I am puzzled by the behaviour of the numpy.where command with None values -- it does not seem to return the correct indices of
None and "not None" values:
>>> import numpy as np
>>> a = np.array([None, 1])
# The following works as expected
>>> a[0]==None
True
>>> a[0]!=None
False
>>> np.where(a)
(array([1]),)
# What's going on here? None of the output seems correct here
>>> np.where(a != None)
(array([0]),)
>>> np.where(a == None)
(array([], dtype=int32),)
Is this intentional or a bug? How can I use numpy.where to find indices equal to None?
I am aware of this question, that talks about using masked arrays to use numpy with arrays with NoneType values, but my question is specifically about numpy.where.
Python: 2.7 Numpy: 1.6.1
