CONDITIONING AND CONCENTRATION OF PRINCIPAL COMPONENTS

summary This paper uses various gauges to construct principal variables that satisfy criteria of maximal scatter. The solutions coincide with Hotelling's (1933) principal components in structured ensembles and mixtures, including heavy-tailed distributions not having moments. Thus, normal-theory tests are exact in level and power under nonstandard models allowing for correlated vector observations and for certain mixtures having neither moments nor unimodal marginals.

[1]  H. Hotelling Analysis of a complex of statistical variables into principal components. , 1933 .

[2]  Z. Birnbaum On Random Variables with Comparable Peakedness , 1948 .

[3]  M. Bartlett TESTS OF SIGNIFICANCE IN FACTOR ANALYSIS , 1950 .

[4]  M. Bartlett A Note on the Multiplying Factors for Various χ2 Approximations , 1954 .

[5]  D. Lawley TESTS OF SIGNIFICANCE FOR THE LATENT ROOTS OF COVARIANCE AND CORRELATION MATRICES , 1956 .

[6]  C. L. Mallows Latent vectors of random symmetric matrices , 1961 .

[7]  T. W. Anderson ASYMPTOTIC THEORY FOR PRINCIPAL COMPONENT ANALYSIS , 1963 .

[8]  P. Krishnaiah,et al.  Simultaneous tests for equality of latent roots against certain alternatives-II , 1971 .

[9]  On the Exact Distributions of the Ratios of the Extreme Roots of the Real and Complex Wishart Matrices. , 1974 .

[10]  Irving John Good,et al.  Invariant Distributions Associated with Matrix Laws Under Structural Symmetry , 1981 .

[11]  G. Simons,et al.  On the theory of elliptically contoured distributions , 1981 .

[12]  B. Everitt,et al.  Finite Mixture Distributions , 1981 .

[13]  Geoffrey J. McLachlan,et al.  Mixture models : inference and applications to clustering , 1989 .

[14]  Donald R. Jensen,et al.  The structure and analysis of spherical time-dependent processes , 1989 .

[15]  Vijay K. Rohatgi,et al.  Robustness of statistical tests , 1989 .

[16]  K. Fang,et al.  Generalized Multivariate Analysis , 1990 .

[17]  B. Lindsay Mixture models : theory, geometry, and applications , 1995 .