Sumof previous columns (in pivot table)

  • I need to determine the weekly stock amounts. I know the weekly inputs and outputs. I made a pivot table to get the products and weekly inputs/outputs. the pivot works well but at any week, the amount of stock is equal (weekly input - weekly output) + previous weeks inputs/outputs. inputs are (+), outputs are (-) so they are summable. hete's pivot code:

    DECLARE @columns VARCHAR(8000)

    SELECT @columns = COALESCE(@columns + ',[' + cast(WeekOffset as varchar) + ']',

    '[' + cast(WeekOffset as varchar)+ ']')

    FROM HaftalikStokRaporu_HaftalikStoklar_View

    GROUP BY WeekOffset

    order by Weekoffset

    --select @columns

    DECLARE @query VARCHAR(8000)

    SET @query = '

    SELECT urun, ' + @columns + '

    FROM

    (SELECT urun, Miktar, WeekOffset

    FROM HaftalikStokRaporu_HaftalikStoklar_View ) AS p

    PIVOT

    (

    SUM (Miktar)

    FOR WeekOffset IN

    ( ' + @columns +' )

    ) AS pvt

    WHERE urun like ''' + @UrunAdi + '''

    ORDER BY urun;

    '

    --EXECUTE(@query)

    print @query

    this produces following sample:

    urun 5217566356935694

    %5 Dekstroz 500ml3 7 2 5

    Abilify 10mg 167 75

    Abilify 15mg 307 3 5

    Abilify 30mg 307 3 5

    Accolate 20mg -617 45

    now I can explain simpler way:

    5693's total must be 7+3+2

    5694's total must be 7+3+5..... this goes on for every column

    ..... and for every row.

    in other words, SUM function in pivot code should be something like:

    SUMOFPREVIOSROWS(...)

    its a bit confusing but changing pivor code or more simpler, creating a view to chage column values to sum off all previous colums are welcome.

    But my friend, remember this: column number are varies. for all summing must be done with code. meaning SUM(column5217 + 5694 +....) is forbidden

  • Since you have no data that we can actually play with, all I can do is give you a suggestion... see the following article...

    http://www.sqlservercentral.com/articles/Crosstab/65048/

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • wow, quick answer. yes I have no data, sorry I know it would be easy. okay, let me put in that way. here's sample data:

    Product Col1 col2 col3 col4

    a 1 2 3 4

    b 2 3 4 5

    c 3 3 4 4

    that would be transformed to this:

    Product Col1 col2 col3 col4

    a 1 3 (1+2) 6 (3+3) 10 (6+4)

    b 2 5 (2+3) 9 (5+4) 14 (9+5)

    c 3 6 (3+3) 10 (6+4) 14 (10+4)

    each cell is sum of itself and left cell. I can take it go on from here. forget pivots, etc. I temporarly solved the problem by loading data to grid in client side (spread component, like excel) and added cells with code. but my friend, its slow as a hell. I rellay want to make this calculations in sql server side.

    thanks.

  • aykut canturk (8/28/2010)


    wow, quick answer. yes I have no data, sorry I know it would be easy. okay, let me put in that way. here's sample data:

    Product Col1 col2 col3 col4

    a 1 2 3 4

    b 2 3 4 5

    c 3 3 4 4

    that would be transformed to this:

    Product Col1 col2 col3 col4

    a 1 3 (1+2) 6 (3+3) 10 (6+4)

    b 2 5 (2+3) 9 (5+4) 14 (9+5)

    c 3 6 (3+3) 10 (6+4) 14 (10+4)

    each cell is sum of itself and left cell. I can take it go on from here. forget pivots, etc. I temporarly solved the problem by loading data to grid in client side (spread component, like excel) and added cells with code. but my friend, its slow as a hell. I rellay want to make this calculations in sql server side.

    thanks.

    What happened to the "previous row" requirement? That's the data I'm looking for. And please help me help you better... take a look at the article at the first link in my signature line below for how to post some good test data. Thanks.

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • no my friend, forget about everything I said before and just consider my last post. I managed my pivot code and my only concern is now "sum of itself and left cell". other rows and cells are irrelevant now.

    (and it not previous row, its previous cell)

  • Can't you just add up the columns then?

    SELECT

    Product,

    col1,

    col1 + col2,

    col1 + col2 + col3,

    col1 + col2 + col3 + col4

    FROM YourTransformedTable

  • yes actually. but column count and column names are changing all the time. so I beleive a kind of "for each column..." structure needed. and the the sql statement you wrote must be created by stored proc. that is what I cannot do because of knowledge lack.

