Relations between information criteria for model-structure selection Part 2. Modelling by shortest data description
作者:
SÁNDORM. VERES,
期刊:
International Journal of Control
(Taylor Available online 1990)
卷期:
Volume 52,
issue 2
页码: 409-421
ISSN:0020-7179
年代: 1990
DOI:10.1080/00207179008953543
出版商: Taylor & Francis Group
数据来源: Taylor
摘要:
The structure selection of stochastic dynamic models by the predictive least squares (PLS) criterion introduced by Rissanen and Wertz (1985) is investigated in this paper in view of its relations to the bayesian estimator and to information criteria with a penalty term. It is shown that in the autoregressive case the strong consistency of the PLS criterion can be derived from the strong consistency of the bayesian order estimator.
点击下载:
PDF (374KB)
返 回