www.pudn.com > Classification_toolbox.part01.rar > calculate_error.m

function [train_err, test_err] = calculate_error (D, train_patterns, train_targets, test_patterns, test_targets, region, Nclasses) 
 
% Calculate error (used by the main calculation functions) 
 
train_err   = zeros(Nclasses+1,1); 
test_err    = zeros(Nclasses+1,1); 
 
if ~isempty(train_targets),     
    [classify, err]  = classification_error(D, train_patterns, train_targets, region); 
    for j = 1:Nclasses, 
        train_err(j)   = 1 - classify(j,j);     
    end 
    if (Nclasses>=2) 
        train_err(Nclasses+1) = err;      
    else 
        train_err(3) = err; 
    end 
end 
 
if ~isempty(test_targets), 
    [classify, err]  = classification_error(D, test_patterns, test_targets, region); 
    for j = 1:Nclasses, 
       test_err(j)   = 1 - classify(j,j);     
    end 
    if (Nclasses>=2) 
        test_err(Nclasses+1) = err;      
    else 
        test_err(3) = err; 
    end 
end