How to control the output of groupby in pandas to show the whole row not a single column

Question

this is a sample my dataframe(seasonal_name):

    season  winner  count   team_api_id team_long_name
1   2008/2009   10260   28  10260.0 Manchester United
2   2008/2009   8634    27  8634.0  FC Barcelona
3   2008/2009   8548    26  8548.0  Rangers
4   2008/2009   8650    25  8650.0  Liverpool
5   2008/2009   8636    25  8636.0  Inter

• I want to show highest count rows in each season with team_long_name, and for that i try this code:

 seasonal_name.groupby("season")["count"].max()

and its output is like this:

season
2008/2009    28
2009/2010    31
2010/2011    30
2011/2012    32
2012/2013    32

I want the output to contain the corresponding team_long_name and the other columns as well not just the count column.

Answer 1

If you're okay with dropping duplicates (eg if 2 different teams have the same highest score for a season, only display 1 of the teams), you can do the following:

new_df = df.loc[df.groupby(["Season"])["Count"].idxmax()] 

This first groups by Season and gets the indices of the rows having the maximum count for each season. We then locate those rows in df using loc .

The only issue is that idxmax only returns 1 row even if the maximum Count value for a season exists in multiple rows. Still, I hope this is a good starting point for you.

This is the code in action. The following is the example dataframe I used:

  Group  Count  Season
0     G      2  spring
1     G      3  summer
2     H      1  spring
3     H      4  summer

And this is what my line of code gives for new_df :

  Group  Count  Season
0     G      2  spring
3     H      4  summer

If you'd like to change the index of new_df , you can just do:

new_df = new_df.reset_index(drop=True)

Hope that helps.