function out = kmeans2(m,k)
%% kmeans.m
%% a quick implementation of k-means algorithm
%%
%% input:
%%   - m, data matrix
%%   - k, number of clusters
%%
%% output:
%%   - [m,g], data matrix plus group membership (1..k)
%%
[nr,nc] = size(m)
p = randperm(size(m,1));      % random initialization
for i=1:k
    c(i,:)=m(p(i),:)  
end
tmp = zeros(nr,1)

while 1,
    d = dist(m,c);            % object centroïd distances 
    [z,g] = min(d,[],2);      % group matrix
    if g == tmp, break;
    else tmp = g;
    end
    for i=1:k
    	f = find(g == i)
	if f                  % compute centroïd if not empty set
	   c(i,:) = mean(m(find(g == i),:),1);
	end
    end
end

out = [m,g]