Analytical function need to work based on previous rows result.

Question

My data are sorted based on sartdatetime column.but i need the sum(value) based on below condition.
Condition :
i need to check with current row with previous rows
if they are same then i need to sum the value with previous row.
else i need to start the summation.
My data like below:
Rownum     Code     Sartdatetime     Value
1     4619     12/9/2012 11:00      219
2     5344     12/9/2012 14:39      27
3     4619     12/9/2012 15:11     20
4     4619     12/9/2012 19:33     14
5     4619     12/9/2012 20:53      16
6     6851     12/9/2012 21:21      1
7     4619     12/9/2012 21:22     35
8     5623     12/10/2012 1:59     4
9     5623     12/10/2012 2:03 3
10     5623     12/10/2012 2:06 5
11     4619     12/10/2012 2:17 1
12     5623     12/10/2012 2:18     5
13     5623     12/10/2012 2:25      2
14     5623     12/10/2012 2:27 2
15     4619     12/10/2012 2:29      30
Eg:
Take the first row. For that no previous rows. So we directly added the value like below.
Rownum     Code     Sartdatetime     Value sum(val)
1     4619     12/9/2012 11:00      219 219
take the seconds row, we can check the code value (5344)with first row value (4619). the values are different . So our summation like.
2     5344     12/9/2012 14:39      27 27
for third row , we can check the code value (4619)with second row value (5344). the values are different . So our summation like.
3     4619     12/9/2012 15:11     20 20
for fourth row we can check the code value (4619)with second row value (4619). the values are same. So our summation like.
4     4619     12/9/2012 19:33     14 34 - The sum of previous row result.
5 th row also same value so
5     4619     12/9/2012 20:53      16 50T he sum of previous row result.
then the same value come again in 7 th row so here we need to start the summation like below.
7     4619     12/9/2012 21:22     35 35.
i want the result like below.
Rownum     Code     Sartdate time     Value Expected result
1     4619     12/9/2012 11:00      219 219
2     5344     12/9/2012 14:39      27 27
3     4619     12/9/2012 15:11     20 20
4     4619     12/9/2012 19:33     14 34
5     4619     12/9/2012 20:53      16 50
6     6851     12/9/2012 21:21      1 1
7     4619     12/9/2012 21:22     35 35
8     5623     12/10/2012 1:59     4 4
9     5623     12/10/2012 2:03 3 7
10     5623     12/10/2012 2:06 5 12
11     4619     12/10/2012 2:17 1 1
12     5623     12/10/2012 2:18     5 5
13     5623     12/10/2012 2:25      2 7
14     5623     12/10/2012 2:27 2 9
15     4619     12/10/2012 2:29      30 30
Please help

Answer 1

Hi,
Welcome to the forum!
Here's one way to do what you want:
WITH     got_grp_id     AS
     SELECT  code, starttime, value
     ,     ROW_NUMBER () OVER ( ORDER BY      starttime )     AS r_num
     ,     ROW_NUMBER () OVER ( ORDER BY      starttime )
           - ROW_NUMBER () OVER ( PARTITION BY  code
                                     ORDER BY          starttime
                       )                    AS grp_id
     FROM    table_x
SELECT       r_num
,       code
,       starttime
,       value
,       SUM (value) OVER ( PARTITION BY  code, grp_id
                                ORDER BY          starttime
                  )         AS sum_value
FROM      got_grp_id
ORDER BY  starttime
;For an explanation of the Fixed Difference technique, see {message:id=9953384} and/or {message:id=9957164}
Whenever you have a problem, please post CREATE TABLE and INSERT statements for your same data. For example:
CREATE TABLE     table_x
(   code     NUMBER (4)
,   starttime     DATE          UNIQUE
,   value     NUMBER
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/9/2012 11:00', 'MM/DD/YYYY HH24:MI'),
                                        219);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5344, TO_DATE ( '12/9/2012 14:39', 'MM/DD/YYYY HH24:MI'),
                                        27);
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/9/2012 15:11', 'MM/DD/YYYY HH24:MI'),
                                        20);
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/9/2012 19:33', 'MM/DD/YYYY HH24:MI'),
                                        14);
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/9/2012 20:53', 'MM/DD/YYYY HH24:MI'),
                                        16);
INSERT INTO table_x (code, starttime, value)
       VALUES         (6851, TO_DATE ( '12/9/2012 21:21', 'MM/DD/YYYY HH24:MI'),
                                        1);
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/9/2012 21:22', 'MM/DD/YYYY HH24:MI'),
                                        35);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5623, TO_DATE ( '12/10/2012 1:59', 'MM/DD/YYYY HH24:MI'),
                                        4);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5623, TO_DATE  ('12/10/2012 2:03' , 'MM/DD/YYYY HH24:MI' ),
                                                   3);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5623, TO_DATE ( '12/10/2012 2:06', 'MM/DD/YYYY HH24:MI'),
                                        5);
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/10/2012 2:17', 'MM/DD/YYYY HH24:MI'),
                                        1);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5623, TO_DATE ( '12/10/2012 2:18', 'MM/DD/YYYY HH24:MI'),
                                        5);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5623, TO_DATE ( '12/10/2012 2:25', 'MM/DD/YYYY HH24:MI'),
                                        2);
INSERT INTO table_x (code, starttime, value)
       VALUES         (5623, TO_DATE  ('12/10/2012 2:27' , 'MM/DD/YYYY HH24:MI' ),
                                                   2);
INSERT INTO table_x (code, starttime, value)
       VALUES         (4619, TO_DATE ( '12/10/2012 2:29', 'MM/DD/YYYY HH24:MI'),
                                        30);
COMMIT; See the forum FAQ {message:id=9360002}
Output:
R_NUM       CODE STARTTIME             VALUE  SUM_VALUE
    1       4619 12/9/2012 11:00         219        219
    2       5344 12/9/2012 14:39          27         27
    3       4619 12/9/2012 15:11          20         20
    4       4619 12/9/2012 19:33          14         34
    5       4619 12/9/2012 20:53          16         50
    6       6851 12/9/2012 21:21           1          1
    7       4619 12/9/2012 21:22          35         35
    8       5623 12/10/2012 01:59          4          4
    9       5623 12/10/2012 02:03          3          7
   10       5623 12/10/2012 02:06          5         12
   11       4619 12/10/2012 02:17          1          1
   12       5623 12/10/2012 02:18          5          5
   13       5623 12/10/2012 02:25          2          7
   14       5623 12/10/2012 02:27          2          9
   15       4619 12/10/2012 02:29         30         30Edited by: Frank Kulash on Dec 24, 2012 6:44 AM
Added sample data.