基于元组时间戳创建访问矩阵作为输入

Question

I am doing a project for university and I am blocked from some days on a problem.

For start, after some manipulation on the entry data, I have this:

d = pd.DataFrame({
    'ID':["007",  "001", "009"], 
    'users': [[("us1", "us2", "1577839066196", '1589200898463'), 
               ('us2', "us3", '1589476569647', '1589476734542'), 
               ('us5', 'us1', '1586234607616', '1589195456609'),
               ('us5', 'us1', '1586234607618', '1589195456689')], 
              [("us2", "us3", '1589301928018', '1589463287633'),
               ("us3", "us2", '1589463287633', '1589469006691')], 
              [('us1', 'us2', '1589931863229', '1589931878670')]] })

The 'users' are a list of tuples of (user1, user2, timestamp user1, timestamp user2). This is the list of users that, on these timestamps, accessed the ID.

What I want to do is to create a matrix with the accesses count, that I am calling 'access_interest'. So it would be:

For each (user1, ID, timestamp1) where to_date(timestamp1) < T
  For each user user2:
    If (user1, user2) exists for this ID
      access_interest(user)[ID] += 1

Edit

The expected output should be:

On the picture you can see that 'us2' for '007' has the number 1. This is because on the first 'for each' when we fix 'us1' and '007' we will have (us1, us2) existent for 007, so we add 1 on the us2.

The same for us3, when we fix 'us2' and '007' on the first 'for each' we will have (us2, us3), so we add one on us3 for the 007.

Answer 1

In [223]: d['users_list'] = d['users'].apply(lambda x: [(y[0]) for y in x ]if isinstance(x,list) else [x[0]])

In [224]: all_users = sorted(list(set(sum([x for x in d['users_list']],[]))))

In [225]: for us in all_users:
     ...:     d[us] = d['users_list'].apply(lambda x :  1 if us in x else 0)
     ...:

In [226]: d
Out[226]:
    ID                                              users            users_list  us1  us2  us3  us5
0  007  [(us1, us2, 1577839066196, 1589200898463), (us...  [us1, us2, us5, us5]    1    1    0    1
1  001  [(us2, us3, 1589301928018, 1589463287633), (us...            [us2, us3]    0    1    1    0
2  009           (us1, us2, 1589931863229, 1589931878670)                 [us1]    1    0    0    0

output:

In [227]: d.set_index(['ID'])[all_users]
Out[227]:
     us1  us2  us3  us5
ID
007    1    1    0    1
001    0    1    1    0
009    1    0    0    0