Question

I’m trying to make some code run faster by using pypy. However, whenever I run the code using pypy, I get an error from pypy‘s numpy equivalent (numpypy).

In my usual code I use…

numpy.average(array, axis=0)

in order to average an array along a particular axis.

When running the same code using pypy and numpypy, I get the following error:

TypeError: average() got an unexpected keyword argument 'axis'

I could use a for loop to go through the array and average individual elements but this would take a long time and (I would imagine) not provide the speed that I would like.

Is there an alternative to numpy.average() that can average along an axis whilst using numpypy?

Answer 1

If you don’t need to use the weights parameter, you can use mean instead (1.9.1-dev0 — not sure when it was introduced):

>>>> import numpypy as np
>>>> a = np.arange(2*3).reshape(2,3)
>>>> a
array([[0, 1, 2],
       [3, 4, 5]])
>>>> np.mean(a, axis=0)
array([ 1.5,  2.5,  3.5])
>>>> np.mean(a, axis=1)
array([ 1.,  4.])
>>>> a.mean(axis=0)
array([ 1.5,  2.5,  3.5])
>>>> a.mean(axis=1)
array([ 1.,  4.])