从带有条件的 dataframe 中选择列

Question

I was learning to use pandas library today and I came across this error that I couldn't fully understand. This is the dataframe I was using.

      name   kda  combat  econ
0   Austin  1.45   270.0    67
8   Austin  1.70   300.0    90
4   Justin  1.36   230.0    50
11  Justin  1.50   270.0    60
1    Kevin  1.40   230.0    55
6    Kevin  1.00   100.0   120
3     Matt  1.00   180.0    65
9     Matt  1.40   280.0    70
2     Nick  2.10   360.0    87
7     Nick  2.50   340.0    88
5     Will  1.20   185.0    45
10    Will  1.60   260.0    75

I was trying to get name and kda columns who have average combat scores greater than 250 which I tried to achieve by doing

temp = df.groupby('name').mean()
temp = temp[temp['combat'] > 250]
print(temp['name', 'kda'])

but it returned this key error instead

KeyError: "['name'] not in index"

Could someone explain why I can't grab columns from these temporary dataframes? Or did I do something wrong in my code? Luckily my friend helped me out and I could do it by

temp = df.loc[df['combat'] > 250, ['name','kda']]
print(temp.groupby('name').mean())

This did the trick to give

          kda
name         
Austin  1.575
Justin  1.500
Matt    1.400
Nick    2.300
Will    1.600

Thank you in advance

Answer 1

When you do a groupby("col_name") , the default behaviour is for pandas to set the col_name as the index

In your case, you can name will be set as the dataframe index

You can use

temp = df.groupby('name').mean()
temp = temp[temp['combat'] > 250]
print(temp['kda'])

to get your desired result (it will return a Series)

Another option would be to use as_index=False with groupby

groupby('col_name', as_index=False)

This will return a dataframe with 'name' as a column and your first solution will work

Have a look at the intermediate steps and you'll see what's going on

Answer 2

Alternative answer.

.reset_index() can be used after .groupby() as in code below. Also while printing you may need to add [[]] instead of [] if more than two columns need to be printed.

# Import libraries
import pandas as pd

# Create DataFrame
df = pd.DataFrame({
    'name': ['Austin','Austin','Justin','Justin','Kevin','Kevin',
            'Matt','Matt','Nick','Nick','Will','Will'],
    'kda': [1.45,1.70,1.36,1.50,1.40,1.40,1.0,1.30,2.10,2.50,1.20,1.60],
    'combat':[270.0,300.0,230.0,270.0,230.0,100.0,180,280,360,340,185,260],
    'econ':[67,90,50,60,55,120,65,70,87,88,45,75]
})

# Groupby (copy pasted code from question and modified)
temp = df.groupby('name').mean().reset_index()
temp = temp[temp['combat'] > 250]
print(temp[['name', 'kda']])

Output

     name    kda
0  Austin  1.575
4    Nick  2.300