|
Activity Number:
|
171
- New Nonparametric Methods for Correlated Data
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2018 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
|
Abstract #329196
|
|
|
Title:
|
Sparse Single Index Models for Multivariate Responses
|
|
Author(s):
|
Yuan Feng* and Luo Xiao and Eric Chi
|
|
Companies:
|
North Carolina State University and North Carolina State University and North Carolina State University
|
|
Keywords:
|
multivariate response;
matrix penalization;
spline;
ADMM;
sparsity
|
|
Abstract:
|
Joint models are popular statistical models for multivariate responses data. To address the insufficiency of classical multivariate linear regression models, we propose multivariate single index models, where responses and covariate indexes are linked by unspecified functions. We further incorporate matrix level penalties to select group variables across responses to deal with challenges of high dimensionality. An algorithm based on alternating direction method of multipliers is used for optimization, and the degree of freedom for the estimation procedure is derived for tuning parameter selection. We demonstrate the effectiveness of proposed methods in simulation study and with real data.
|
Authors who are presenting talks have a * after their name.
