pandas: Grouping by two columns and then sorting it by the values of a third column

Question

I have the following line:

genre_df.groupby(['release_year', 'genres']).vote_average.mean()

This gives me the following:

release_year  genres         
1960          Action             6.950000
              Adventure          7.150000
              Comedy             7.900000
              Drama              7.600000
              Fantasy            7.300000
              History            6.900000
              Horror             8.000000
              Romance            7.600000
              Science Fiction    7.300000
              Thriller           7.650000
              Western            7.000000
1961          Action             7.000000
              Adventure          6.800000
              Animation          6.600000
              Comedy             7.000000
              Crime              6.600000
              Drama              7.000000
              Family             6.600000
              History            6.700000
              Music              6.600000
              Romance            7.400000
              War                7.000000
...

What I'd like to see is the df grouped by release year and genre, but sorted by the highest vote average first.

AKA:

 release_year  genres         
    1960          Horror             8.000000
                  Comedy             7.900000
                  Action             6.950000
                  Thriller           7.650000
                  Drama              7.600000
                  Romance            7.600000
                  Fantasy            7.300000
                  Science Fiction    7.300000
                  Adventure          7.150000
                  Western            7.000000
                  History            6.900000

How can this be achieved?

Answer 1

Solution for 0.23.0+ - first create one column DataFrame by to_frame and then sort_values :

df = df.to_frame().sort_values(['release_year','vote_average'], ascending=[True, False])
print (df)
                              vote_average
release_year genres                       
1960         Horror                   8.00
             Comedy                   7.90
             Thriller                 7.65
             Drama                    7.60
             Romance                  7.60
             Fantasy                  7.30
             Science Fiction          7.30
             Adventure                7.15
             Western                  7.00
             Action                   6.95
             History                  6.90
1961         Romance                  7.40
             Action                   7.00
             Comedy                   7.00
             Drama                    7.00
             War                      7.00
             Adventure                6.80
             History                  6.70
             Animation                6.60
             Crime                    6.60
             Family                   6.60
             Music                    6.60

For oldier versions of pandas is necessary reset_index and set_index :

df = (df.reset_index()
       .sort_values(['release_year','vote_average'], ascending=[True, False])
       .set_index(['release_year','genres']))

Answer 2

try this:

   genre_df = genre_df.reset_index()
   genre_df.sort_values(['vote_average'],ascending=False)