what are the meanings of every ROW

[pro_octave]

I am a newcomer for Machine Learning, below is the algorithm of k-Medoids
clustering in Octave, but I am confused by the meanings of every VARs and
ROWs, here is the CODE, is there anyone can explain for me, I will be very
appreciate, please note that with more DETAILS



function [Medoids, clusvector, DIST]=clusterKMedoids(M, N)

[points, dim]=size(M);

% disp('calculating distances');
DIST=zeros(points, points);
        for i=1:points-1
                for j=i+1:points
                        DIST(i, j)=norm(M(i, :)-M(j, :), 2);
                        DIST(j, i)=DIST(i, j);
                end
        end

%compute medoids

Medoids=[];%
sumdist=sum(DIST);%
[Min Medoids(1)] = min(sumdist);%

for k=2:N %
        notselected=setdiff([1:points], Medoids);%
        Cj=zeros(points,1);%
        for i=notselected%
                for j=notselected(find(notselected~=i))
                %for j=not yet select
                        Dj=min(DIST(j, Medoids));
                        dij=DIST(j, i);
                        Cj(i)=Cj(i)+max([Dj-dij 0]);
                end
        end
        [Max Medoids(k)]=max(Cj);%selected ones have score zero
end


% do swap
notselected=setdiff([1:points], Medoids);
ns=points-N; %

minT=-1;
T=zeros(N, ns);%
iter=0;
sumD=[];%

while (minT<0 && iter<=20)
        for i=Medoids
                for h=notselected
                        Cjih=0;
                        for j=union(notselected, i)
                                Dj=min(DIST(j, Medoids));
                                dij=DIST(j,i);
                                dhj=DIST(j,h);
                                if Dj==dij
                                        Ej=min(DIST(j, 
Medoids(find(Medoids~=i))));
                                        if dhj<Ej
                                                Cjih=Cjih+dhj-dij;
                                        else
                                                Cjih=Cjih+Ej-Dj;
                                        end
                                elseif Dj>dhj
                                        Cjih=Cjih+dhj-Dj;
                                end
                        end
                        T(find(Medoids==i),find(notselected==h))=Cjih;
                end
        end

        [minT, Ind]=min(T(:));
        if minT<0
                colh=ceil(Ind/N);
                rowi=Ind-(colh-1)*N;
                i=Medoids(rowi);
                Medoids(rowi)=notselected(colh);
                notselected(colh)=i;
        end
        iter=iter+1;
        sumiter=0;

        for j=notselected
                sumiter=sumiter+min(DIST(j, Medoids));
        end
        
        sumD=[sumD sumiter];
end%END for WHILE

clusvector=zeros(points,1);

for i=Medoids
        clusvector(i, 1)=find(Medoids==i);
end
for j=notselected
        [Dj Ind]=min(DIST(j, Medoids));
        clusvector(j, 1)=Ind;
end

end



--
View this message in context: 
http://octave.1599824.n4.nabble.com/what-are-the-meanings-of-every-ROW-tp4655577.html
Sent from the Octave - General mailing list archive at Nabble.com.