Difference between number of unique values in collaborative filtering movie titles and len(data.classes.value)

I noticed there was a difference between those two figures, with there being fewer titles in data.classes.values, I would guess that perhaps movies that occur too rarely in the dataset have been replaced with #na#. If that’s the case, I would really love to know if that behaviour is something I can change, or is there a threshold on the number of minimum occurrences a title has to have before being removed?