Sorting unique values according to another column in pandas

Question

I'm trying to sort unique values in pandas dataframe with group by;

df = pd.DataFrame({
...     'gr1': ['A', 'A', 'A','A', 'B', 'B', 'B','B'],
        'gr1_sum' : [100,100 ,100,100, 200,200,200,200],
        'rank_gr1': [2, 2, 2, 2, 1, 1, 1, 1],

...     'gr2': ['a1', 'a1', 'a2','a2',  'b1', 'b1', 'b2','b2'],
        'gr2_sum' : [30,30 ,40,40, 20,20,10,10]})

#df.sort_values(by=['col2'],inplace = True)    
rank_gr1_sort = pd.unique(df['rank_gr1'].values)
rank_gr2_sort = df.sort_values(['rank_gr1']).groupby(['gr1','gr2'])['gr2_sum'].unique()

rank_gr1_sort
array([2, 1], dtype=int64)

rank_gr2_sort
gr1  gr2
A    a1     [30]
     a2     [40]
B    b1     [20]
     b2     [10]
Name: gr2_sum, dtype: object

What I need to have is this;

   gr1  gr2
    B    b1     [20]
         b2     [10]
    A    a1     [30]
         a2     [40]
    Name: gr2_sum, dtype: object

How I do achive this output ?

Thx!

pandas groupby sort within groups

Pandas Number of Unique Values and sort by the number of unique

Answer 1

Pass sort=False under the groupby.

From docs:

sort : bool, default True Sort group keys. Get better performance by turning this off. Note this does not influence the order of observations within each group. Groupby preserves the order of rows within each group.

rank_gr2_sort = df.sort_values(['rank_gr1']).groupby(
                       ['gr1','gr2'],sort=False)['gr2_sum'].unique()

---

gr1  gr2
B    b1     [20]
     b2     [10]
A    a1     [30]
     a2     [40]
Name: gr2_sum, dtype: object