When is it not possible to apply vectorization to pandas processing?

Question

I understand this might be a paricularly ambiguous question, but I have been experimenting with optimizing code using pandas vectorization because I work with Bigdata. I also un