Closed cxd closed 6 years ago
work is progressing on this, the first step is an implementation of the canonical discriminant analysis which is very closely related to the linear discriminant analysis but does not take into account priors and does not estimate probabilities of class membership. it utilises the mahalanobis distance after it extracts the canonical functions and projects both group means and data set into the resulting basis.
The implementation for lda is provided in the canonical discriminant analysis. The second method of classification "classifydiscriminant" provides the decision functions using the projections derived from the original decomposition.
The ground work is in place to implement lda