find unique rows of cell array

I have a cell array as attached. I can see that rows 3 and 4 are repeated, so I would like to keep only the first occurrence of this repetition. I have tried a few things, but calling unique(x, 'rows') doesn't work on cell arrays and each row is a different size, so i have found a few issues when trying to cell index.

Thankyou.

Answer 1

[Guillaume]

This is possibly more efficient than the other solutions:

[r1, r2] = ndgrid(1:size(matching__, 1));
duplicates = any(triu(arrayfun(@(r1, r2) isequal(matching__(r1, :), matching__(r2, :)), r1, r2), 1))
matching__(duplicates, :) = []

In your example, only rows 3 and 5 are identical.

[Matlab User]

Thankyou, this is great.

Answer 2

[Jos (10584)]

One, not so elegant option:

% A is your N-by-2 cell array holding arrays of doubles
% convert to strings
C = cellfun(@(x) sprintf('%.99f ',x),A,'un',0) 
C = strcat(C(:,1),C(:,2)) ; 
[~,k] = unique(C,'stable')
uniqueA = A(k,:)

Btw, I only see that rows 3 and 5 are the same ...

[Matlab User]

Thanks for your answer :)

Answer 3

[Titus Edelhofer]

Hi Matlab User,

that's probably not very easily solved. This should work:

i = 1;
while i<size(matching__, 1)
  idx = cellfun(@(x) isequal(x, matching__{i,1}), matching__(i+1:end,1)) & cellfun(@(x) isequal(x, matching__{i,2}), matching__(i+1:end,2));
  if any(idx)
      matching__(find(idx)+i, :) = [];
  end
  i = i + 1;
end

although it's admittedly not very nice (and I hope your matching__ is not too huge).

Titus

[Matlab User]

Thankyou for your answer :)