DMCA
Variable Selection with The Modified Buckley–James Method and The Dantzig Selector for High–dimensional Survival Data
BibTeX
@MISC{Hasinur_variableselection,
author = {Md Hasinur and Rahaman Khan and J. Ewart and H. Shaw},
title = {Variable Selection with The Modified Buckley–James Method and The Dantzig Selector for High–dimensional Survival Data},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
We develop a group of algorithms for variable selection using the accelerated failure time (AFT) models that are based on the synthesis of the Buckley–James estimating method and the Dantzig selector. In particular, first two algorithms are based on two modified Buckley–James estimating methods that are developed for high–dimensional data. The last three algorithms are based on a weighted Dantzig selector that uses weights obtained from the two synthesis based algorithms and another obtained from a proposed form. The adaptive Dantzig selector is chosen because it satisfies the oracle properties. The methods are easy to understand and scalable and they do estimation and variable selection simultaneously. They also can deal with collinearity among the covariates and the groups of covariates. We conducted several simulation studies and one empirical analysis with a microarray dataset.
Keyphrases
variable selection dantzig selector high dimensional survival data modified buckley james method adaptive dantzig selector accelerated failure time buckley james several simulation study high dimensional data weighted dantzig selector empirical analysis microarray dataset oracle property modified buckley james
