|
Activity Number:
|
294
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #309654 |
|
Title:
|
Testing for Lack of Dependence in a Functional Linear Model
|
|
Author(s):
|
Inga Maslova*+ and Piort Kokoszka
|
|
Companies:
|
Utah State University and Utah State University
|
|
Address:
|
3900 Old Main Hill, Logan, UT, 84322-3900,
|
|
Keywords:
|
functional ; linear ; model ; test of independence
|
|
Abstract:
|
We study a functional linear model, $Y = \Psi X + \epsilon$ with functional response and explanatory variables; $\Psi$ is a bounded operator in the Hilbert space $L^2$. We propose a simple test of the nullity of $\Psi$ based on the principal component decomposition. The test statistic has asymptotic chi-square distribution, which is also an excellent approximation in finite samples. The methodology is applied to magnetometer data. Records from twelve magnetic observatories during disturbed and quiet periods are analyzed to check if there is functional linear dependence between observations at different latitude stations.
|
