ABSTRACT

In this and following chapters we will deal with covariance structures of multivariate distributions. Principal components, canonical correlations, and factor models are three interrelated concepts dealing with covariance structure. All these concepts aim at reducing the dimension of observable random variables. The principal components will be treated in this chapter. Canonical analysis and Factor analysis will be treated in Chapters 11 and 12 respectively. Though these concepts will be developed for any multivariate population, statistical inferences will be made under the assumption of normality. Proper references will be given for elliptical distributions.