Model Selection with Information Criteria

This thesis is on model selection using information criteria. The information criteria include generalized information criterion and a family of Bayesian information criteria. The properties and improvement of the information criteria are investigated. We analyze nonasymptotic and asymptotic properties of the information criteria for linear models, probabilistic models, and high dimensional models, respectively. We give probability of selecting a model and compute the probability by Monte Carlo methods. We derive the conditions under which the criteria are consistent, underfitting, or overfitting. We further propose new model selection procedures to improve the information criteria. The procedures combine the information criteria with the probability of selecting a model and overfitting level, respectively. In addition, we develop model selection software packages in R and examine applications to real data.

Name thesis:degree_name

Ph D

Discipline thesis:degree_discipline

Statistics and Actuarial Sciences

Grantor dc:publisher

The University of Western Ontario

Year dc:date.issued

2010

Author dc:creator

• Xu, Changjiang

Advisor dc:contributor.advisor

• McLeod, A. Ian

• Statistical modeling
• model selection
• variable selection
• penalized likelihood
• model selection criterion
• information criteria

Language dc:language.iso

en_ca

Handle dc:identifier.uri

https://hdl.handle.net/20.500.14721/34724

OAI identifier oai:identifier

oai:uwo.scholaris.ca:20.500.14721/34724