Replace a certain value with np.nan or NA or nan in Numpy in python

Question

I've got a numpy array that contains data, but I want to replace some certain values with NA. I don't want to deal with masked data so I don't mask it. Is this possible?

My data is a numpy array with data type uint16:

array([[[1, 1, 1, ..., 1, 1, 1],
        [1, 1, 1, ..., 1, 1, 1],
        [1, 1, 1, ..., 1, 1, 1],
        ...,
        [1, 1, 1, ..., 1, 1, 1],
        [1, 1, 1, ..., 1, 1, 1],
        [1, 1, 1, ..., 1, 1, 1]]], dtype=uint16)

I want to replace those "1"s with NA values, I have used the code below, but I'm getting this error:

aa [aa == 1] = np.nan
ValueError: cannot convert float NaN to integer

Please include a small sample input and the expected output. What do you me mean by "replace some certain values with NA"? — Dani Mesejo, Feb 25 '19 at 12:33
The edited post still doesn't contain an example output of what you expect. — Alfe, Feb 25 '19 at 12:44

score 2 · Answer 1 · answered Feb 25 '19 at 12:45

2

type(np.nan) returns <class 'float'>.

You need your array to be of float type.

See: Numpy integer nan

answered Feb 25 '19 at 12:45

Spoutnovitch

119
10

score 1 · Answer 2 · answered Feb 25 '19 at 12:47

1

You try to put NaN values into an array of type uint16 which is a type that doesn't know a NaN value. Try the same with float32 instead and it will work.

answered Feb 25 '19 at 12:47

Alfe

56,346
20
107
159

Replace a certain value with np.nan or NA or nan in Numpy in python

2 Answers2