Comment by pacofvf

my point is that the benchmark is deceiving, again if you are doing data analytics or ML, then you already are using numpy/pandas/scipy, so thats not a valid argument.