MEANCOV

Estimation of the means and covariances from multiclass data

    [U,G] = MEANCOV(A,N)

   INPUT
    A Dataset
    N Normalization to use for calculating covariances: by M, the number
     of samples in A (N = 1) or by M-1 (default, unbiased, N = 0).

Output
U	Mean vectors
G	Covariance matrices

Description

Computation of a set of mean vectors U and a set of covariance matrices G of the C classes in the dataset A. The covariance matrices are stored as a  3-dimensional matrix G of the size K x K x C, the class mean vectors as a  labeled dataset U of the size C x K.

The use of soft labels or target labels is supported.

See also

datasets, nbayesc, distmaha,