Online Program Home
My Program

Abstract Details

Activity Number: 654
Type: Contributed
Date/Time: Thursday, August 4, 2016 : 8:30 AM to 10:20 AM
Sponsor: Biometrics Section
Abstract #321069 View Presentation
Title: Two-Step Parsimonious Variable Selection for Right-Censored Survival Time Models
Author(s): Anju Menon* and A. Gregory DiRienzo
Companies: University of Wyoming and
Keywords: Survival ; Variable Selection ; SIS ; multiple hypothesis testing ; prediction error ; parsimonious model

Variable selection is fundamental in any kind of statistical modeling.There has been extensive research by different authors on methods of variable selection from linear regression models to more complex non-linear applications. Modeling survival data especially poses challenges because of a more complicated data structure as the time variable T is usually subject to censoring. This paper presents a two step objective approach to choose between several candidate models based on the the ability of the model to predict survival times using loss functions. Once potentially important variables are selected using a screening method called Iterative Sure Independence Screening (ISIS) the method attempts to select a parsimonious model by using multiple hypothesis testing using generalized family wise error rate to compare models based on estimates of average prediction error. Inverse probability weighted complete case estimator is used for the calculation of average prediction error. Several simulation studies and a real data case analysis is provided .

Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

Copyright © American Statistical Association