Optimal Regression Models for Multivariate Analysis (Factor Analysis).

摘要

Multivariate regression models are presented which are optimally responsive to measured variation in experimental data. Tabulated values of fitting functions are taken from selected eigenvectors of the cross-products matrix. Models are developed for bivariate and trivariate experiments. The lattice and non-lattice situations are examined. Numerical applications include the reduction,interpolation,and graphical presentation of experimental data. Statistical applications originated with the 'factor analysis'techniques employed in experimental psychology. (Author)