|
Activity Number:
|
353
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #304669 |
|
Title:
|
Wavelet-Based Functional Linear Regression
|
|
Author(s):
|
Yihong Zhao*+ and Todd Ogden
|
|
Companies:
|
Columbia University and Columbia University
|
|
Address:
|
, , NY, 10032,
|
|
Keywords:
|
Functional data analysis ; Wavelets ; Variable selection ; Dimension reduction
|
|
Abstract:
|
In linear regression with a functional predictor and a scalar response, we seek to estimate the regression coefficient function linking the response with the functional predictor. We consider the situation in which the functional predictors are observed at N points, where N is much larger than the number of samples. Therefore dimension reduction is necessary to obtain an interpretable model. We propose a novel dimension reduction technique by performing the wavelet transform on the functional predictors. We then fit the regression model using a two-step procedure in the wavelet domain. First, we screen the coefficients to choose a subset of potentially important predictors. Then the coefficient function estimate is obtained using penalized least squares approach. We discuss various possibilities for screening criteria and penalties, as well as methods for determining tuning parameters.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
