JSM 2012 Home

JSM 2012 Online Program

The views expressed here are those of the individual authors and not necessarily those of the JSM sponsors, their officers, or their staff.

Online Program Home

Abstract Details

Activity Number: 462
Type: Contributed
Date/Time: Wednesday, August 1, 2012 : 8:30 AM to 10:20 AM
Sponsor: Section on Nonparametric Statistics
Abstract - #306269
Title: Bootstrap Variance Estimates for Bagged Predictors
Author(s): Dennis D. Boos*+ and Jiangtao Duan and Leonard Stefanski
Companies: North Carolina State University and BBVA Compass and North Carolina State University
Address: Department of Statistics, Raleigh, NC, 27695-8203, United States
Keywords: double bootstrap ; jackknife-after-bootstrap ; residual-based bootstrap ; random-pair bootstrap
Abstract:

Bagging, averaging over bootstrap replications, is a method to improve predictions based on procedures that are known to be sensitive to small changes in the observed data. Two bootstrap methods are introduced for estimating the variance of bagged predictions in linear regression models when the base estimation method uses variable selection. A key feature of both methods is the use of the residual-based bootstrap rather than the standard random-pair bootstrap. The first method is called the bootstrap-after-bootstrap (BaB) because two levels of bootstrapping are employed, the residual-based bootstrap in the outer loop and the random-pair bootstrap in the inner loop. Although very computationally intensive, this method performs nearly unbiasedly. The second method is called the parallel bootstrap (PB) because only a second set of B resamples are required in parallel to the original B resamples used to define the bagged estimator. It tends to be slightly biased upwards, but is computationally attactive and has low coefficient of variation compared to the BaB method. Simulation comparisons are made using the adaptive LASSO and forward selection.


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 2012 program




2012 JSM Online Program Home

For information, contact jsm@amstat.org or phone (888) 231-3473.

If you have questions about the Continuing Education program, please contact the Education Department.