cosmo classify knn hdrΒΆ

function predicted=cosmo_classify_knn(samples_train, targets_train, samples_test, opt)
% k-nearest neighbor classifier
%
% predicted=cosmo_classify_nn(samples_train, targets_train, samples_test[, opt])
%
% Inputs:
%   samples_train      PxR training data for P samples and R features
%   targets_train      Px1 training data classes
%   samples_test       QxR test data
%   opt                struct with field:
%     .knn             Number of nearest neighbors to be considered
%     .norm            distance norm (default: 2)
%
% Output:
%   predicted          Qx1 predicted data classes for samples_test. Each
%                      predicted sample is laballed as having the most
%                      samples in the training set within the nearest
%                      opt.knn samples.
%
% Example:
%     ds=cosmo_synthetic_dataset('ntargets',5,'nchunks',15);
%     test_chunk=1;
%     te=cosmo_slice(ds,ds.sa.chunks==test_chunk);
%     tr=cosmo_slice(ds,ds.sa.chunks~=test_chunk);
%     opt=struct();
%     opt.knn=2;
%     pred=cosmo_classify_knn(tr.samples,tr.sa.targets,te.samples,opt);
%     % show targets and predicted labels (40% accuracy)
%     disp([te.sa.targets pred])
%     %||      1     1
%     %||      2     3
%     %||      3     3
%     %||      4     4
%     %||      5     5
%     %
%     opt.norm=1; % city-block distance
%     pred=cosmo_classify_knn(tr.samples,tr.sa.targets,te.samples,opt);
%     % show targets and predicted labels (40% accuracy)
%     disp([te.sa.targets pred])
%     %||      1     1
%     %||      2     3
%     %||      3     3
%     %||      4     4
%     %||      5     5
%
% Notes:
%   - in the case of knn=1, this function is identical to cosmo_classify_nn
%
% See also: cosmo_crossvalidate, cosmo_crossvalidation_measure
%
% #   For CoSMoMVPA's copyright information and license terms,   #
% #   see the COPYING file distributed with CoSMoMVPA.           #