MATLAB: Delete row not following pattern

delete

Hi,
I have….
a = [ 2 5 1 0.504; 
      3 6 2 0.507; 
      3 4 1 0.589; 
      9 4 2 0.503; 
      8 3 1 0.592; 
      3 2 2 0.203; 
      9 5 2 0.341; 
      4 8 1 0.592]
Notice how the 2nd to the last column, column #3, follows a pattern of 1, 2, 1, 2..and so on. The 7th row however has a 2 in the 2nd last column just like the 6th row before…thus does not follow the pattern. I want to delete one of these two rows based on the value in the last column. I want to delete which ever row has a value in the last column further away from the number 0.500.
I would want to delete any row where the 2nd last column does not follow the pattern 1,2,1,2 and when deciding between which row out of the two rows to delete…base it on whichever is further from the number 0.500. Thanks!

Best Answer

Assuming the pattern in the next-to-last column must be 1-2-1-2-...etc exactly
[m,n] = size(a);
n1 = n - 1;
expected = 1; % initialize expected value in 1st row

x = false(m,1); % initialize the deletion flag array
keep = 0; % initialize row to keep when marking for deletion
v = 0.5; % discriminator value for deciding which line to delete
for k=1:m
    if( a(k,n1) ~= expected ) % if not as expected, mark one line for deletion
        if( keep == 0 || abs(a(k,n)-v) > abs(a(keep,n)-v) )
            x(k) = true; % mark current line for deletion
        else
            x(keep) = true; % mark previous line for deletion
            keep = k;
        end
    else
        expected = 3 - expected; % if as expected, update expected
        keep = k;
    end
end
a(x,:) = []; % delete the unexpected pattern rows
If the first line can start with a 2 rather than a 1, change this line:
expected = a(1,n1); % initialize expected value in 1st row
Related Question