I want all of the accounts and amounts for accounts that occur in at least three different week ending dates. Can you help?WeekEnding Amt Acct ---------- ---- ---- 2007-10-06 5.00 XYZ 2007-10-13 3.00 XYZ 2007-10-20 0.23 XYZ 2007-10-06 5.00 ZZZ 2007-10-13 0.23 ZZZ
The difficulty with this type of problem is that it requires returning detail rows for a situation which is easily solved with HAVING. However, to use HAVING we have to use GROUP BY, and then the detail rows are aggregated. The way to approach this problem is to break it down into components. The first component of the solution does involve the HAVING clause.
select Acct from Weekly group by Acct having count(*) >= 3
Note that COUNT(DISTINCT WeekEnding) also works. Whether we use this or the simpler COUNT(*) depends on whether the combination of WeekEnding and Acct is unique. Counting distinct values may be necessary if, for example, there are other columns you didn't mention, such as customer. In that case, we might have one row per customer per account per week ending date, so COUNT(DISTINCT WeekEnding) is necessary if the GROUP BY is only on the account.
Now that we know which accounts qualify, we can build the other component, which is the selection of all detail rows for the accounts which qualify. This can be accomplished in two ways. The first method uses an IN subquery:
select WeekEnding , Amt , Acct from Weekly where Acct in ( select Acct from Weekly group by Acct having count(*) >= 3 )
The second method uses the same subquery but as a derived table in the FROM clause:
select WeekEnding , Amt , Acct from Weekly inner join ( select Acct from Weekly group by Acct having count(*) >= 3 ) as ok_accts on ok_accts.Acct = Weekly.Acct
What is the difference between these two solutions? As far as execution goes, they should perform the same. But the derived table has an advantage. Suppose that if, instead of one column, the HAVING condition needed to be based on a GROUP BY involving two columns. What if you wanted details for all customer accounts that occurred at least 3 times?
The first solution would be:
select WeekEnding , Amt , Acct , Customer from Weekly where ( Acct, Customer ) in ( select Acct , Customer from Weekly group by Acct , Customer having count(*) >= 3 )
This solution now uses what is called a row constructor, and this is perfectly valid SQL. The problem with row constructors is that not every database engine supports them. Yet.
Now imagine what the query involving the join to the derived table would look like.
Dig Deeper on Oracle and SQL
Related Q&A from Rudy Limeback
Read an example of an SQL case expression from our SQL expert Rudy Limeback. Continue Reading
Read SQL expert Rudy Limeback's advice for counting combinations in a table with SQL's GROUP BY clause Continue Reading
Read about the Mimer Validator, a tool used to verify your SQL code, in this tip from SQL expert Rudy Limeback. Continue Reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.