clear all
clc

%%

mu_1 = [0 0];
mu_2 = [1 2];
sigma = [.8 0;0 .8];

prior = 0.7; p = 1;q = 1;r = 1;

for i = 1 : 3000
    rnd = unifrnd (0, 1);

    if (rnd < prior)
        data (i, [1 2 3]) = [mvnrnd (mu_1', sigma, 1) 1];
    else
        data (i, [1 2 3]) = [mvnrnd (mu_2', sigma, 1) 2];
    end
end

training_data = data (1 : 2000, :);
class_1 = data (find (training_data (:, 3) == 1), :);
class_2 = data (find (training_data (:, 3) == 2), :);

testdataclassified = data(2001:3000,:);
testset = data(2001:3000,[1 2]);

plot (class_1 (:, 1), class_1 (:, 2), 'g.')
hold on
plot (class_2 (:, 1), class_2 (:, 2), 'r.')

%%

mean_1 = mean (class_1 (:, [1 2]));
mean_2 = mean (class_2 (:, [1 2]));

covariance_1 = cov (class_1 (:, [1 2]));
covariance_2 = cov (class_2 (:, [1 2]));

%%

[x, y] = Euclidclassifier (testset, mean_1, mean_2);
euclidclass = [x, y'];

delta_euclid = (euclidclass (:, [3]) == testdataclassified (:, [3]));
error_euclid = sum (delta_euclid == 0) / 1000 * 100
display (error_euclid);

%%

[x, y] = Mahalnobisclassifier (testset, mean_1, mean_2, covariance_1, covariance_2);
mahalnobisclass = [x y'];

delta_mahalanobis = (mahalnobisclass (:, [3]) == testdataclassified (:, [3]));
error_mahalanobis = sum (delta_mahalanobis == 0) / 1000 * 100
display (error_mahalanobis);