How to apply SUM function multiple times on a column?

  • Hi,

    I have a table with the following data.

    DID-------ID--------Value

    1---------1-----------2

    1---------2-----------12

    1---------3-----------45

    1---------4-----------8

    1---------5-----------63

    1---------6-----------24

    2---------1-----------89

    2---------2-----------20

    2---------3-----------46

    2---------4-----------96

    I have to write a procedure with two parameters @startposition and @endposition. If @StartPosition contains 1 and @endposition contains 3 then i have to sum first three rows i.e. 2+12+45 then next three rows i.e. 12+45+8 then 45+8+63 then 8+63+24 then 63+24 and then only 24. Same goes for next DID i.e 89+20+46 then 20+46+96 then 46+96 and then only 96. I would have same number of rows in output.

    How can i do it in one query.

    I am using SQL Server 2005. If you have any question please let me know.

    Regards,

    Sulaman

  • Hmm... And what if @startposition is 2 and @endposition 3?

    Does that mean you skip rows with 1 altogether and only need sums of rows with ID 2+3, 3+4, 4+5, 5+6, 6?

    Also, it would be nice to supply us with table definition (CREATE TABLE) and test data in form of INSERT INTO ...

  • yes you got it right.

  • What do you want the output to look like if @StartPosition contains 1 and @endposition contains 3?

    How do the @StartPosition and @endposition correspond to values in the table? It's not clear.

    “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail Shaw

    For fast, accurate and documented assistance in answering your questions, please read this article.
    Understanding and using APPLY, (I) and (II) Paul White
    Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden

  • Try something like this...this works i think

    while @startposition<=6

    (

    select SUM(Value) from

    where id >= @startposition and id<=@endposition and DID=@DID

    select @startposition=@startposition+1

    select @endposition=@endposition+1

    )

  • Here is a solution that doesn't use a loop.

    It uses a tally table.

    The only thing missing from the query is that I'm not 100% sure how the StartPostion comes into play

    DECLARE @Table TABLE

    ([DID] INT,

    [ID] INT ,

    [Value] INT)

    DECLARE @startposition INT

    DECLARE @endposition INT

    SELECT

    @startposition = 1,

    @endposition = 3

    INSERT INTO @Table

    SELECT 1,1,2 UNION ALL

    SELECT 1,2,12 UNION ALL

    SELECT 1,3,45 UNION ALL

    SELECT 1,4,8 UNION ALL

    SELECT 1,5,63 UNION ALL

    SELECT 1,6,24 UNION ALL

    SELECT 2,1,89 UNION ALL

    SELECT 2,2,20 UNION ALL

    SELECT 2,3,46 UNION ALL

    SELECT 2,4,96

    SELECT * FROM @Table

    SELECT DID,n.n,SUM([VALUE])

    FROM Tally n

    INNER JOIN @Table i

    ON n.n <= i.Id AND i.id < n.n + @endposition

    GROUP BY n.n,DID

    ORDER BY DID,n

    ----------------------------------------------
    Try to learn something about everything and everything about something. - Thomas Henry Huxley

    :w00t:
    Posting Best Practices[/url]
    Numbers / Tally Tables[/url]

    SQL-4-Life
  • If @startposition = 1 and @endposition = 3 then Output column would be something like following

    DID-------ID--------Value---------Output

    1---------1-----------2------------2+12+45 = 59

    1---------2-----------12-----------12+45+8 = 65

    1---------3-----------45-----------45+8+63

    1---------4-----------8------------8+63+24

    1---------5-----------63-----------63+24

    1---------6-----------24-----------24

    2---------1-----------89-----------89+20+46

    2---------2-----------20-----------20+46+96

    2---------3-----------46-----------46+96

    2---------4-----------96-----------96

    I have to do it in one query without using loop. My be it is possible with recursive CTE but i am not able to do it yet.

  • if the table has one millinon rows:-)

  • does my solution not work???

    I'll test it with a million rows and see what happens...

    OH and you haven't given us an example of a different start position?

    ----------------------------------------------
    Try to learn something about everything and everything about something. - Thomas Henry Huxley

    :w00t:
    Posting Best Practices[/url]
    Numbers / Tally Tables[/url]

    SQL-4-Life
  • Christopher Stobbs (3/17/2009)


    does my solution not work???

    I'll test it with a million rows and see what happens...

    OH and you haven't given us an example of a different start position?

