Calculation between columns per group in pandas-CodePudding

Hello I have a dataframe such as :

COL1 COL2 COL3 
A    30   400
A    32   400
A    70   400
B    32   700
B    10   700

And I would like for each COL1, to calculate the sum of COL2 / COL3.unique()

Example = 30 32 70/400 = 0.33

Here I should then get:

COL1 COL2 COL3 New_col
A    30   400  0.33
A    32   400  0.33
A    70   400  0.33
B    32   700  0.06
B    10   700  0.06

Does someone have an idea please ?

CodePudding user response：

In your case do groupby with transform

df['new'] = df.COL3.rdiv(df.groupby('COL1')['COL2'].transform('sum'))
df
Out[19]: 
  COL1  COL2  COL3   new
0    A    30   400  0.33
1    A    32   400  0.33
2    A    70   400  0.33
3    B    32   700  0.06
4    B    10   700  0.06