Lag with partition by.

Question

Hi,
I have a question on LAG analytical function. Here is my test scenario.
SCOTT@CRP1 > CREATE TABLE test1 AS 
SELECT 123 document_id, 0 bill_id, 12.40 prev_balance
  FROM DUAL
UNION ALL
SELECT 123, 1, 13.2
  FROM DUAL
UNION ALL
SELECT 123, 2, 1.2
  FROM DUAL
UNION ALL
SELECT 122, 1, 15.5
  FROM DUAL
Table created.
Elapsed: 00:00:00.09
SCOTT@CRP1 > select * from test1;
DOCUMENT_ID    BILL_ID PREV_BALANCE
        123          0         12.4
        123          1         13.2
        123          2          1.2
        122          1         15.5
Elapsed: 00:00:00.02
SCOTT@CRP1 > select t.*,
lag(t.prev_balance) over (partition by document_id order by document_id, bill_id) prev_bal1 from test1 t;
DOCUMENT_ID    BILL_ID PREV_BALANCE  PREV_BAL1
        122          1         15.5
        123          0         12.4
        123          1         13.2       12.4
        123          2          1.2       13.2
Elapsed: 00:00:00.01
  1* select * from
(select t.*,
lag(t.prev_balance) over (partition by document_id order by document_id, bill_id) prev_bal1 from test1 t)
where bill_id<>0
SCOTT@CRP1 > /
DOCUMENT_ID    BILL_ID PREV_BALANCE  PREV_BAL1
        122          1         15.5
        123          1         13.2       12.4
        123          2          1.2       13.2
Elapsed: 00:00:00.00Now, my problem is, for scenario where document_id = 122, I need prev_bal1 to be 15.5 instead of zero (since there is no value to be carried over from bill_id = 0 for that document_id).
Everything else should remain the same.
Is this even possible with 'LAG' function? or am I gonna have to look something exotic like 'MODEL' clause?
Thank you,
Rahul.
Message was edited by:
RPuttagunta
Message was edited by:
RPuttagunta

Answer 1

Did you try?
SQL> CREATE TABLE test1 AS  SELECT 123 document_id, 0 bill_id, 12.40 prev_balance
  2    FROM DUAL
  3  UNION ALL
  4  SELECT 123, 1, 13.2
  5    FROM DUAL
  6  UNION ALL
  7  SELECT 123, 2, 1.2
  8    FROM DUAL
  9  UNION ALL
10  SELECT 122, 1, 15.5
11    FROM DUAL
12  ;
Table created.
SQL> select *
  2    from (select t.*
  3               , lag(t.prev_balance, 1, t.prev_balance)
  4                  over (partition by document_id order by document_id, bill_id) new_prev_bal1
  5           from test1 t
  6         )
  7   where bill_id<>0
  8  /
DOCUMENT_ID    BILL_ID PREV_BALANCE NEW_PREV_BAL1
        122          1         15.5          15.5
        123          1         13.2          12.4
        123          2          1.2          13.2
SQL>