Activity Number:
|
256
- Contributed Poster Presentations: Section on Bayesian Statistical Science
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Bayesian Statistical Science
|
Abstract #330580
|
|
Title:
|
Bayesian Functional Quantile Regression
|
Author(s):
|
Yusha Liu* and Jeffrey S Morris and Meng Li
|
Companies:
|
Rice University and The University of Texas M.D. Anderson Cancer Center and Rice University
|
Keywords:
|
Bayesian modeling;
quantile regression;
shrinkage priors
|
Abstract:
|
We propose a unified Bayesian functional quantile regression framework which simultaneously performs quantile regression and adaptively regularizes the functional coefficients by projection into a basis space and employing an appropriate shrinkage prior on the basis coefficients. Our framework is highly general in that any basis functions and shrinkage priors can be chosen, depending on the characteristics of the functional data to be analyzed. We developed an efficient Gibbs sampler to fit this fully Bayesian hierarchical model automatically with no tuning required, which yield posterior samples that can be used to perform pointwise or joint inference on any model quantity of interest. Our approach is computationally efficient and can handle functional data observed on grids of hundreds to thousands.
|
Authors who are presenting talks have a * after their name.