How to generate unique id and sub_id for each group

问题

My goal is to generate an id (id trajectory) and a sub id (under trajectory) for each group (u_uuid and p_uuid).

I tried the ngroup function and it didn't work

data = [
{'u_uuid': 110, 'p_uuid': 'aaa', 'mode': 'walk', 'dest': 'work'}, 
{'u_uuid': 110, 'p_uuid': 'aaa', 'mode': 'walk', 'dest': 'work'}, 
{'u_uuid': 110, 'p_uuid': 'aaa', 'mode': 'bus',  'dest': 'work'},
{'u_uuid': 110, 'p_uuid': 'aaa', 'mode': 'bus',  'dest': 'work'}, 
{'u_uuid': 110, 'p_uuid': 'aaa', 'mode': 'walk', 'dest': 'work'}, 
{'u_uuid': 110, 'p_uuid': 'bbb', 'mode': 'walk', 'dest': 'home'}, 
{'u_uuid': 110, 'p_uuid': 'bbb', 'mode': 'bus',  'dest': 'home'},
{'u_uuid': 110, 'p_uuid': 'bbb', 'mode': 'bus',  'dest': 'home'}, 
{'u_uuid': 110, 'p_uuid': 'bbb', 'mode': 'walk', 'dest': 'home'}] 

df = pd.DataFrame(data)

df['id'] = df.groupby(['u_uuid', 'p_uuid', 'dest']).ngroup()
df['sub_id'] = df.groupby(['u_uuid', 'p_uuid', 'mode']).ngroup()

My dataframe:

What I m looking for:

回答1:

Use:

s1 = df.groupby(['u_uuid', 'p_uuid', 'dest'],sort=False).ngroup().add(1)
s2 = df.groupby(['u_uuid','p_uuid',
                 df['mode'].ne(df2['mode'].shift()).cumsum()],sort=False).ngroup()
df['sub_id']=s2.sub(s2.where(s1.ne(s1.shift())).ffill()).add(1).astype(int)
df['id']=s1
print(df)

   u_uuid p_uuid  mode  dest  sub_id  id
0     110    aaa  walk  work       1   1
1     110    aaa  walk  work       1   1
2     110    aaa   bus  work       2   1
3     110    aaa   bus  work       2   1
4     110    aaa  walk  work       3   1
5     110    bbb  walk  home       1   2
6     110    bbb   bus  home       2   2
7     110    bbb   bus  home       2   2
8     110    bbb  walk  home       3   2

来源：https://stackoverflow.com/questions/59772370/how-to-generate-unique-id-and-sub-id-for-each-group