How do you bar plot the nlargest of aggregated groupby data?

Question

I'm sure this has been answered somewhere, but I'm not great with pandas and need someone to break it down for me.

I have this function:

def process_data(data):
    data = data[data['Bucket Number'] == 25.0].groupby(['Activity Month', 'Agent Sign']).agg({'Total Ping Current Forecast': [np.sum]})
    return data

EDIT: Fixed Sample

That produces this output:

                          Total Ping Current Forecast
                                                  sum
Activity Month Agent Sign                            
202001         CRCDIF                        618485.0
               CRCTLD                      47746137.0
               DFW1DF                      16220228.0
               HDQ9LO                         19995.0
               HDQBDE                         10739.0
...                                               ...
202007         HDQRPT                             0.0
               HDQZED                             0.0
               MSCUHD                             0.0
               RDUSMD                         56732.0
               {Various}                  245079251.0

How would I make a bar plot where the X-axis is the Activity Month, the Y-axis is the sum, and for each tick is the two largest Agent Sign bars?

So for example the tick at 202001 would have two bars, one for Agent Sign CRCTLD and one for DFW1DF.

Answer 1

Let's try extracting the rows with groupby().cumcount (also possible with nlargest), then plot with sns.barplot():

processed = process_data(data)

sns.barplot(data=processed.sort_values('sum',ascending=False)
                     .assign(rank=lambda x: x.groupby(level=0).cumcount())
                     .reset_index()
                     .loc[lambda x: x['rank']<=1],
            x='Activity Month',y='sum', hue='rank'     
           )

Output: