MySQL Get all users who performed same actions as current user - mysql

Question

I have to do a query and I can't figure it out. I have an actions table ( user_id , action, created_at ), and I need to retrieve all users who performed the same actions as current_user ( in exact order).
ex.
current_user delete 2022/03/19 13:40
current_user add_post 2022/03/19 13:45
current_user write_comment 2022/03/22 13:48
Query result:
user_5 delete 2021/03/15 14:50
user_5 add_post 2021/05/15 13:50
user_5 write_comment 2022/06/06 14:30
user_6 delete 2021/03/15 14:50
user_6 add_post 2021/05/15 13:50
user_6 write_comment 2022/06/06 14:30
( all users with same actions )

Answer 1

You don't stipulate that the exact matching has to be in groups of 3. The following query will identify exact action sequences of 1, 2, 3 or even more then 3:
CREATE TABLE actions(
user_id VARCHAR(50) NOT NULL
,action VARCHAR(50) NOT NULL
,created_at DATE NOT NULL
);
INSERT INTO actions(user_id,action,created_at)
VALUES
('current_user','delete','2022/03/19 13:40')
, ('current_user','add_post','2022/03/19 13:45')
, ('current_user','write_comment','2022/03/22 13:48')
, ('other_user','delete','2022/02/19 13:40')
, ('other_user','add_post','2022/02/19 13:45')
, ('other_user','write_comment','2022/02/22 13:48')
, ('diff_user','delete','2022/02/19 12:40')
, ('diff_user','add_post','2022/02/19 12:42')
, ('diff_user','other_action','2022/02/19 12:45')
, ('diff_user','write_comment','2022/02/22 12:48')
;
It appears (through a comment) that you can use the lag() so I suggest using row_number() to match the action sequences of current user to other users, as follows:
SELECT ou.*
FROM (
SELECT *
, row_number() OVER (
PARTITION BY user_id ORDER BY created_at
) AS rn
FROM actions
WHERE user_id <> 'current_user'
) AS OU
INNER JOIN (
SELECT *
, row_number() OVER (
ORDER BY created_at
) AS rn
FROM actions
WHERE user_id = 'current_user'
) AS CU ON ou.rn = cu.rn
AND ou.action = cu.action
result
+------------+---------------+---------------------+----+
| user_id | action | created_at | rn |
+------------+---------------+---------------------+----+
| diff_user | delete | 2022-02-19 12:40:00 | 1 |
| diff_user | add_post | 2022-02-19 12:42:00 | 2 |
| other_user | delete | 2022-02-19 13:40:00 | 1 |
| other_user | add_post | 2022-02-19 13:45:00 | 2 |
| other_user | write_comment | 2022-02-22 13:48:00 | 3 |
+------------+---------------+---------------------+----+
Now if you really did want to limit this to sequence match of just 3, then you could subsequently count(*) over(partition by user_id) then filter for when that calculation is 3:
SELECT *
FROM (
SELECT ou.*
, count(*) OVER (PARTITION BY ou.user_id) AS cn
FROM (
SELECT *
, row_number() OVER (
PARTITION BY user_id ORDER BY created_at
) AS rn
FROM actions
WHERE user_id <> 'current_user'
) AS OU
INNER JOIN (
SELECT *
, row_number() OVER (
ORDER BY created_at
) AS rn
FROM actions
WHERE user_id = 'current_user'
) AS CU ON ou.rn = cu.rn
AND ou.action = cu.action
) d
WHERE cn = 3
result
+------------+---------------+---------------------+----+----+
| user_id | action | created_at | rn | cn |
+------------+---------------+---------------------+----+----+
| other_user | write_comment | 2022-02-22 13:48:00 | 1 | 3 |
| other_user | add_post | 2022-02-19 13:45:00 | 2 | 3 |
| other_user | delete | 2022-02-19 13:40:00 | 3 | 3 |
+------------+---------------+---------------------+----+----+
for reference db<>fiddle (nb: using postgres as MySQL 8 wasn't available at the time)
btw: you probably also need to introduce the concept of "session" into this logic but that is left for you to consider.