Parallel factoring of strata
Marianna Bolla · 2001
A method for compromise factor extraction from covariance/correlation matrices corresponding to different strata is introduced. Compromise factors are independent and on this constraint they explain the largest possible part of the variables' total variance over the strata. Binary samples are also investigated. The so-called compromise representation of the strata is introduced. A practical application for parallel factoring of medical data in different strata is also presented.