1.
两个数据集的代码一样，所以只放了一个dataset
Result:
Data1
Code:
clear; close all;
load(‘H:\\aaa\\1_data1′)
X = Yn’;
[N, I] = size(X);
figure(1);
plot(X(:,1), X(:,2), ‘.’);
pause;
K = 5; % number of clusters
C(:,:,1) = X(1:K,:); % assign the first K points as the means
E = 1; % update error
m = 1;
while (E > 1e-3)
for n = 1:N
dis = sqrt(sum((ones(K,1)*X(n,:) – C(:,:,m)).^2,2));
[min_dis(m,n), ind(m,n)] = min(dis);
end
for k = 1:K
C(k,:,m+1) = mean(X(ind(m,:)==k,:));
end
E = norm(C(:,:,m+1)-C(:,:,m));
% plot the process
figure(1); clf;
color = ‘rbgcmyk’;
for k = 1:K
plot(X(ind(m,:)==k,1),X(ind(m,:)==k,2),[color(k) ‘.’],’MarkerSize’,12);
hold on;
plot(C(k,1,m),C(k,2,m),’kx’, ‘MarkerSize’,12,’LineWidth’,2)
end
% compute the sum of squares
ss(m) = sum(min_dis(m,:).^2);
pause;
m = m+1;
end
figure(2);
plot(ss, ‘o-‘);
title(‘The sum of squared error versus the iteration index’);