Far East Journal of Theoretical Statistics
Volume 3, Issue 1, Pages 35 - 48
(July 1999)
|
|
A BAYES CRITERION FOR DETERMINING INFLUENTIAL OBSERVATIONS IN
MDA
Myung-Cheol Kim (Korea) and Hea Jung Kim (Korea)
|
Abstract: This paper suggests a new diagnostic measure for detecting
influential observations In multiple discriminant analysis (MDA). It is
developed from a Bayesian pint of view using a default Bayes factor obtained
from the imaginary training sample methodology. The Bayes factor is taken as a
measure of discriminant criterion for estimating linear discriminant function (LDF).
It is shown that the effect of an observation over the discriminant criterion is
fully explained by the diagnostic measure. As a tool for interpreting the
measure a graphical method is suggested. Performance of the methods is examined
through as illustrative example. |
Keywords and phrases: multiple discriminant analysis,. Diagnostic measure,
influential observation, Bayes criterion, imaginary training sample method,
graphical method. |
|
Number of Downloads: 5 | Number of Views: 886 |
|