Help with aging query. not sure how to acomplish this???

Question

Guys im a bit stuck usually i have a slight idea how to even start my query but for this particular one I just dont even know where to start to tackle
i have a table like so.
       create table TRRACCD (
       TRRACCD_GRNT_CODE varchar2(6),
       TRRACCD_TRANS_dATE date,
       TRRACCD_AMOUNT number(17,2),
       trraccd_detail_code varchar2(6))with data like so
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('15-FEB-10', 'DD-MM-RR'),19764.77,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('12-MAR-10', 'DD-MM-RR'),2054.29,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('20-APR-10', 'DD-MM-RR'),4111.46,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('14-MAY-10', 'DD-MM-RR'),2570.1,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('10-JUN-10', 'DD-MM-RR'),2176.16,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('12-JUL-10', 'DD-MM-RR'),4756.29,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('13-AUG-10', 'DD-MM-RR'),28500.62,'GCSH');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('16-AUG-10', 'DD-MM-RR'),3602.3,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('17-SEP-10', 'DD-MM-RR'),3995.79,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('11-OCT-10', 'DD-MM-RR'),3413.6,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('11-NOV-10', 'DD-MM-RR'),2119.86,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('10-DEC-10', 'DD-MM-RR'),5905.52,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('10-DEC-10', 'DD-MM-RR'),12008.21,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('11-JAN-11', 'DD-MM-RR'),-10574.05,'GBIL');
INSERT INTO TRRACCD (TRRACCD_GRNT_CODE, TRRACCD_TRANS_DATE, TRRACCD_AMOUNT, TRRACCD_DETAIL_CODE) VALUES ('400364',to_date('11-JAN-11', 'DD-MM-RR'),20241.9,'GBIL');however what i need to do is end up with an aging for the items summarize so the expected output should be something like
tbrraccd_grnt_Code      30-60                    60-90         over90
400364                      20,241.90              2119.86       7,370.09 
if your confused about how i arrived at the 7370.90 is because i have to subtract entries with GCSH on them and i also need to take
entries that are - and applied them to the oldest period if the balance of the period is positive. im inclined to say that i can do this with sql but the only other way i can think of is using plsql?... any suggestions or solutions would be greatly appreciated.
the closest that i have gotten to acomplishing this query without pl is this. i know is lame but i need help.
SELECT TRRACCD_GRNT_CODE,
       TRRACCD_TRANS_dATE,
       TRRACCD_AMOUNT,
       trraccd_detail_code,
       round(sysdate - TRRACCD_TRANS_dATE) "total days"
       --sum(case when round(sysdate - TRRACCD_TRANS_dATE)>190 then "greater then 190"end )
       FROM TRRACCD
       where trraccd_grnt_Code = '400364'
       ORDER BY TRRACCD_GRNT_CODE,
                TRRACCD_TRANS_dATE,
                trraccd_detail_code;  

