Assigning sequential numbers to multiple rows with a common attribute

Question

Oracle 10i, looking to modify a query to only select a total of 96 unique compoundnames and number each group of batchids 1-96.
Input
Table EX
Unique_ID, CompoundName, BatchID, Quantity
1, EX 1000, EX 1000 100, 100
2, EX 1000, EX 1000 101, 100
3, EX 1000, EX 1000 102, 100
4, EX 1001, EX 1001 100, 100
5, EX 1001, EX 1001 101, 100
6, EX 1001, EX 1001 102, 100
7, EX 1001, EX 1001 103, 100
8, EX 1001, EX 1001 104, 100
9, EX 1001, EX 1001 105, 100
10, EX 1002, EX 1002 100, 100
11, EX 1002, EX 1002 101, 100
12, EX 1002, EX 1002 102, 100
100+, EX 1095, EX 1095 100, 100
Using
select distinct * from
Select Unique_id, CompoundName, batchid, quantity, SUM(QUANTITY) over (partition by CompoundName) sumqty
FROM EX
where sumqty < 500
Order by 2,3;
Gives the data batched together by compoundnames and the total quantity of all the data sharing the same compound name (and only results whose total quantity is < 500).
I'm looking for a way to only grab 96 unique compoundnames (not total rows of data) at a time and numbering them to look like this desired output;
Unique_ID, CompoundName, BatchID, Quantity, SUM(Quantity), ???
1, EX 1000, EX 1000 100, 100, 300, 1
2, EX 1000, EX 1000 101, 100, 300, 1
3, EX 1000, EX 1000 102, 100, 300, 1
10, EX 1002, EX 1002 100, 100, 300, 2
11, EX 1002, EX 1002 101, 100, 300, 2
12, EX 1002, EX 1002 102, 100, 300, 2
100+, EX 1095, EX 1095 100, 100, 100, 96
Thanks in advance.
Edited by: user13112928 on Jun 26, 2010 11:49 AM
Edited by: user13112928 on Jun 26, 2010 11:50 AM
Edited by: user13112928 on Jun 26, 2010 12:01 PM

Answer 1

Hi,
If you want the numbers to be assigned only to the groups that remain after the groups where sumqty >= 500 or more have been excluded, then you have to assign the numbers after the groups where sumqty >= 500 or more have been excluded. All analytic funtions in the same sub-query are computed at the same time, so if you want to use analytic functions only, you have to do something like this:
SELECT DISTINCT
FROM     (
          SELECT      s.*
          ,     DENSE_RANK () OVER (ORDER BY compoundname)     AS rnk
          FROM     (
                    SELECT     Unique_id
                    ,      CompoundName
                    ,      batchid
                    ,      quantity
                    ,      SUM (quantity) OVER (PARTITION BY CompoundName)     AS sumqty
                    FROM     ex
               )     s
          WHERE     sum_qty     < 500
WHERE       rnk     <= 96
ORDER BY  batchid
,       quantity
;Since analytic funtions are evaluated after the GROUP BY and HAVING clauses, you could also do it this way, with only one sub-query:
WITH     got_rnk          AS
     SELECT       CompoundName
     ,       SUM (quantity)               AS sumqty
     ,       ROW_NUMBER () OVER (ORDER BY  CompoundName)     AS rnk
     FROM       ex
     GROUP BY  CompoundName
     HAVING       SUM (quantity)     < 500
SELECT DISTINCT
       e.unique_id
,       e.CompoundName
,       e.batchid
,       e.quantity
,       r.sumqty
,       r.rnk
FROM       ex          e
,       got_r_num     r     ON     e.CompoundName     = r.CompoundName
WHERE       r.rnk     <= 96
ORDER BY  e.batchid
,       e.quantity
;I'm not sure the latter will be any faster, since it needs to make two passes through ex.