|
Activity Number:
|
452
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #302756 |
|
Title:
|
Inference Post-Model Selection: The Good, The Bad, and The Ugly
|
|
Author(s):
|
Hannes Leeb*+
|
|
Companies:
|
Yale University
|
|
Address:
|
Department of Statistics , New Haven, CT, 06510,
|
|
Keywords:
|
Model selection ; inference ; ill-posed problems ; shrinkage methods ; adaption ; confidence and prediction intervals
|
|
Abstract:
|
In situations where a data-driven model selection step precedes subsequent model fitting, we study the potential, the limitations, and also the caveats of inference procedures that are based on the selected model. In particular, we review some insurmountable obstructions to inference in such settings, discuss certain myths about the performance of post-model-selection estimators, and describe scenarios in which inference procedures post model selection can perform well.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
