Pandas Possibly unexpected behaviour when using apply and custom funcs

total=pd.DataFrame({"all_eternal":[1,2,1,1,0,1],"country_name":["United States","Germany","Germany","Germany","France","United States"],"amount":[20,40,30,10,5,6]})

When I apply

totalbycountrynew=total.groupby(["country_name"],as_index=False).apply(
        lambda g:
            pd.Series({
                    "LTVav":g.amount.mean(),
                    "LTVav_without_trials":g.amount[g.all_eternal!=1].mean()
                    })
        )

I would like to see:

country_name    LTVav   LTVav_without_trials
France            5.0                         5.0
Germany          26.7                       40.0
United States   13.0                           nan

Instead I see:

LTVav   LTVav_without_trials
 5.0                         5.0
 26.7                       40.0
 13.0                          nan (with index 0,1,2)

Without as_index=False works as_expected - countries' names are there, but they are in index While there is a workaround, I would say it is an unexpected behaviour (at least non-intuitive one)

(sorry for styling, I don't know Markdown)

Comment From: jorisvandenbossche

(sorry for styling, I don't know Markdown)

https://guides.github.com/features/mastering-markdown/#GitHub-flavored-markdown (first example show how to make code blocks)

Comment From: jreback

you don't need ``as_index=False)

In [25]: total.groupby(["country_name"]).apply(
    ...:         lambda g:
    ...:             pd.Series({
    ...:                     "LTVav":g.amount.mean(),
    ...:                     "LTVav_without_trials":g.amount[g.all_eternal!=1].mean()
    ...:                     })
    ...:         )
    ...:        
Out[25]: 
                   LTVav  LTVav_without_trials
country_name                                  
France          5.000000                   5.0
Germany        26.666667                  40.0
United States  13.000000                   NaN

Comment From: cryptotvync

@jreback did you real all my comment?) Countries' names are in index then, not in column, as usual

Comment From: TomAugspurger

@cryptotvync can you take the time to properly format your desired output then? @jreback's suggestion looks correct.