How to group variables in a same cluster ?

if I have

A=[1 2; 2 4; 3 4; 4 1; 5 1; 6 4; 7 4; 8 2; 9 1; 10 1]

where 1st column represents customer ID and 2nd column shows Clustering Number. How can I group customer ID into a same cluster such as

cluster1=[4 5 9 10], cluster2=[1,8], cluster3=[ ], cluster4=[2,3,6,7]

Answer 1

[ANKUR KUMAR]

A=[1 2; 2 4; 3 4; 4 1; 5 1; 6 4; 7 4; 8 2; 9 1; 10 1]
id=unique(A(:,2))
for ii=1:length(id)
   iid{ii}=A(find(A(:,2)==id(ii)),1)  
end

you can see the outputs just by entering iid{1}

[Guillaume]

Never mind that this answer is incorrect (try it when the customer id is not equal to the row number such as A = [1234 1;4567 1]), find is usually unnecessary.

If you wanted to use a loop, the body of the loop should be:

for ii = 1 : numel(id)
   iid{ii} = A(A(:, 2) == id(ii), 1)
end

Also, you should preallocate the cell array rather than growing it in the loop, wasting time reallocating a new cell array at each step. So before the loop:

iid = cell(numel(id), 1);
for ...

Or use accumarray which avoids all these issues...

edit: I see the answer has been edited to correct the bug so the answer now works. find is still unnecessary. It just slows the code for no benefit.

[ANKUR KUMAR]

Sir, I am just a student. That was my mistake. Sorry for this.

[Pradya Panyainkaew]

It's fine. Don't worry. I'm a rookie also. It's not your fault. Let's share.

Answer 2

[Guillaume]

Assuming your clustering numbers are all strictly positive integers:

accumarray(A(:, 2), A(:, 1), [], @(custs) {custs})

[Pradya Panyainkaew]

Thank you very much. it is excellent code.