[Python] How do I replace "NaN" in a pandas dataframe with a dictionary?

Sample dataframe:

df = pd.DataFrame([[1, {'t': 1}], [2, {'t': 2}], [3, np.nan], [4, np.nan]], columns=['A', 'B'])


A B
0 1 {'t': 1}
1 2 {'t': 2}
2 3 NaN
3 4 NaN


What I want:

A B
0 1 {'t': 1}
1 2 {'t': 2}
2 3 {'t': 0}
3 4 {'t': 0}


For some reason, the below does not work -

df.fillna({'B': {'t': 0}}, inplace=True)


Is there a restriction on what values can be used for fillna?

I have this solution which works,

df['B'] = df['B'].mask(df['B'].isna(), {'t': 0})


but I want to know if there is a better way to do this? For example, this fails if 'B' does not exist in the dataframe, whereas the fillna doesn't fail in similar case

Continue reading...