Pandas Pivot table: How to sort top n nows in the pivot table by category and generate new dataframe? - pandas

Question

df = pd.DataFrame({'country': ['AUD','CAD', 'IND','JPY', 'AUD', 'CHY', 'IND', 'KRL', 'SRI', 'KRW', 'CAD'],
'area': ['N','S','N','E','W','S','NE','N','S','SE','N'], 'gdp': [349,65,60,88,75,100,200,250,150,210,160], 'income': [7000,2300,5000,1000,550,1000,2060,2750,1450,2610,1650], 'expense': [500,300,700,600,500,900,206,275,1405,210,150]})
df = df.pivot_table(index=['country','area'],values=['gdp'],aggfunc='sum').sort_values(by = ['gdp'], ascending = False, axis = 0).head(5)
By applying above method, I am unable to see top 5 'country' based on 'gdp'. My expected output should be like below in dataframe. Please suggest. However, I tried to populate the expected output through MS excel to get a feel.
new_df = country gdp expense income
AUD 424 1000 7550
N 349 500 7000
W 75 500 550
IND 260 906 7060
N 60 700 5000
NE 200 206 2060
KRL 250 275 2750
N 250 275 2750
CAD 225 450 3950
N 160 150 1650
S 65 300 2300
KRW 210 210 2610
SE 210 210 2610
new_df = country gdp expense income area
AUD 424 1000 7550 N, W
IND 260 906 7060 N, NE
KRL 250 275 2750 N
CAD 225 450 3950 N,S
KRW 210 210 2610 SE

Answer 1

Use -
df.groupby('country', as_index=False).agg({'gdp': 'sum', 'area': ','.join}).sort_values(by='gdp', ascending=False).head(5)
Output
country gdp area
0 AUD 349 N,W
5 KRL 250 N
6 KRW 210 SE
3 IND 200 N,NE
1 CAD 160 S,N

Answer 2

you don't need to use pivot for this, it can be done this way
df.groupby('country').agg({'gdp':'sum',
'area':','.join}).sort_values('gdp',ascending = False).head(5)
Output:
country gdp area
AUD 424 N,W
IND 260 N,NE
KRL 250 N
CAD 225 S,N
KRW 210 SE