Improve speed of for loop

I have a for loop as follows:

    for z = 1:r      %run through data from start to end 
    x = row14 >= (z-1)*tp & row14 <= z*tp;  % run through in steps the size of your time bins
    s(z) = nnz(x); 
    end  
 
    
   

However, I'm working with really large matrices, r = 56043, and row14 is 1x19609285 double. As a result this particular for loop takes absolutely ages. Is there any way that I can improve the performance to speed up the process a bit?

Answer 1

[Walter Roberson]

idx = floor(row14 ./ tp) + 1;
counts = accumarray(idx(:), 1);
counts(end+1:r) = 0;

Or

counts = histcounts(row14, (0:r)*tp);

or

counts = histcounts(row14, 'binwidth', tp);

Caution: the binwidth will not be respected if the data implies more than 2^16 bins.

[Anna van der Stap]

Thank you, brilliant!