fusion.m

function [ acc, Pred ]=fusion(classifiers, data) 
X = data(:, 1:end-1);
Y = data(:, end);
decisionMatrix = ones(size(X,1), length(classifiers));
index = 1;
for i=1:length(classifiers)
    try
        if strcmp(classifiers{1,i}.name, 'SVM') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'KNN') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'DT') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'NB') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'DISCR') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'ANN') == 1
            decisionMatrix(:,index) = getNNPredict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'NN') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);           
        elseif strcmp(classifiers{1,i}.name, 'GAM') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'GLR') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'RBFNN') == 1
            decisionMatrix(:,index) = getNNPredict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'BAG') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);   
        elseif strcmp(classifiers{1,i}.name, 'SUBSPACE') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);  
        elseif strcmp(classifiers{1,i}.name, 'TOTALBOOST') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);   
        elseif strcmp(classifiers{1,i}.name, 'ADABOOST') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);
        elseif strcmp(classifiers{1,i}.name, 'LPBOOST') == 1
            decisionMatrix(:,index) = predict(classifiers{1,i}.model, X);            
        end        
        index = index + 1;
    catch ME
        disp(sprintf('fusion causing errors: %s',ME.identifier));
    end
end
Pred = decisionMatrix;
allCombined = mode(decisionMatrix,2);
acc = mean(allCombined == Y);
% [Xfpr,Ytpr,~,auc]  = perfcurve(double(Y), double(allCombined), '1','TVals','all','xCrit', 'fpr', 'yCrit', 'tpr');
end