k-means 算法matlab实现

function [gIdx,c]=k_means(X,k)
[n,m]=size(X);
g0=ones(n,1);
gIdx=zeros(n,1);
D=zeros(n,k);
c=X(ceil(rand(k,1)*n),:);
while any(g0~=gIdx)
    g0=gIdx;
    for t=1:k
        d=zeros(n,1);
        for s=1:m
            d=d+(X(:,s)-c(t,s)).^2;
        end
        D(:,t)=d;
    end
    [z,gIdx]=min(D,[],2);
    for t=1:k
        c(t,:)=mean(X(gIdx==t,:));
    end

end

输出gIdx为分类的数组，c为最后迭代的均值点。k为分组数。