|
Activity Number:
|
162
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 13, 2002 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
WNAR
|
| Abstract - #300367 |
|
Title:
|
Wavelet Estimation of Deterministic Trend
|
|
Author(s):
|
Todd Ogden*+
|
|
Affiliation(s):
|
Columbia University
|
|
Address:
|
PH18W 622 168th Street, New York, New York, 10032, USA
|
|
Keywords:
|
wavelets ; nonparametric regression ; functional data analysis
|
|
Abstract:
|
The model considered here is that each series of observations consists of a deterministic trend component, plus a random "session effect," plus random noise. The primary goals of the analysis are to use data from several sessions to estimate the overall trend component and to model the behavior of the session effects. By first transforming the original data to the wavelet domain, the question of estimating trend can be treated one coefficient at a time as a problem of estimating a mean of several random variables, each of which may include random session effects. Modeling the behavior of random effects is aided by studying the deviations of the wavelet coefficients from their estimated means. The motivating example for this talk consists of data from a psychological study in which the error in judgment of the collinearity of line segments for various angles is recorded.
|
- 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 2002 program |
