Different behaviour when assigning None using loc[] and []

[socheon]

import pandas as pd
import numpy as np 

ser = pd.Series(['a', 'b', np.nan])
print ser

ser.loc[ser.isnull()] = None
print ser

ser[ser.isnull()] = None
print ser 

Output

0      a
1      b
2    NaN
dtype: object
0       a
1       b
2    None
dtype: object
0      a
1      b
2    NaN
dtype: object

I am not sure whether this behaviour is intentional. Using .loc[], None was correctly assigned but when using [], NaN was assigned instead. Thanks

I am using version 0.14.0-169-g7684b6e

[jreback]

FYI, you should create the series new each time before printing it.

I think the .loc is correct, but the [] might be historically motivated. marking as a bug.

[jreback]

was closed by #7941 (as well)