Multivariate Analysis and Multidimensional Geometry

Many multivariate statistical methods may be regarded as techniques for investigating a sample space in which each sample member is represented by a point. Some aspects of the geometrical interpretation of Principal Components, Principal Coordinates, Canonical Variate and Factor Analysis are discussed, with stress given to implicit assumptions not always realized; some special cases of principal components analysis are included. The angular representation of samples is discussed and the inter‐relationship between the hierarchical representation of a sample and its spatial representation briefly described.