Not sure what you're arguing. If you use numpy and pandas properly, most of your computations happen in optimized C routines.

So you will write Python, but get very good resource (both time and space) efficiency.