The Heteroscedastic Method: Theory and Applications.
Abstract
Statistical procedures for many problems have in the past usually, in fact almost exclusively, assumed that the random variables observed have equal variance-convariance matrices. In practice the author often finds substantially unequal variance-covariance matrices arising in such practical problems as processing signals (e.g., sonar, radio, etc.) in noise, but classically only imprecise rule-of-thumb methods, or else methods involving some loss of information, were available for these problems. Research was undertaken to develop the general theory for solving such problems and apply it to important problems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1983
- Accession Number
- ADA129525
Entities
People
- Edward J. Dudewicz
Organizations
- Ohio State University