Numerical evaluation of cytologic data. VI. Multivariate distributions and matrix notation.

To represent multivariate data, matrix notation is required. Its equivalence to the arithmetic notation is shown with a bivariate example. With matrix notation one can then readily represent and evaluate multivariate data sets; however, most multivariate procedures require that one find that inverse of the variance-covariance matrix. This is shown using the Cholesky factorization with a numerical example.