Convert python list with None values to numpy array with nan values

Question

I am trying to convert a list that contains numeric values and None values to numpy.array, such that None is replaces with numpy.nan.

For example:

my_list = [3,5,6,None,6,None]
# My desired result: 
my_array = numpy.array([3,5,6,np.nan,6,np.nan]) 

Naive approach fails:

>>> my_list
[3, 5, 6, None, 6, None]
>>> np.array(my_list)
array([3, 5, 6, None, 6, None], dtype=object) # very limited 
>>> _ * 2
Traceback (most recent call last):
 File "<stdin>", line 1, in <module>
TypeError: unsupported operand type(s) for *: 'NoneType' and 'int'
>>> my_array # normal array can handle these operations
array([ 3., 5., 6., nan, 6., nan])
>>> my_array * 2
array([ 6., 10., 12., nan, 12., nan])

What is the best way to solve this problem?

Answer 1

You simply have to explicitly declare the data type:

>>> my_list = [3, 5, 6, None, 6, None]
>>> np.array(my_list, dtype=float)
array([ 3., 5., 6., nan, 6., nan])

• Akavall: I had a feeling that I was missing something simple...Thank You
• Andrzej Pronobis: The problem with this approach is that it only works for float for which nan is defined and not for dtype int.
• Babak: Justy a heads-up that np.float is deprecated. use dtype=float instead
• matanox: Interestingly this turns them to plain python float nan, not np.nan. I wonder why as it may seem a little irregular, even though both types of nan qualify the same under np.isnan() . This is so even when using dtype=np.float64 which is explicitly a numpy data type. This conversion is not even explicitly documented for np.array()
• Gideon Kogan: DeprecationWarning: np.float is a deprecated alias for the builtin float

Answer 2

What about

my_array = np.array(map(lambda x: numpy.nan if x==None else x, my_list))

• alexey: I'd say x is None as == might be more quirky
• Esteban: What worked for me was my_array = np.array(list(map(lambda x: np.nan if x is None else x, my_list)))