% Extract YTrain from tblTrain
YTrain = tblTrain.Class;

% Extract YValidation from tblValidation
YValidation = tblValidation.Class;

% Extract YTest from tblTest
YTest = tblTest.Class;

layer = 'fc_3';
featuresTrain = activations(net,tblTrain,layer,'OutputAs','rows','ExecutionEnvironment','cpu');
featuresValidation = activations(net,tblValidation,layer,'OutputAs','rows','ExecutionEnvironment','cpu');
featuresTest = activations(net,tblTest,layer,'OutputAs','rows','ExecutionEnvironment','cpu');

trainData=[featuresTrain;featuresValidation]; 
%trainData= reshape(trainData, [49288, 1]);
testData=featuresTest;
%testData= reshape(testData, [8700, 1]);

YTrain = [YTrain;YValidation];
YTrain = grp2idx(YTrain);
YTest = grp2idx(YTest);%testing data
trainlabel=YTrain;            %training labels
testlabel=YTest;                   %classes
attributesData=trainData;          %wine data
% for ii=1:size(attributesData,2)       %normalize the data
%     attributesData(:,ii)=normalize(attributesData(:,ii));
% end
 
X = [trainData;testData]; % Features
Y = [YTrain;YTest]; % Outputs - converted to categorical

% Split train and test data
[train, ~, test] = dividerand(length(Y), 0.8, 0,0.2);
X_train = X(train, :);
Y_train = Y(train);
X_test = X(test, :);
Y_test = Y(test);

% Adjust data


linear = templateSVM('KernelFunction','linear');
linear_svm = @(x, y)fitcecoc(x, y, 'Learners', linear);
gaussian = templateSVM('KernelFunction','gaussian');
gaussian_svm = @(x, y)fitcecoc(x, y, 'Learners', gaussian);
knn1 = @(x, y)fitcknn(x, y, 'NumNeighbors', 1);
knn3 = @(x, y)fitcknn(x, y, 'NumNeighbors', 3);
knn5 = @(x, y)fitcknn(x, y, 'NumNeighbors', 5);
tree = @(x, y)fitctree(x, y);
Bayes= @(x, y)fitcnb(x,y);
discriminant=@(x, y)fitcdiscr(x,y);
%learners = {linear_svm, gaussian_svm, knn1, knn3, knn5, tree,Bayes,discriminant};
learners = {linear_svm, gaussian_svm, knn1, knn3, knn5, tree,Bayes};
%% Single Model

mdl = tree(X_train, Y_train); % Train Model
y_ens = mdl.predict(X_test); % Predict
Y_test1 = grp2idx(Y_test);
y_ens = grp2idx(y_ens);
[c_matrixp333,Result233]= confusion.getMatrix(Y_test1,y_ens);fprintf("---------------------\n");
fprintf("Decision Tree: %.2f%%\n", ...
    100 * sum(y_ens == Y_test) / length(Y_test));
fprintf("---------------------\n");

mdl1666 = knn5(X_train, Y_train); % Train Model
y_ens199 = mdl1666.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens199 = grp2idx(y_ens199);
[c_matrixp3335,Result2335]= confusion.getMatrix(Y_test1,y_ens199);
% Print result
fprintf("---------------------\n");
fprintf("Decision knn5: %.2f%%\n", ...
    100 * sum(y_ens199 == Y_test) / length(Y_test));
fprintf("---------------------\n");

mdl2 = knn3(X_train, Y_train); % Train Model
y_ens28 = mdl2.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens28 = grp2idx(y_ens28);
[c_matrixp33356,Result23356]= confusion.getMatrix(Y_test1,y_ens28);
% Print result
fprintf("---------------------\n");
fprintf("Decision knn3: %.2f%%\n", ...
    100 * sum(y_ens28 == Y_test) / length(Y_test));
fprintf("---------------------\n");

mdl3 = knn1(X_train, Y_train); % Train Model
y_ens380 = mdl3.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens380 = grp2idx(y_ens380);
[c_matrixp33359680,Result23356980]= confusion.getMatrix(Y_test1,y_ens380);
% Print result
fprintf("---------------------\n");
fprintf("Decision knn1: %.2f%%\n", ...
    100 * sum(y_ens380 == Y_test) / length(Y_test));
fprintf("---------------------\n");

mdl4 = gaussian_svm(X_train, Y_train); % Train Model
y_ens4900 = mdl4.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens4900 = grp2idx(y_ens4900);
[c_matrixp33359786,Result2335699]= confusion.getMatrix(Y_test1,y_ens4900);% Print result
fprintf("---------------------\n");
fprintf("Decision gaussian_svm: %.2f%%\n", ...
    100 * sum(y_ens4900 == Y_test) / length(Y_test));
fprintf("---------------------\n");


mdl5 =linear_svm(X_train, Y_train); % Train Model
y_ens59009 = mdl5.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens59009 = grp2idx(y_ens59009);
[c_matrixp33359606,Result2335609]= confusion.getMatrix(Y_test1,y_ens59009);% Print result
% Print result
fprintf("---------------------\n");
fprintf("Decision linear_svm: %.2f%%\n", ...
    100 * sum(y_ens59009 == Y_test) / length(Y_test));
fprintf("---------------------\n");

mdl6 =Bayes(X_train, Y_train); % Train Model
y_ens6908 = mdl6.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens6908 = grp2idx(y_ens6908);
[c_matrixp33300596,Result27033569]= confusion.getMatrix(Y_test1,y_ens6908);% Print result
% Print result
fprintf("---------------------\n");
fprintf("Decision Bayes: %.2f%%\n", ...
    100 * sum(y_ens6908 == Y_test) / length(Y_test));
fprintf("---------------------\n");

mdl7 =discriminant(X_train, Y_train); % Train Model
y_ens7900 = mdl7.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_ens7900 = grp2idx(y_ens7900);
[c_matrixp333596,Result233569]= confusion.getMatrix(Y_test1,y_ens7900);% Print result
% Print result
fprintf("---------------------\n");
fprintf("Decision discriminant: %.2f%%\n", ...
    100 * sum(y_ens7900 == Y_test) / length(Y_test));
fprintf("---------------------\n");


learners = {tree, knn5};
ensggg = classification_ensemble(learners); % Initialize Ensemble
ensggg = ensggg.fit(X_train, Y_train); % Train Ensemble
y_enshhh = ensggg.predict(X_test); % Predict
%Y_test1 = grp2idx(Y_test);
y_enshhh = grp2idx(y_enshhh);
[c_matrixp,Result2]= confusion.getMatrix(Y_test1,y_enshhh);