Python3.x Pandas groupby基于索引的连续列值相等

您可以构建两个掩码，一个用于标识角色更改，另一个用于标识索引高于阈值(diff)的差异.然后在运行cumsum:

m1 = df.Role!= df.Role.shift()
m2 = df['Index'].diff().gt(5)

out = df.groupby((m1|m2).cumsum()).agg(
            Role = ('Role', 'first'),
            Name = ('Name', ' '.join), 
            Grade = ('Grade', 'mean')
        ).reset_index(drop=True)

输出:

          Role               Name     Grade
0     Provider  Alex William Juan  6.666667
1     Provider              Pedro  4.500000
2       Client             George  8.000000
3     Provider               Mark  9.400000
4       Client              James  8.100000
5  Transporter            Anthony  9.500000
6  Transporter              Jason  7.000000

中间体:

   Index         Role     Name  Grade     m1     m2  m1|m2  cumsum
0      1     Provider     Alex    7.0   True  False   True       1
1      2     Provider  William    7.5  False  False  False       1
2      7     Provider     Juan    5.5  False  False  False       1
3     15     Provider    Pedro    4.5  False   True   True       2
4     25       Client   George    8.0   True   True   True       3
5     26     Provider     Mark    9.4   True  False   True       4
6     37       Client    James    8.1   True   True   True       5
7     39  Transporter  Anthony    9.5   True  False   True       6
8     50  Transporter    Jason    7.0  False   True   True       7