Some Measures for Discriminating between Normal Multivariate Distributions with Equal Covariance Matrices.
Abstract
Suppose that a statistician is permitted access to data which are more precise under (H sub 1) than under (H sub 2) where each hypothesis specifies a multivariate normal distribution. He is also allowed a choice between additional data more precise under (H sub 1) than under (H sub 2) or data in which the reverse is true. In a previous paper it was shown that if a linear discriminant function is used there is a premium on selecting the additional data to be more precise under (H sub 1). In the paper this result is extended to the case where the likelihood-ratio test is used. The results involve several alternate measures for discriminating between normal multivariate distributions with unequal covariance matrices. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 28, 1972
- Accession Number
- AD0750689
Entities
People
- Herman Chernoff
Organizations
- Stanford University