    It matches OP's desired result set, Chris:cool:

    “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail Shaw

    For fast, accurate and documented assistance in answering your questions, please read this article.
    Understanding and using APPLY, (I) and (II) Paul White
    Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden

  • As a quick solution, I would use this:

    /*Create environment*/

    create table #test(did int, id int, myval int)

    /*create test data*/

    insert into #test(did, id, myval)

    SELECT 1, 1, 2

    UNION SELECT 1,2,12

    UNION SELECT 1,3,45

    UNION SELECT 1,4,8

    UNION SELECT 1,5,63

    UNION SELECT 1,6,24

    UNION SELECT 2,1,89

    UNION SELECT 2,2,20

    UNION SELECT 2,3,46

    UNION SELECT 2,4,96

    /*set parameters*/

    DECLARE @startposition INT, @endposition INT, @rows INT

    SELECT @startposition = 2, @endposition = 3

    SET @rows = @endposition - @startposition

    /*This does the work*/

    SELECT t1.did, t1.id, SUM(t2.myval)

    FROM #test t1

    JOIN #test t2 ON t2.did = t1.did

    AND t1.id >= @startposition

    AND t2.id >= t1.id

    AND t2.id <= t1.id + @rows

    GROUP BY t1.did, t1.id

    /*cleanup*/

    DROP TABLE #test

    But if you're going to work with large amounts of data, solution using "Tally" table might be more effective. My solution uses triangular self-join, which can sometimes be pretty nasty.

  • I see that Christopher already posted the solution with Tally table, so just try to tweak the parameters there to suit your needs...

    BTW, there seems to be some problem with code display, all "INT" in declaration of parameters are doubled.

  • @christopher

    Yes your solution works but as you said, @startposition is not going to play any role in it.

    And as long as different @startpositions are concerned, it is very tricky scenerio [Smile]

    if @startposition is 2 and @endposition is 3 then

    DID-------ID--------Value----------Output

    1---------1-----------2------------12+45

    1---------2-----------12-----------45+8

    1---------3-----------45-----------8+63

    1---------4-----------8------------63+24

    1---------5-----------63-----------24

    1---------6-----------24-----------Null

    2---------1-----------89-----------20+46

    2---------2-----------20-----------46+96

    2---------3-----------46-----------96

    2---------4-----------96-----------Null

    Please ignore my previous reply where i said that if @startposition is 2 then first row will be ignored all together.

    If you still have any question please let me know.

  • @christopher

    Yes your solution works but as you said, @startposition is not going to play any role in it.

    And as long as different @startpositions are concerned, it is very tricky scenerio 🙂

    if @startposition is 2 and @endposition is 3 then

    DID-------ID--------Value----------Output

    1---------1-----------2------------12+45

    1---------2-----------12-----------45+8

    1---------3-----------45-----------8+63

    1---------4-----------8------------63+24

    1---------5-----------63-----------24

    1---------6-----------24-----------Null

    2---------1-----------89-----------20+46

    2---------2-----------20-----------46+96

    2---------3-----------46-----------96

    2---------4-----------96-----------Null

    Please ignore my previous reply where i said that if @startposition is 2 then first row will be ignored all together.

    If you still have any question please let me know.

  • You can do it without a number table though I'd love to know what the @startPosition variable does...

    DECLARE @Table TABLE

    ([DID] INT,

    [ID] INT ,

    [Value] INT)

    DECLARE @startposition INT

    DECLARE @endposition INT

    SELECT

    @startposition = 1,

    @endposition = 3

    INSERT INTO @Table

    SELECT 1,1,2 UNION ALL

    SELECT 1,2,12 UNION ALL

    SELECT 1,3,45 UNION ALL

    SELECT 1,4,8 UNION ALL

    SELECT 1,5,63 UNION ALL

    SELECT 1,6,24 UNION ALL

    SELECT 2,1,89 UNION ALL

    SELECT 2,2,20 UNION ALL

    SELECT 2,3,46 UNION ALL

    SELECT 2,4,96

    SELECT t1.DID, t1.ID, min(t1.value) as value, sum(t2.Value) as output FROM @Table t1

    inner join @Table t2 on t1.DID=T2.did and t2.id between t1.id and T1.id+@endposition-1

    group by t1.DID, t1.id

    DID ID value output

    ----------- ----------- ----------- -----------

    1 1 2 59

    1 2 12 65

    1 3 45 116

    1 4 8 95

    1 5 63 87

    1 6 24 24

    2 1 89 155

    2 2 20 162

    2 3 46 142

    2 4 96 96

    (10 row(s) affected)

    (Later: Oops- Vladen beat me to it!)

    Best wishes,
    Phil Factor

Viewing 15 posts - 1 through 15 (of 17 total)

You must be logged in to reply to this topic. Login to reply