计数为零时的熊猫分组以及如何在结果中包含零值

Question

Please consider any solution without usage of pivot_table() or unstack()

For the following dataframe:

import pandas as pd

df = pd.DataFrame({
    'name': ['Mason', 'Ali', 'Si', 'Pedram'],
    'continent': ['Europe', 'Asia', 'Asia', 'Asia'],
    'blood_type': ['AB', 'O+', 'AB', 'AB']
})

Which is:

    name     continent  blood_type
0   Mason    Europe     AB
1   Ali      Asia       O+
2   Si       Asia       AB
3   Pedram   Asia       AB

The following groupby count:

df.groupby(['continent', 'blood_type']).count()

Will produce:

                           name
continent   blood_type  
Asia        AB             2
            O+             1
Europe      AB             1

Instead, how can I include zero value count like the table below? (by not using pivot_table or unstack )

                           name
continent   blood_type  
Asia        AB             2
            O+             1
Europe      AB             1
            O+             0

Answer 1

As of pandas 0.25 (or 0.24.2, not sure now), if you groupby a Category ( pd.Categorical ), it will show all values in the final count.

df.groupby([pd.Categorical(df.continent), 'blood_type']).count().fillna(0)

                   name  continent
       blood_type                 
Asia   AB           2.0        2.0
       O+           1.0        1.0
Europe AB           1.0        1.0
       O+           0.0        0.0