|
Activity Number:
|
501
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #302939 |
|
Title:
|
Functional Additive Modeling and Gradient Estimation
|
|
Author(s):
|
Hans-Georg Müller*+ and Fang Yao
|
|
Companies:
|
University of California, Davis and University of Toronto
|
|
Address:
|
One Shields Avenue, Davis, CA, 95616,
|
|
Keywords:
|
Derivative ; Eigenfunction ; Functional Regression
|
|
Abstract:
|
The functional linear regression model imposes structural constraints which may be restrictive. Completely nonparametric modeling on the other hand encounters a severe form of the curse of dimension due to the infinite dimensional nature of the functional predictors. The proposed functional additive model is both flexible and not subject to the curse of dimension. It is easy to implement, asymptotically consistent and can be extended to the estimation of functional directional derivatives (i.e., derivatives with regard to a predictor function). Functional derivatives indicate how changes in the predictor function in a specified functional direction are associated with corresponding changes in a scalar response. The method is used to construct a representation of the functional gradient field for lifetime fertility of fruit flies in dependence on early life reproductive trajectories.
|
