Python 如果包含特定值，则筛选Groupby

使用groupby.transform与设置操作:

out = df[df.groupby('NAME')['CLASS']
           .transform(lambda x: ((S:=set(x))>={'EQ', 'AI'}) and ('BO' not in S))]
# or
# out = df[df.groupby('NAME')['CLASS']
#            .transform(lambda x: ((S:=set(x))>={'EQ', 'AI'})
#                       and not S.intersection({'BO'}))]

groupby.agg和isin的变体:

g = df.groupby('NAME')['CLASS'].agg(set)
out = df[df['NAME'].isin(g.index[(g >= {'EQ', 'AI'}) & ~(g >= {'BO'})])]

输出:

   NAME CLASS
4     B    LI
5     B    EQ
6     B    AI
11    D    EQ
12    D    AI

Intermediates

为了完整性，

# transform approach
   NAME CLASS  g >= {'EQ', 'AI'}  ~(g >= {'BO'})      &
0     A    LI               True           False  False
1     A    BO               True           False  False
2     A    EQ               True           False  False
3     A    AI               True           False  False
4     B    LI               True            True   True
5     B    EQ               True            True   True
6     B    AI               True            True   True
7     C    LI               True           False  False
8     C    BO               True           False  False
9     C    EQ               True           False  False
10    C    AI               True           False  False
11    D    EQ               True            True   True
12    D    AI               True            True   True
13    E    AI              False            True  False

# agg approach
                     g  (g >= {'EQ', 'AI'})  ~(g >= {'BO'})      &
NAME                                                              
A     {LI, EQ, BO, AI}                 True           False  False
B         {LI, EQ, AI}                 True            True   True
C     {LI, EQ, BO, AI}                 True           False  False
D             {EQ, AI}                 True            True   True
E                 {AI}                False            True  False