A bias correction to the Akaike information criterion, AIC, is derived for regression and autoregressive time series models. The correction is of particular use when the sample size is small, or when the number of fitted parameters is a moderate to large fraction of the sample size. The corrected method, called AICC, is asymptotically efficient if the true model is infinite dimensional. Furthermore, when the true model is of finite dimension, AICC is found to provide better model order choices than any other asymptotically efficient method. Applications to nonstationary autoregressive and mixed autoregressive moving average time series models are also discussed.
%0 Journal Article
%1 hurvich_regression_1989
%A Hurvich, Clifford M.
%A Tsai, Chih-Ling
%D 1989
%J Biometrika
%K AIC, model sample selection, size small
%N 2
%P 297--307
%R 10.1093/biomet/76.2.297
%T Regression and time series model selection in small samples
%U http://biomet.oxfordjournals.org/content/76/2/297
%V 76
%X A bias correction to the Akaike information criterion, AIC, is derived for regression and autoregressive time series models. The correction is of particular use when the sample size is small, or when the number of fitted parameters is a moderate to large fraction of the sample size. The corrected method, called AICC, is asymptotically efficient if the true model is infinite dimensional. Furthermore, when the true model is of finite dimension, AICC is found to provide better model order choices than any other asymptotically efficient method. Applications to nonstationary autoregressive and mixed autoregressive moving average time series models are also discussed.
@article{hurvich_regression_1989,
abstract = {A bias correction to the Akaike information criterion, AIC, is derived for regression and autoregressive time series models. The correction is of particular use when the sample size is small, or when the number of fitted parameters is a moderate to large fraction of the sample size. The corrected method, called AICC, is asymptotically efficient if the true model is infinite dimensional. Furthermore, when the true model is of finite dimension, AICC is found to provide better model order choices than any other asymptotically efficient method. Applications to nonstationary autoregressive and mixed autoregressive moving average time series models are also discussed.},
added-at = {2017-01-09T13:57:26.000+0100},
author = {Hurvich, Clifford M. and Tsai, Chih-Ling},
biburl = {https://www.bibsonomy.org/bibtex/2b2ae03d77ecc4c0dcea0e0ac190b7a16/yourwelcome},
doi = {10.1093/biomet/76.2.297},
interhash = {e674071264c15725d25b2d91ce9a63ec},
intrahash = {b2ae03d77ecc4c0dcea0e0ac190b7a16},
issn = {0006-3444, 1464-3510},
journal = {Biometrika},
keywords = {AIC, model sample selection, size small},
language = {en},
month = jun,
number = 2,
pages = {297--307},
timestamp = {2017-01-09T14:01:11.000+0100},
title = {Regression and time series model selection in small samples},
url = {http://biomet.oxfordjournals.org/content/76/2/297},
urldate = {2016-06-29},
volume = 76,
year = 1989
}