Python Groupby 使用序列并计算总和

您需要构建一个定制的Grouper，因此需要一个循环，这里使用一个定制函数:

def group_consecutive(s, target):
    out = []
    i = 0

    g = 1
    while i < len(s)-len(target)+1:
        if target == set(s.iloc[i:i+len(target)]):
            out.extend([g]*len(target))
            g += 1
            i += len(target)
        else:
            out.append(0)
            i += 1
    out.extend([0]*(len(target)-1))

    return pd.Series(out, index=s.index)

group = group_consecutive(df['Name'], {'a', 'b', 'c'})

out = df[group>0].groupby(group).agg({'COST': 'sum', 'Timestamp': 'last'})

输出:

   COST            Timestamp
1    27  2023-09-20 15:15:00
2    16  2023-09-20 15:15:12

中级与集团:

  Name  COST            Timestamp  group
0    c     8  2023-09-20 15:14:46      0
1    a     8  2023-09-20 15:14:48      1
2    c     9  2023-09-20 15:14:55      1
3    b    10  2023-09-20 15:15:00      1
4    c     4  2023-09-20 15:15:02      2
5    a     9  2023-09-20 15:15:04      2
6    b     3  2023-09-20 15:15:12      2
7    a     3  2023-09-20 15:15:17      0
8    c     6  2023-09-20 15:15:20      0
9    c     6  2023-09-20 15:15:29      0