Skip to content
Advertisement

How to look for missing rows within a group?

Consider the data below. I am trying to find situations where in a specific RequestID, there is an Attempt: 0, but Attempt: 2 is missing.

I’ve tried looking for Attempt: 0 with a WHERE predicate and doing NOT EXISTS ... Attempt: 2 in a subquery, but it doesn’t return the right data.

How do I find the RequestIDs with missing Attempt: 2?

ID      Message       RequestID
635828  Attempt: 0    1
635968  Attempt: 1    1
641085  Attempt: 2    1
641230  Attempt: 3    1
643859  Attempt: 0    2
645991  Attempt: 1    2
650255  Attempt: 3    2
652388  Attempt: 0    3
654520  Attempt: 1    3
658785  Attempt: 3    3

Advertisement

Answer

You can use not exists like this:

select t.*
from t
where t.message = 'Attempt: 0' and
      not exists (select 1
                  from t t2
                  where t2.requestid = t.requestid and
                        t2.message = 'Attempt: 2'
                 );

Another possibility is aggregation:

select requestid
from t
where message in ('Attempt: 0', 'Attempt: 2')
group by requestid
having sum(case when message = 'Attempt: 2' then 1 else 0 end) = 0;
User contributions licensed under: CC BY-SA
8 People found this is helpful
Advertisement