How to suffle rows of specific rows from a matrix

Hello I have a matrix with 2 columns and 120 rows. the first columns has numbers 2-6 and the second one is 0 and 1 like the following

[2 1]
[3 0]
[2 0]
[4 0]
[6 1]
[2 1] etc.

I want to shuffle randomly the second column to produce a new column. In this new column I want to shuffle the rows of the second column, that their first column is between 2 and 5. Also shuffle the rows on the second column that the first column is 6 between them. More specifically if I have a sequence in the following way

[2 1] [3 0] [6 0] [2 0]

one new shuffled column will make the matrix

[2 0 0] [3 0 1] [6 0 0] [2 0 0]

but NOT

[2 0 0] [3 0 0] [6 0 1] [2 0 0]

because i don't wont to shuffle the rows that have 6 with the rows that have 2-5. I want with this process to create 100 new columns and to do this 2008 times with different matrixes of the same size. All this shuffling has to be in a random way except for the condition of 2-5 and 6 that I mentioned.

Answer 1

[the cyclist]

I think this does what you want. I wrote it from more of a didactic than efficiency approach. I imagine there are some efficiencies here, but one should generally get a working solution first, then consider optimization.

NCOL = 100;

M = [2 1;
     3 0;
     2 0;
     4 0;
     6 1;
     2 1;
     6 0];

M_new = [M, zeros(size(M,1),NCOL)];

isRowSix = M(:,1)==6;
isRowNot = M(:,1)~=6;

M2Six = M(isRowSix,2);
M2Not = M(isRowNot,2);

numberRowsSix = sum(isRowSix);
numberRowsNot = sum(isRowNot);

for nc = 1:NCOL

      newOrderSix = randperm(numberRowsSix);
      newOrderNot = randperm(numberRowsNot);

      newColSix = M2Six(newOrderSix);
      newColNot = M2Not(newOrderNot);

      M_new(isRowSix,nc+2) = newColSix;
      M_new(isRowNot,nc+2) = newColNot;

end

[Georgios Tertikas]

thanx for the answer could you please explain what you do in each step?

[the cyclist]

% The number of additional columns
NCOL = 100;

% The original data
M = [2 1;
     3 0;
     2 0;
     4 0;
     6 1;
     2 1;
     6 0];

% Preallocate the new array
M_new = [M, zeros(size(M,1),NCOL)];

% Identify the rows that have a 6 in the first column (and those that do not)
isRowSix = M(:,1)==6;
isRowNot = M(:,1)~=6;

% Get the 2nd-column values for the 6 rows (and the not-6 rows)
%    These values will get separately shuffled
M2Six = M(isRowSix,2);
M2Not = M(isRowNot,2);

% The count of rows with 6's (and not)
numberRowsSix = sum(isRowSix);
numberRowsNot = sum(isRowNot);

% For each additional column ...
for nc = 1:NCOL

      % Create a random permutation of the values from the rows with 6's (and those not)
      newOrderSix = randperm(numberRowsSix);
      newOrderNot = randperm(numberRowsNot);

      newColSix = M2Six(newOrderSix);
      newColNot = M2Not(newOrderNot);

      % Insert those pemutations into the new array
      M_new(isRowSix,nc+2) = newColSix;
      M_new(isRowNot,nc+2) = newColNot;

end

[Georgios Tertikas]

thank you very much