zero values to nan values in an array of Python

Question

I have an array with some zero values that i wish to convert in nan values. When i apply the code
all values become nan

myarray
array([[ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       ..., 
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.]])

myarray.shape
(64L, 52L)
myarray.max()
4563.666015625
myarray.min()
0.0

i wish to convert zero values in nan. I use an example from stackoverflow

a = np.arange(3.0)
a
array([ 0.,  1.,  2.])
a[a==0] = np.nan
a
array([ nan,   1.,   2.])

when i apply the example to my array all values became nan

myarray[myarray == 0.] = nan
myarray
array([[ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       ..., 
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan]])

myarray.max()
nan
myarray.min()
nan

DSM · Accepted Answer · 2014-12-30 06:47:59Z

5

It's not that all the values become nan, it's that (1) you're only looking at the parts that do, and (2) min and max don't work well with nans.

For example, if we make an array resembling yours:

>>> myarray = np.zeros((64, 52))
>>> myarray[3:-3,3:-3] = np.random.uniform(0, 5000, (64-6,52-6))
>>> myarray
array([[ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       ..., 
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.],
       [ 0.,  0.,  0., ...,  0.,  0.,  0.]])
>>> myarray[myarray==0] = np.nan
>>> myarray
array([[ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       ..., 
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan],
       [ nan,  nan,  nan, ...,  nan,  nan,  nan]])

It may look like it's all nan, but it's not:

>>> myarray[2:5, 2:5]
array([[           nan,            nan,            nan],
       [           nan,  1500.05326562,  4583.70521213],
       [           nan,  4896.62420284,   892.83210033]])

You can also use nanmin and nanmax, which ignore nans:

>>> myarray.min()
nan
>>> myarray.max()
nan
>>> np.nanmin(myarray)
0.60474162939361253
>>> np.nanmax(myarray)
4996.8967777356092

answered Dec 30, 2014 at 6:47

DSM

355k67 gold badges606 silver badges504 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Gianni Spear Over a year ago

Thanks @DSM. is it possible with np.nan compute also the standard deviation and the mode?

DSM Over a year ago

There's nanstd; I don't think there's nanmode, though. You could always compute the mode of the non-nan values manually after removing them, though (myarray[~np.isnan(myarray)]).

Collectives™ on Stack Overflow

zero values to nan values in an array of Python

1 Answer 1

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related