Reports sum don't sum as they should

Question

Hi,
I created a report where it displays, in order: COST CENTER, PROJECT, EMPLOYEE, JAN, FEB... DEC
Then I check on sum for JAN, FEB.... DEC under columns attributes.
Then I have a break for the first and second columns (COST CENTER, PROJECT), so it will give me a sum for project total, and a sum for cost center total.
The thing is that there is a case that the project is the same, but in different cost center, what the report is doing is that since the project is the same, even though the cost center is different, it's returning a sum for project total combining 2 different cost centers. This should not happen since cost center is highest in the hierarchy so even 2 different cost centers with the same projects, it should return a sum of the projects separtely, since although the same project, they belong to 2 different cost centers.
How can I fix this?
Thanks,

Answer 1

I have no idea if this is "good" SQL or not, I played with it until it worked. :-)
Going through the SQL a line at a time:
select decode(grouping_id(cost_center)+grouping_id(project)+grouping_id(amount),3, 'Report Total:',2,'Cost Center '||cost_center||' Subtotal:',1,'Project '||project||' Subtotal:',0,null) tag,grouping_id(item) is set to 1 when that item level is being totaled, 0 otherwise. So this is basically a case statement. Column 1 is the 'Tag' field, which is just a label and not in the database.
If the sum of the grouping_id values is 3, then select the text 'Report Total:' for column 1
Else if the sum is 2, select 'Cost Center ' concatenated with the value of the current cost center, concatenated with 'Subtotal:' in for column 1
Else if the sum is 1, you're subtotalling the project, so select 'Project' concatenated with the project value and 'Subtotal:' for the 'Tag' column.
Otherwise, display null.
This brings up a feature I'd like to see in ApEx: It would be nice to be able to display not only the column headers in the break row, but also the value/name of the item currently being totaled. Another nice feature would be if I could get around having to use this 'Tag' field and, instead, have column 1 span several TD cells so that I could get the label over next to the subtotal values instead of in column 1.
On to line/column 2. Similar to before, this is a case statement, determining what goes in the cost_center field:
decode(grouping_id(cost_center)+grouping_id(project)+grouping_id(amount),0,cost_center,null) cost_center, If the sum of all three grouping ids is 0 (not rolling up or subtotalling), show cost_center, otherwise show null.
The next line, for project, works the same way.
decode(grouping_id(cost_center)+grouping_id(project)+grouping_id(amount),0,project,null) project, If the sum of all three grouping ids is 0 (not rolling up or subtotalling), show project, otherwise, show null.
The next field, Amount, is a little tricky:
decode(sum(amount),amount,amount,sum(amount)) amount, Again, take it as a case statement (or if-then-elseif)...
if the sum(amount) is the same as amount (displaying regular row), show the amount, otherwise, show the sum(amount).
Why do this? Because I don't want two columns for amount and sum(amount). I want the sum to display in the Amount field on a report row. I suppose I could have summed up the grouping_id values as before, eh?
In the report attributes, I suppress the display of the last field, sum(amount).
sum(amount) The final bit defines the grouping sets:
from breakreport group by grouping sets((cost_center, project, amount),(cost_center, project), (cost_center),());This says: Aggregate on amount, then project, then cost center, then give an overall total.
Google for "grouping_id oracle" and you'll find a bunch of very nice links to Ask Tom and other places that discuss grouping.