KNNSEARCH search for K nearest neighbors and related statistics Usage: IDX = knnsearch(X,Y); finds for each element (row) in Y, the nearest of all elements in X, such that IDX(k) points to X(IDX(k),:) that is nearest to Y(k,:) IDX has as many elements as Y has rows [IDX,DIST] = knnsearch(X,Y); ... = knnsearch(...,'k',k); search for k nearest neighbors (default: k=2) ... = knnsearch(...,'Scale',Scale); Scaling vector of 'seuclidian' metric default value is std(X) ... = knnsearch(...,'Cov',Cov); Cov is the covariance matrix used for Mahalanobis distance default value is cov(X) ... = knnsearch(...,'Distance',Distance); the following distance metrics are currently supported: 'euclidean' [1], 'seuclidean', (scaled euclidian) 'minkowski' [3], 'cityblock' or 'manhattan' [4], 'hamming' [5], 'mahalanobis' [6], 'cosine' [7] (one minus the cosine of the angle between the two samples), 'correlation' (one minus the linear correlation between each pair f data vectors), 'spearman' (one minus the rank correlation between each pair of data vectors), SEE ALSO: corrcoef, spearman, rankcorr, cov, std Reference(s): [1] https://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm [2] https://en.wikipedia.org/wiki/Euclidean_distance [3] https://en.wikipedia.org/wiki/Minkowski_distance [4] https://en.wikipedia.org/wiki/Taxicab_geometry [5] https://en.wikipedia.org/wiki/Hamming_distance [6] https://en.wikipedia.org/wiki/Mahalanobis_distance [7] https://en.wikipedia.org/wiki/Cosine_similarity
Package: nan