The use of the generalized inverse in the general linear statistical model
Ackley, Frederick Roberts, Jr.
Larson, Harold J.
MetadataShow full item record
The general linear model of statistical inference is formulated in terms of the Moore-Penrose generalized inverse. The matrix algebra of the generalized inverse which is essential to the model is presented. A methodology for estimation and hypothesis testing is derived which permits identical manipulation of both the full rank and the less-than-full rank cases of the model.
Approved for public release, distribution is unlimited