KNNC

K-Nearest Neighbor Classifier

    [W,K,E] = KNNC(A,K)
    [W,K,E] = KNNC(A)

Input
A	Dataset
K	Number of the nearest neighbors (optional; default: K is  optimised with respect to the leave-one-out error on A)

Output
W	k-NN classifier
K	Number of the nearest neighbors used
E	The leave-one-out error of the KNNC

Description

Computation of the K-nearest neighbor classifier for the dataset A.  The resulting classifier W is automatically evaluated by KNN_MAP.

For multi-class problems with many classes (e.g. > 100) evaluation is  slow if K = 1. For that situation the use of FNNC may be considered.

Warning: class prior probabilities in A are neglected.

See also

mappings, datasets, knn_map, fnnc,