|
Activity Number:
|
134
- Bayesian Modeling
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
|
Abstract #318266
|
|
|
Title:
|
A Bayesian Nonparametric Regression with Multiple Periodic Predictors
|
|
Author(s):
|
Duchwan Ryu* and Alan M. Polansky and Devrim Bilgili
|
|
Companies:
|
Northern Illinois University and Northern Illinois University and University of North Florida
|
|
Keywords:
|
ANOVA Decomposition;
High-Dimensional Smoothing;
Spherical Splines;
Tensor Product Splines
|
|
Abstract:
|
The nonparametric regression models with multiple periodic predictors are highly desirable to estimate the shape of objects in three-dimensional space, and to analyze sequential data dominated by more than one cyclic factor. However, their high-dimensional reproducing kernel or hat matrix are computationally difficult to handle and have hindered the utilization. We propose a novel method to estimate the regression functions without computational difficulty by introducing the pulled effects that decompose the regression function into additive regression functions. Through the simulated data we have demonstrated the capability of the proposed method in the estimation of regression functions.
|
Authors who are presenting talks have a * after their name.
