pandas groupby：如何计算总数的百分比？

Question

How can I calculate a column showing the % of total in a groupby?

One way to do it is to calculate it manually after the gorupby, as in the last line of this example:

import numpy as np
import pandas as pd
df= pd.DataFrame(np.random.randint(5,8,(10,4)), columns=['a','b','c','d'])
g = df.groupby('a').agg({'b':['sum','mean'], 'c':['sum'], 'd':['sum']})
g.columns = g.columns.map('_'.join)
g['b %']=g['b_sum']/g['b_sum'].sum()

However, in my real data I have many more columns, and I'd need the % right after the sum, so with this approach I'd have to manually change the order of the columns.

Is there a more direct way of doing it so that the % is the column right after the sum? Note that I need the agg(), or something equivalent, because in all my groupbys I apply different aggregate functions to different columns (eg sum and avg of x, but only the min of y, etc.).

Thanks!

Answer 1

I think you need lambda function in agg and then replace column names to % :

np.random.seed(78)
df= pd.DataFrame(np.random.randint(5,8,(10,4)), columns=['a','b','c','d'])

g =(df.groupby('a')
      .agg({'b':['sum',lambda x: x.sum()/ df['b'].sum(),'mean'],
            'c':['sum'], 
            'd':['sum']}))
g.columns = g.columns.map('_'.join).str.replace('<lambda>','%')

print (g)

   d_sum  c_sum  b_sum       b_%  b_mean
a                                       
5     25     24     24  0.387097       6
6     11     11     14  0.225806       7
7     22     23     24  0.387097       6