Does the count() in pyspark depend on the #columns in dataset?

Question

I am making a function using pyspark that is using count at a place.

I ran 2 versions of my function, one with onehotencoding done before the count, and other post oh