerical experiments are performed to illustrate the superiority of the proposed method over some commonly used order selection criteria. Finally, the methodology is applied to a real data example to predict the weekly rate of return on the stock price of Taiwan Semiconductor Manufacturing Company and the results indicate that the proposed method is satisfactory.

AB - The autoregressive (AR) model is a popular method for fitting and prediction in analyzing time-dependent data, where selecting an accurate model among considered orders is a crucial issue. Two commonly used selection criteria are the Akaike information criterion and the Bayesian information criterion. However, the two criteria are known to suffer potential problems regarding overfit and underfit, respectively. Therefore, using them would perform well in some situations, but poorly in others. In this paper, we propose a new criterion in terms of the prediction perspective based on the concept of generalized degrees of freedom for AR model selection. We derive an approximately unbiased estimator of mean-squared prediction errors based on a data perturbation technique for selecting the order parameter, where the estimation uncertainty involved in a modeling procedure is considered. Some numerical experiments are performed to illustrate the superiority of the proposed method over some commonly used order selection criteria. Finally, the methodology is applied to a real data example to predict the weekly rate of return on the stock price of Taiwan Semiconductor Manufacturing Company and the results indicate that the proposed method is satisfactory.

UR - http://www.scopus.com/inward/record.url?scp=84859188968&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84859188968&partnerID=8YFLogxK

U2 - 10.1080/02664763.2011.636418

DO - 10.1080/02664763.2011.636418

M3 - Article

AN - SCOPUS:84859188968

VL - 39

SP - 913

EP - 922

JO - Journal of Applied Statistics

JF - Journal of Applied Statistics

SN - 0266-4763

IS - 4

ER -