Consistent bootstrap procedures for post-model-selection inference under linear regression models


Grant Data
Project Title
Consistent bootstrap procedures for post-model-selection inference under linear regression models
Principal Investigator
Professor Lee, Stephen Man Sing   (Principal investigator)
Duration
36
Start Date
2016-01-01
Completion Date
2018-12-31
Amount
631972
Conference Title
Presentation Title
Keywords
post-model-selection, bootstrap, regression
Discipline
Probability & Statistics
Panel
Physical Sciences (P)
Sponsor
RGC General Research Fund (GRF)
HKU Project Code
17303715
Grant Type
General Research Fund (GRF)
Funding Year
2015/2016
Status
On-going
Objectives
2 Develop a bootstrap procedure for estimating distributions of post-model-selection least squares estimators. 3 Study the asymptotic properties, and prove consistency, of the procedure under a moving-parameter asymptotic framework, which facilitates a theoretical investigation of greater practical relevance. 4 Conduct empirical studies to compare the proposed procedure with existing methods in standard inference applications, under various model selection settings including, in particular, those most susceptible to inconsistency problems.