Combine GroupBy Objects (not the output but the actual objects)-CodePudding

Admittedly, probably not an frequent case scenario. In essence, I am wondering if it is possible to take groups_1=df.groupby('something') and groups_2=df.groupby('something_else') and put them into groups = [groups_1,groups_2] so that I can iterate over one iterable groups object instead of iterating over the two objects separately. Here's an example:

# create data frame    
df = pd.DataFrame({'A':np.random.random(20), 'B':np.random.random(20), 'date':pd.date_range("2021-05-01",periods = 20, freq='2W')})

# groupby at two separate frequencies
groups_1 = df.loc[df.date<'2021-11-01',:].groupby(pd.Grouper(key='date', freq='M'))
groups_2 = df.loc[df.date>='2021-11-01',:].groupby(pd.Grouper(key='date', freq='2W'))

Given the above, I would love to do something like:

groups = some_concat_function([groups_1,groups_2])

for label, group in groups:
        # do something here

CodePudding user response：

I think you can use itertools.chain for that:

from itertools import chain

groups = chain(groups_1, groups_2)
...

from itertools import chain

groups = chain.from_iterable([groups_1, groups_2])
...