How to separate a column into rows based on the element of a different column?

Input: a = {1 2; 2 3; 3 5; -1 6; 5 7; 6 8; 7 9; 8 10; -1 11};

I have a 9x2 cell array matrix. I need my output this way: if the element in column 1 == -1, it will create a new row from the 1st element of column 2 to the element of column 2 until -1 in column 1 (inclusive). Then if -1 appears again in the first column, it will create a new row starting after the previous endpoint of column 2 to the element in column 2 until -1 appears in column 1 (inclusive).

Expected output: b = {2 3 5 6; 7 8 9 10 11}

I was trying b = a(cat(1,a{:,1}) >= -1 ,2:2).' but couldn't make separate rows. How can I do it?

[Stephen23]

b = {2 3 5 6; 7 8 9 10 11}

This cell array is not possible because each row has a different number of elements.

Why are you storing numeric scalars in a cell array? This just makes processing the data pointlessly complex.

[Md Shahidullah Kawsar]

Thanks for your reply. Besides cell array, is it possible to deal with a different number of elements in each row?

Answer 1

[Stephen23]

This is a lot easier if you simply store the scalar numeric values in one numeric array (use cell2mat if required):

>> A = [1,2;2,3;3,5;-1,6;5,7;6,8;7,9;8,10;-1,11]
A =
    1    2
    2    3
    3    5
   -1    6
    5    7
    6    8
    7    9
    8   10
   -1   11

Method one: diff and mat2cell:

>> X = diff([0;find(A(:,1)==-1)]);
>> C = mat2cell(A(:,2),X,1);
>> C{:}
ans =
    2
    3
    5
    6
ans =
    7
    8
    9
   10
   11

Method two: cumsum and accumarray:

>> X = cumsum([true;A(1:end-1,1)==-1]);
>> C = accumarray(X,A(:,2),[],@(v){v});
>> C{:}
ans =
    2
    3
    5
    6
ans =
    7
    8
    9
   10
   11

[Md Shahidullah Kawsar]

Thanks a lot