Can I return 'nan' value for bad data?

I am writing an algorithm for a big database. I am running the algorithm in a for loop over the database trials. My algorithm works for >99% of the trials in the database. But it stops for a few due to some bad data entry. So I have to keep eye to check if my loop has stopped running. Can I force the algorithm to output 'nan' value for the cases where there are bad data? If not how should I avoid the situstion?

Here is the error message given by Matlab.

% Unable to perform assignment because the left and right sides have a different number of elements.

Answer 1

[Star Strider]

One option would be to create the left side as a cell array (so the dimensions do not have to match) and then deal with any inconsistencies when that part of the code (apparently the loop) finishes.