Search code examples
pythonpandasmulti-index

Create index from multiindex pandas dataframe

I have a data frame

   0        1      
   A   B    A    B   
1  4   6    5    6  
2  4   8    15   3
3  8   10   10   2

I wanted to create an index so all of the columns are divided by the total of the A column for their persprective level 0 i.e. 0 or 1

so

0=16
1=30

so the output is

   0              1      
   A   B          A       B   
1  4/16   6/16    5/30    6/30  
2  4/16   8/16    15/30   3/30
3  8 /16  10/16   10/30   2/30

so far I have

df_in.divide(df_in.iloc[:, df_in.columns.get_level_values(1)=='A'].sum(axis=0),level=0,axis=1)

But its not working

Solution

  • Use DataFrame.xs for select level A with sum and pass to DataFrame.div:

    df = df.div(df.xs('A', axis=1, level=1).sum(), axis=1, level=0)
print (df)
      0                1          
      A      B         A         B
1  0.25  0.375  0.166667  0.200000
2  0.25  0.500  0.500000  0.100000
3  0.50  0.625  0.333333  0.066667