pandas groupby and rank within groups that start with 1 for each group

问题

I have a dataframe:

import pandas as pd

df = pd.DataFrame([[1, 'a'],
                    [1, 'a'],
                    [1, 'b'],
                    [1, 'a'],
                    [2, 'a'],
                    [2, 'b'],
                    [2, 'a'],
                    [2, 'b'],
                    [3, 'b'],
                    [3, 'a'],
                    [3, 'b'],

                   ], columns=['session', 'issue'])
df

I would like to rank issues within sessions. I tried with:

df.groupby(['session', 'issue']).size().rank(ascending=False, method='dense')

session  issue
1        a        1.0
         b        3.0
2        a        2.0
         b        2.0
3        a        3.0
         b        2.0
dtype: float64

What I need is result like this one:

1. for group session=1, there are three a issues and one b issue, so for group 1, ranks are a = 1 and b = 2
2. for group session=2, both ranks are equal so their rank should be the same = 1
3. for group session=3, there are to b issues and one a issue, so ranks should be b=1 and a=2

Anyway, why for each group ranks don't start from 1, 2, 3...?

回答1:

Use DataFrameGroupBy.rank by first level of MultiIndex (session):

s = (df.groupby(['session', 'issue'])
        .size()
        .groupby(level=0)
        .rank(ascending=False, method='dense'))
print (s)
session  issue
1        a        1.0
         b        2.0
2        a        1.0
         b        1.0
3        a        2.0
         b        1.0
dtype: float64

来源：https://stackoverflow.com/questions/54530503/pandas-groupby-and-rank-within-groups-that-start-with-1-for-each-group