Python Group by multiple columns and keep other columns

55 views Asked by At

I have a table that looks like below:

City_code     City_name     Site_code      Site_capacity
AAA100      City_A          Site001         300
AAA100      City_A          Site002         600
AAA100      City_A          Site003         500
AAA200      City_B          Site004         350
AAA200      City_B          Site005         250
AAA300      City_C          Site006         800
AAA300      City_C          Site007         150
AAA300      City_C          Site008         450
AAA400      City_D          Site009         300
AAA400      City_D          Site0010        400

I want to select the site with with the Highest value for Site_capacity for each city

I have tried the following code:

df.groupby(['City_code', 'City_name'])['Site_capacity'].max()

This this the output it generates:

City_code     City_name     
AAA100      City_A          600
AAA200      City_B          350
AAA300      City_C          800
AAA400      City_D          400

How do i create some output that looks like this?

City_code     City_name     Site_code      Site_capacity
AAA100      City_A          Site002         600
AAA200      City_B          Site004         350
AAA300      City_C          Site006         800
AAA400      City_D          Site0010        400
3

There are 3 answers

0
BENY On BEST ANSWER

We can do sort_values + drop_duplicates

s = df.sort_values('Site_capacity').drop_duplicates(['City_code', 'City_name'],keep='last')
Out[334]: 
  City_code City_name Site_code  Site_capacity
3    AAA200    City_B   Site004            350
9    AAA400    City_D  Site0010            400
1    AAA100    City_A   Site002            600
5    AAA300    City_C   Site006            800
0
Umar.H On

try idxmax() and .loc

print(df.loc[df.groupby(['City_code', 'City_name'])['Site_capacity'].idxmax()])

  City_code City_name Site_code  Site_capacity
1    AAA100    City_A   Site002            600
3    AAA200    City_B   Site004            350
5    AAA300    City_C   Site006            800
9    AAA400    City_D  Site0010            400
2
Malan On

Try this:

df.sort_values(by=['City_name','Site_capacity'], inplace=True,ascending = (True, False)) 
df = df.drop_duplicates('City_name', keep='first')
print(df)

Result:

  City_code City_name Site_code  Site_capacity
  AAA100    City_A    Site002    600
  AAA200    City_B    Site004    350
  AAA300    City_C    Site006    800
  AAA400    City_D    Site0010   400

Or if you want to keep the lowest value.

df = df.drop_duplicates('City_name', keep='last')