Online Program Home
  My Program

Abstract Details

Activity Number: 435 - Introductory Overview Lecture: Quantile Regression
Type: Invited
Date/Time: Wednesday, August 2, 2017 : 8:30 AM to 10:20 AM
Sponsor: JSM Partner Societies
Abstract #325065
Title: Computer-Intensive Methods for Inference
Author(s): Xuming He*
Companies: University of Michigan
Keywords:
Abstract:

We provide an overview of two computer-intensive approaches for interval estimation of quantile regression parameters. The first is resampling-based methods that use the signs of residuals as exchangeable units. This exchangeability holds in quantile regression models that accommodate heterogeneity in the data, and permits several forms of resampling schemes with attractive computation. The second is Bayesian quantile regression where an efficient Gibbs sampler can be used to obtain posterior intervals. We show that the validity of the posterior inference requires an adjustment in the posterior variance calculation.


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

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association