FNNC

Fast Nearest Neighbor Classifier

    W = A*FNNC(LOO)
    W = FNNC(A,LOO)

Description

This is a fast version of KNNC([],1), especially useful for many-class  problems. For reasons of speed it will NOT return proper confidences on  evaluation. D = B*W will be a 0/1 dataset. Classification results are  otherwise identical to KNNC([],1).

LOO is a logical that should be TRUE in case W is applied to the same  dataset (A) to enforce Leave-One-Out processing. Default LOO is FALSE.

See also

mappings, datasets, knnc, testk,