Answer 1

Hi, Miguel,
I should have mentioned this before: I'm assuming SYSDATE is some point on January 22, 2011, after 00:00. I don't think this has caused any confusion so far, but if the discussion continues for a day or two, let's be clear about that.
I think I see what you want regarding the rows where trraccd+detailcode='GSCH'. If we completly ignored all the GSCH rows, we would get this output:
TRRACC       0-29      30-59      60-89    90-over
400364   ...
400404   19904.97    6625.27    4757.01   67618.68You want GSCH amounts to cancel other amounts in the last column only until that column reaches 0 ; the remaining GSCH amount will cancel amounts in to "60-89" column, again, only until that amount reaches 0 . If there is anything left over, it will cancel amounts in the "30-59" column, and if there is still some left over, the "0-29" column.
To be specific, trraccd_grnt_code=400404 has a total of 72475.69 from GSCH rows. That counts against the "90-over" total.
67618.68 - 72475.69 = -4757.01; you want to dispaly 0 in the "90-over" column, and apply the remaining 4757.01 to the next column to the left.
That column contains 4757.01, which happens to be the exact maount left over from the previous subtraction, so you want to diesplay 0 in the "60-89" column.
The remaining columns to the left should contain their regular totals.
Here's one way to do that:
WITH     got_col_num     AS
     SELECT     trraccd_grnt_code
     ,     CASE
               WHEN  trraccd_detail_code != 'GCSH'     THEN  trraccd_amount
          END          AS net_amount
     ,     SUM ( CASE
                    WHEN trraccd_detail_code = 'GCSH'     THEN  trraccd_amount
                END
              ) OVER (PARTITION BY trraccd_grnt_code)     
                             AS gsch_total
     ,     CASE
              WHEN  trraccd_amount < 0               THEN  4
              WHEN  trraccd_trans_date > SYSDATE - 30     THEN  1
              WHEN  trraccd_trans_date > SYSDATE - 60     THEN  2
              WHEN  trraccd_trans_date > SYSDATE - 90     THEN  3
                                        ELSE  4
          END          AS col_num
     FROM     trraccd
     WHERE     trraccd_trans_date     <= SYSDATE
,     got_sum          AS
     SELECT       trraccd_grnt_code
     ,       col_num
     ,       LEAST ( SUM (net_amount)
                 , GREATEST ( 0
                            , SUM (SUM (net_amount)) OVER ( PARTITION BY  trraccd_grnt_code
                                                             ORDER BY        col_num   DESC
                              - gsch_total
               )  AS total
     FROM       got_col_num
     GROUP BY  trraccd_grnt_code
     ,       col_num
     ,       gsch_total
SELECT       trraccd_grnt_code
,       SUM (CASE WHEN col_num = 1 THEN total END)     AS "0-29"
,       SUM (CASE WHEN col_num = 2 THEN total END)     AS "30-59"
,       SUM (CASE WHEN col_num = 3 THEN total END)     AS "60-89"
,       SUM (CASE WHEN col_num = 4 THEN total END)     AS "90-over"
FROM       got_sum
GROUP BY  trraccd_grnt_code
ORDER BY  trraccd_grnt_code
;Whenever you want to understand a query like this, run just the first sub-query (in this case, got_col_num) and make sure you understand what it's doing. When you do, add one more sub-query, and make sure you understand what it's doing.
Add morte columns to the display to help you understand it better. For example, when you get to the point where you're trying to understand the 2nd sub-query above, got_sum, you might run this query:
WITH     got_col_num     AS
     SELECT     trraccd_grnt_code
     ,     CASE
               WHEN  trraccd_detail_code != 'GCSH'     THEN  trraccd_amount
          END          AS net_amount
     ,     SUM ( CASE
                    WHEN trraccd_detail_code = 'GCSH'     THEN  trraccd_amount
                END
              ) OVER (PARTITION BY trraccd_grnt_code)     
                             AS gsch_total
     ,     CASE
              WHEN  trraccd_amount < 0               THEN  4
              WHEN  trraccd_trans_date > SYSDATE - 30     THEN  1
              WHEN  trraccd_trans_date > SYSDATE - 60     THEN  2
              WHEN  trraccd_trans_date > SYSDATE - 90     THEN  3
                                        ELSE  4
          END          AS col_num
     FROM     trraccd
     WHERE     trraccd_trans_date     <= SYSDATE
,     got_sum          AS
     SELECT       trraccd_grnt_code
     ,       col_num
     ,       gsch_total
     ,       SUM (net_amount)               AS period_sum
     ,       SUM (SUM (net_amount)) OVER ( PARTITION BY  trraccd_grnt_code
                                          ORDER BY      col_num     DESC
                               )         AS running_sum
     ,       LEAST ( SUM (net_amount)
                 , GREATEST ( 0
                            , SUM (SUM (net_amount)) OVER ( PARTITION BY  trraccd_grnt_code
                                                             ORDER BY        col_num   DESC
                              - gsch_total
               )  AS total
     FROM       got_col_num
     GROUP BY  trraccd_grnt_code
     ,       col_num
     ,       gsch_total
SELECT       *
FROM       got_sum
ORDER BY  trraccd_grnt_code
,            col_num
;which produces this output:
TRRACC    COL_NUM GSCH_TOTAL PERIOD_SUM RUNNING_SUM      TOTAL
400364          1   28500.62    20241.9     76146.2    20241.9
400364          2   28500.62   17913.73     55904.3   17913.73
400364          3   28500.62    2119.86    37990.57    2119.86
400364          4   28500.62   35870.71    35870.71    7370.09
400404          1   72475.69   19904.97    99005.93   19904.97
400404          2   72475.69    6625.27    79100.96    6625.27
400404          3   72475.69    4857.01    72475.69          0
400404          4   72475.69   67618.68    67618.68          0The total column in that sub-query is very complicated, so break it down some. You can see that it is based on an analytic SUM whose argumenet is an aggregate SUM, all wrapped in LEAST and GREATEST. That can be very hard to underrstand, so break it down. Display the aggregate SUM by itself (period_sum) and also display the analytic SUM, without the GREATEST and LEAST.
mlov83 wrote:
where can i get more information about these types of querys?Can you be more specific? What types of queries do you mean? Complicated queries? Pivots? GROUP BY?
... I'm always leary of using a solution that i dont understand Excellent reaction!
would you mind expalining it to me a little.Again, could you be more specific? I don't mind explaining things to you, but I don't want to explain parts you already understand. If something is new, you'll want to read the SQL Language manual, or other sources, anyway, so it's more efficient to do that first, and then post specific questions on this forum.
An example of a more specific question is: "When I looked up "WITH clause" in the SQL Language reference, I found this page:
http://download.oracle.com/docs/cd/B19306_01/server.102/b14200/statements_10002.htm#sthref9758
but I don't understand where it says ...
The query above says ... Is that what the manual is talking about when it says ...?"
Another example of a good question is
"The query you posted says ... I don't understand what that's doing. I looked for ... and ... but that didn't seem to have anything to do with this query. What should I look for?
The Oracle 10 SQL Language manual is indexed pretty poorly. If you can't find something, it may not be your fault.