A fairly simple SQL question

I have an Oracle table like so:

dealer, dealnum, lastname, firstname, emplid, date

I believe that in a few cases the emplid for a given person has changed. In this example Joe Smith's emplid changed from 00000 to 12345:

dealer dealnum lastname firstname emplid, date
10001 7700 SMITH JOE 12345 2013-10-24
10001 4533 SMITH JOE 00000 2013-07-01

Open in new window


How can I determine the dealer, dealnum, lastname and firstname of rows where there is more than one emplid associated with a given dealer+lastname+firstname combination?

In other words I'm looking for people whose emplid has changed over time.

I know how to use row_number() with group by to get the most recent row for a given person, which would give me the current value, 12345 in this example. But that's expensive and I want first to size up how many such cases there are.
FelineConspiracyAsked:
Who is Participating?
 
Steve WalesConnect With a Mentor Senior Database AdministratorCommented:
select dealer, lastname, firstname, count(*)
from mytable
group by dealer, lastname, firstname
having count(*) > 1

Open in new window


This will show you rows where that combination of columns appears more than once.
0
 
awking00Commented:
What do you want to do with the information? Delete or update the earlier records or something else?
0
 
FelineConspiracyAuthor Commented:
No deletes or updates. To oversimplify a bit: Someone will be doing some manual updates, once we're confident about the data to act on. This query is part of some precautions I'm taking to avoid updating rows where, for example, someone has had multiple valid emplids at different times, as we cannot necessarily know which one to keep. So I'll just screen those people out.
0
 
FelineConspiracyAuthor Commented:
Thank you.
0
All Courses

From novice to tech pro — start learning today.