NMSC

Nearest Mean Scaled Classifier

    W = NMSC(A)
    W = A*NMSC

Input
A	Trainign dataset

Output
W	Nearest Mean Scaled Classifier mapping

Description

Computation of the linear discriminant for the classes in the dataset A assuming normal distributions with zero covariances and equal class variances.  The use of soft labels is supported.

The difference with NMC is that NMSC is based on an assumption of normal  distributions and thereby automatically scales the features and is  sensitive to class priors. NMC is a plain nearest mean classifier that is  feature scaling sensitive and unsensitive to class priors.

As NMSC is a linear classifier, a non-linear combiner might give an  improvement in multi-dimensional problems, e.g. by  W = A*(NMC*QDC([],[],1e-6)).

See also

datasets, mappings, nmc, ldc, fisherc, qdc, udc,