How to use groupby in pandas to calculate a percentage / proportion total based on a criteria in another column

问题

I'm trying to work out how to use the groupby function in pandas to work out the proportions of values per year with a given Yes/No criteria.

For example, I have a dataframe called names:

  Name  Number  Year   Sex Criteria
0  name1     789  1998  Male      N
1  name1     688  1999  Male      N
2  name1     639  2000  Male      N
3  name2     551  1998  Male      Y
4  name2     499  1999  Male      Y

I can use

namesgrouped = names.groupby(["Sex", "Year", "Criteria"]).sum()

to get:

                   Number
Sex    Year      Criteria
Male   1998 N        14507
            Y         2308
       1999 N        14119
            Y         2331

and so on. I would like the 'Number Criteria' column to show the % of the total for each gender and year - so instead of N = 14507 and Y = 2308 for 1998 above I'd have N = 86.27% and Y = 13.73%.

Can anyone advise how to do this?

回答1:

This question is a direct extension of the suggested duplicate. Borrowing from the accepted answer, this will work:

In [46]: namesgrouped.groupby(level=[0, 1]).apply(lambda g: g / g.sum())
Out[46]: 
                      Number
Sex  Year Criteria          
Male 1998 N         0.588806
          Y         0.411194
     1999 N         0.579612
          Y         0.420388
     2000 N         1.000000

---

Edit: a transform operation might be faster than apply:

namesgrouped / namesgrouped.groupby(level=[0, 1]).transform('sum')

来源：https://stackoverflow.com/questions/36987829/how-to-use-groupby-in-pandas-to-calculate-a-percentage-proportion-total-based