Trouble trying to get correct AP Check Amount total on a report

Question

Okay, the challenge is to try and explain my situation well. I am in Discoverer Plus. I am working to develop a Discoverer report that will show, for accounts payable application, a summary report by bank name and job number, a check count, invoice count, and check amount total. For simplicity sake, since I have the problem once I go past one table, I am dealilng with tables ap_checks_all and ap_invoice_payments_all from EBS, the tables joined on check id. Now, if I work just with the ap_checks_all table, I can easily get the correct sum of check amounts (column name in the table is AMOUNT). However when I bring in the invoice payments table, then I run into trouble. I first did the SUM data point for AMOUNT, and got a number that was way too high. Figured out that is because anytime I have multiple invoices on a check (which is most of the time), the check amount gets added in for each invoice. So if check amount is $100.00 and the check pays 5 invoices, $500.00 is added to the total amount, instead of the just the $100.00.
So I tried the SUM_DISTINCT function next - SUM_DISTINCT(AMOUNT). Got me a lot closer, but now my total was too low. Figured out that was because if I had separate checks with the same check amount, that check amount only gets added in one time. Say Check 18 is for $100.00 and Check 39 is for $100.00, I only get $100.00 into the total instead of $200.00. Which when you think about it, is exactly what the SUM_DISTINCT function means. Since the chances of check amounts being duplicated are low, that is why I get close to the right number, but when dealing with tens of thousands of checks, check amount duplication does happen.
So on my totals for the SUM_DISTINCT column, I finally figured out that a CELL SUM got me closer to the right number. I only had a problem now when, on the same bank name and job number, would still have duplicate check amounts at that key (group by) status level. Percentage wise, my error is low - off maybe $200,000 out of $180,000,000. So close to the right number, yet so far.
The same thing happens in raw SQL. If you think about the raw join result, you can see what is happening -
check 1, check 1 amount, invoice 1
check 1, check 1 amount, invoice 2
check 2, check 2 amount, invoice 3
check 3, check 3 amount, invoice 4
So if you do SUM(CHECK AMOUNT), well yeah, it will add in the check 1 amount twice.
If you do SUM_DISTINCT(CHECK AMOUNT), then check 1 amount only gets counted once. Great. Ahh, but what if check 3 amount = check 1 amount? Oops, the check 3 amount is excluded.
Does anyone know of a way to get around this problem? I have been playing around with the functions, but the SQL reference does not give many examples to help understand all the various optional parameters. I know, in vague concept, what I want. Something like -
SUM(AMOUNT) DISTINCT ON CHECK_ID, or
SUM_DISTINCT(AMOUNT) DISTINCT ON CHECK_ID.
CHECK_ID is the internal Oracle id number for each check, from the ap_checks_all table. Check number is not unique enough, since different bank accounts may be using the same check number in the same time period. So I want a way to be able to add my check amount to the total one time when there are multiple invoices on the check, and then get each invidual check amount (one time) added up to get my check total amount.
Like I said, no problems with Discoverer or raw SQL if doing just one table. But once I join checks to invoice payments, poof, I run into trouble.
I did not think I was trying to do anything unusual. So quite surprising that having so much trouble trying to find a way around this problem.
Hope this detail explanation makes some sense. Thanks for any help/insight anyone can give me.
John Dickey

Answer 1

Michael,
Nope, that does not work. It truly just gives me an average. My desired (table) report row format is this -
Bank name , Job Number, Check Count , Invoice Count , Check Amount Total
So this is a SUMMARY report. Not a detail report. If I do detail, I do not have any problems - I get my correct check amount total. If I do just the ap_checks_all table, I get the correct amount (but then cannot get my invoice count). I get the correct check count (count distinct on check id). I get the correct invoice count. When I added a calculation column using the Average function as you suggested, I see on each row what is likely the average check amount. Which is exactly what you would expect to see.
Now I did get a brainstorm and tried this. I calculated both the average and average distinct on check amount, then multiplied each by my check count. Darn. Still does not work. The Average Distinct * Check Count = SUM_DISTINCT(CHECK AMOUNT), which I already have. Average Distinct apparently only includes a duplicate check amount one time in calculating the average (which when you think about it, makes sense, darn it). The AVERAGE function still has the problem, it looks like, of including the check amount each time on checks with multiple invoices (and thus multiple rows in the join result) in determining the average. Which when you think about the raw join result from joining checks to invoice payments tables, makes sense. The result of AVERAGE * CHECK COUNT is better than the SUM(AMOUNT) column, but not as good as the SUM_DISTINCT(AMOUNT) column.
Any other ideas? Interesting that something that seems like a pretty basic thing to do, if it does have a solution, is not a very obvious/simple solution.
John Dickey