Activity Number:
|
210
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistical Education*
|
Abstract - #300754 |
Title:
|
Classification Under Intraclass Correlation Models by Principal Components Analysis
|
Author(s):
|
Hie-Choon Chung*+ and Chien-Pai Han
|
Affiliation(s):
|
Kwangju University and University of Texas, Arlington
|
Address:
|
592-1 Jinwol-Dong Nam-Gu, Kwangju, , 503-703, South Korea
|
Keywords:
|
Discriminant analysis ; Error rates ; Leave-one-out method ; Principal components ; Intraclass correlation ; Bootstrap method
|
Abstract:
|
The use of principal components to reduce the number of dimensions is a procedure for data presentations but may involve the loss of valuable information for discriminant analysis. In this paper, we consider classification with two multivariate normal populations which have different means and equal intraclass correlation matrices. For the two populations, we consider the cases that the means of the variables within each population are equal or unequal. Two testing procedures in the principal component analysis are compared with a regular classification procedure. Error rates are estimated by bootstrap and leave-one-out methods.
|