熊猫：添加像= A1 / SUMIF（B：B，B1，A：A）这样的excel SUMIF列

Question

I have a pandas DataFrame like:

    pet  treats lbs
0   cat    2    5.0
1   dog    1    9.9
2  snek    3    1.1
3   cat    6    4.5
4   dog    1    9.4

I would like to add a fourth column that takes each treat as a percentage of the total treats for pets of that kind. So, the treat value in row 0, divided by the sum of all treats for pets matching "cat" (and so on for each row).

In Excel, I think I would do something like this:

   A    B    C    D
1  cat  2   5.0   =B1/SUMIF(A:A,A1,B:B)
2  dog  1   9.9   =B2/SUMIF(A:A,A2,B:B)
3  snek 3   1.1   =B3/SUMIF(A:A,A3,B:B)
4  cat  6   4.5   =B4/SUMIF(A:A,A4,B:B)
5  dog  1   9.4   =B5/SUMIF(A:A,A5,B:B)

Anyone have an idea how I could add this "treat_percent" column using pandas?

    pet  treats  lbs  treat_percent
0   cat    2    5.0   33.33
1   dog    1    9.9   50.00
2  snek    3    1.1   100.00
3   cat    6    4.5   66.67
4   dog    1    9.4   50.00

So far, I have tried:

df['treat_percent'] = df['pet'] / df.groupby('pet')['treats'].sum()

and

df['treat_percent'] = df['pet'] / df.loc[df['pet'] == df['pet'], 'treats'].sum()

Answer 1

You can using transform

df['treat_rate']=df.treats/df.groupby('pet').treats.transform('sum')
df
Out[153]: 
    pet  treats  lbs  treat_rate
0   cat       2  5.0        0.25
1   dog       1  9.9        0.50
2  snek       3  1.1        1.00
3   cat       6  4.5        0.75
4   dog       1  9.4        0.50

Answer 2

you might try:

df['treat_rate'] = df.groupby('pet').treats.apply(lambda x: x/sum(x)) 

or better:

df['treat_rate'] = df.groupby('pet').treats.transform(lambda x: x/sum(x))