Why is numpy.random.choice so slow?

The take-home message is that vectorizing or parallel computing is very well implemented in numpy. Do take advantage of this!

• Instead of

import numpy as np
for i in range(10000):
    print( np.random.choice(a=np.arrange(10)) )

• do something like

import numpy as np
for i in np.random.choice(a=np.arrange(10), size=10000, replace=True):
    print(i)

• This would save you more than 10000x time