Python Pandas：组内的行数和另一组的行数

这是一个棘手的问题.您希望计算组内的累积计数，但对于所有后续组，您需要跟踪已增加的数量，以便知道要应用的偏移量.这可以通过max+cumsum来完成，其中cumcount组比之前的组多.这里唯一复杂的是，您需要确定之前和后续组标签之间的关系，以防后续组的标签之间没有简单的+1增量.

# Cumcount within group
s = df.groupby(['col_1', 'col_2']).cumcount()

# Determine how many cumcounts were within all previous groups of `col_1' 
to_merge = s.add(1).groupby(df['col_1']).max().cumsum().add(1).to_frame('new')

# Link group with prior group label
df1 = df[['col_1']].drop_duplicates()
df1['col_1_shift'] = df1['col_1'].shift(-1)
df1 = pd.concat([to_merge, df1.set_index('col_1')], axis=1)

# Bring the group offset over
df = df.merge(df1, left_on='col_1', right_on='col_1_shift', how='left')

# Add the group offset to the cumulative count within group.
# First group (no previous group) is NaN so fill with 1.
df['new'] = df['new'].fillna(1, downcast='infer') + s

# Clean up merging column
df = df.drop(columns='col_1_shift')

    col_1 col_2  col_3  new
0       1     A      1    1
1       1     B      1    1
2       2     A      3    2
3       2     A      3    3
4       2     A      3    4
5       2     B      3    2
6       2     B      3    3
7       2     B      3    4
8       3     A      2    5
9       3     A      2    6
10      3     C      2    5
11      3     C      2    6