Read/Search this Article
Abstract
In this paper, we discuss the problem of statistical classification into one of two multivariate normal populations. Two classification rules known as a test procedure (likelihood ratio criterion) and an estimative method (plug-in method) are compared based on their asymptotic risks when the sample sizes are large and based on the risks obtained by simulation when one of the sample sizes is small. The test procedure appears to be better than the estimative method for reasonable values of the population parameters.
Journal
- Journal of the Japanese Society of Computational Statistics [List of Volumes]
-
Journal of the Japanese Society of Computational Statistics 9(1), 15-29, 1996-12 [Table of Contents]
Japanese Society of Computational Statistics