SUBSC

Subspace Classifier

    W = SUBSC(A,N)
    W = SUBSC(A,FRAC)

Input
A	Dataset
N	or FRAC Desired model dimensionality or fraction of retained  variance per class

Output
W	Subspace classifier

Description

Each class in the trainingset A is described by linear subspace of  dimensionality N, or such that at least a fraction FRAC of its variance  is retained. This is realised by calling PCA(AI,N) or PCA(AI,FRAC) for  each subset AI of A (objects of class I). For each class a model is  built that assumes that the distances of the objects to the class  subspaces follow a one-dimensional distribution.

New objects are assigned to the class of the nearest subspace.  Classification by D = B*W, in which W is a trained subspace classifier  and B is a testset, returns a dataset D with one-dimensional densities  for each of the classes in its columns.

If N (ALF) is NaN it is optimised by REGOPTC.

Reference(s)

E. Oja, The Subspace Methods of Pattern Recognition, Wiley, New York, 1984.

See also

datasets, mappings, pca, fisherc, fisherm, gaussm, regoptc,