Varying-Coefficient Single Index Model

• Activity Number: 400
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Feiyang Niu*+ and Jianhui Zhou
• Companies: University of Virginia and University of Virginia
• Keywords: varying-coefficient ; single index ; nonparametric ; canonical correlation ; b spline

Abstract

We consider estimation and variable selection in a flexible varying-coefficient single index model, $y(t)=f(X^T(t)\beta(t), \epsilon)$. The developed nonparametric procedure uses canonical correlation and B-spline approximations to estimate the vector $\beta(t)$, and variables in $X^T(t)$ are selected using group-wise penalty for model interpretability. The developed estimators are easy to compute, and can be generalized to multi-index models. Both simulated and real data are used to illustrate the performance of the developed method.