Business
Jobs
  • About Us
  • Solutions
    • Job Postings
      Post your job and receive qualified candidates in 48h.
    • Candidate Assessments
      500+ technical and psychological tests, plus anti-fraud.
    • Headhunting
      Tailor-made executive search from start to finish.
    • Payroll + EOR
      Payroll dispersal and EOR across 15+ LATAM countries.
  • Pricing
  • Jobs

0

173
Views
Postgresql OVER and GROUP

I have a problem with grouping

I have a table(a lot more stuff in it, but not relevant) that looks something like:

id user
0 1
1 1
2 1
3 2
4 2
5 2
6 1
7 1

I'm trying to get the following value:

user start end
1 0 2
2 3 5
1 6 7

Basicaly, i need the first and last occurance of user, while not messing with order. I know i need to use OVER (PARTITION BY ...), But i've never used it and not sure how to build this query. if i "partition by user", it ignores order. And if i "partition by id, user" it again returns wrong.

Example of what i tried(doesn't even try to get what i need, but is a midpoint that shows me how to do it once i figure out "over" part):

SELECT user, count(user) over (partition by user):
user count
1 5
1 5
1 5
2 3
2 3
2 3
1 5
1 5

SELECT user, count(user) over (partition by id, user):
user count
1 1
1 1
1 1
2 1
2 1
2 1
1 1
1 1
over 4 years ago · Santiago Trujillo
2 answers
Answer question

0

Use a difference of row numbers approach to classify consecutive rows with the same user into one group and start over a new group when a new user is encountered. Thereafter, use group by to get the start and end of each group.

SELECT USER,MIN(ID) AS START,MAX(ID) AS END
FROM (SELECT user,id, row_number() over(order by id) 
                     - row_number() over (partition by user order by id) as grp
      FROM tablename
      ) T
GROUP BY USER,GRP
over 4 years ago · Santiago Trujillo Report

0

to get start, end of id, use:

SELECT user, min(id) over (partition by user) "start", max(id) over (partition by user) "end"
from table_name;

Update My answer was based on wrong predicate and so wrong. To provide right one and not duplicate @vkp one, I made this monstrous construct:

create table so74 as
select * from (values (0, 1), (1, 1), (2, 1), (3, 2), (4, 2), (5, 2), (6, 1), (7, 1)) t(id, u); 

with d as (
    with c as (
        with b as (
            select 
                *
                , case when lag(u) over (order by id) <> u or id = min(id) over() then id end min
                , case when lead(u) over (order by id) <> u or id=max(id) over () then id end max 
            from so74
            )
        select  u, min,max 
        from b 
        where coalesce(min,max) is not null
    ) 
    select u,min,lead(max) over () max  
    from c
) 
select * 
from d 
where coalesce(min,max) is not null
;

 u | min | max
---+-----+-----
 1 |   0 |   2
 2 |   3 |   5
 1 |   6 |   7
(3 rows)

Time: 0.456 ms
over 4 years ago · Santiago Trujillo Report
Answer question
Find remote jobs

Discover the new way to find a job!

Top jobs
Top job categories
Business
Post vacancy Pricing Sales
Legal
Terms and conditions Privacy policy
© 2026 PeakU Inc. All Rights Reserved.
Andres GPT
Show me some job opportunities
There's an error!