  • aykut canturk (8/29/2010)


    yes actually. but column count and column names are changing all the time. so I beleive a kind of "for each column..." structure needed. and the the sql statement you wrote must be created by stored proc. that is what I cannot do because of knowledge lack.

    You keep changing and adding requirements. Let me know when you have a set of requirements properly hammered out.

    --Jeff Moden


    RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
    First step towards the paradigm shift of writing Set Based code:
    ________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

    Change is inevitable... Change for the better is not.


    Helpful Links:
    How to post code problems
    How to Post Performance Problems
    Create a Tally Function (fnTally)

  • I'm not changing requirements my friend. just clearifying the problem for you.

    any way, thanks for your intrerest.

  • If you haven't figured out how to do this. Here's the code --

    Use tempdb

    go

    -- create the above table

    IF EXISTS (SELECT name FROM sysobjects WHERE name = 'temp_model' AND type = 'U')

    DROP TABLE temp_model

    CREATE TABLE temp_model (

    category VARCHAR(100),

    value1 INT,

    value2 INT,

    value3 INT,

    value4 INT)

    go

    INSERT INTO temp_model

    SELECT 'a', 1, 2, 3, 4

    UNION ALL

    SELECT 'b', 2, 3, 4, 5

    UNION ALL

    SELECt 'c', 3, 3, 4, 4

    go

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

    -- 1. get the column numbers and column orders

    DECLARE @tableName VARCHAR(100),

    @sql VARCHAR(8000),

    @columnOrder INT, -- original table column order

    @NestedColumnOrder INT, -- nested column order for value increase from prior column

    @maxColumnOrder INT, -- original table max column numbers

    @maxNestedColumnOrder INT, -- nested column max column numbers for each new column, this is to take care of new value of each column

    @columnName VARCHAR(100), -- that is the column name in as statement, same as original table column

    @columnValue VARCHAR(100), -- that's the new_column name used in value increase

    @columnValue_prior VARCHAR(8000)

    -- that's the prior columns so we can do value1 + value2 + value3

    SET @tableName = 'temp_model'

    IF EXISTS(SELECT name FROM sysobjects WHERE name = 'temp_work1' AND type = 'U' )

    DROP TABLE temp_work1

    SELECT a.name, a.colorder

    INTO temp_work1

    FROM syscolumns a INNER JOIN sysindexes b ON

    a.id = b.id

    WHERE b.name = @tableName AND b.indid <2

    -- cross join to get the columns we need for the new table

    IF EXISTS(SELECT name FROM sysobjects WHERE name = 'temp_work2' AND type = 'U' )

    DROP TABLE temp_work2

    SELECT a.*, b.name as new_column, b.colorder as new_order

    INTO temp_work2

    FROM temp_work1 a INNER JOIN temp_work1 b ON

    a.colorder = b.colorder

    WHERE a.colorder = 1 -- take care of first column

    UNION ALL

    SELECT a.*, b.name as new_column, b.colorder as new_order

    FROM temp_work1 a INNER JOIN temp_work1 b

    ON b.colorder <= a.colorder and b.colorder <> 1 -- the first column don't need the replication

    ORDER BY a.colorder, b.colorder

    -- now will loop through to create the sql select statement looks like

    /*

    SELECT name,

    value1,

    value1 + value2 as value2,

    value1 + value2 + value3 as value3,

    value1 + value2 + value3 + value4 as value4

    into finaltable

    from temp_model

    */

    -- select * from temp_work2

    SET @sql = ''

    SET @columnOrder = 1

    SELECT @maxColumnOrder = MAX(colorder)

    FROM temp_work2

    WHILE @columnOrder <= @maxColumnOrder

    BEGIN

    SELECT @maxNestedColumnOrder = MAX(new_order),

    @NestedColumnOrder = MIN(new_order)

    FROM temp_work2

    WHERE colorder = @columnOrder

    SET @columnValue_prior = ''

    SELECT @columnName = name

    FROM temp_work2

    WHERE colorder = @columnOrder

    WHILE @NestedColumnOrder<= @maxNestedColumnOrder

    BEGIN

    SELECT @columnValue = new_column

    FROM temp_work2

    WHERE new_order = @NestedColumnOrder

    AND

    colorder = @columnOrder

    IF @NestedColumnOrder < @maxNestedColumnOrder

    SET @columnValue = @columnValue + '+ '

    SET @columnValue_prior = @columnValue_prior + @columnValue

    SET @NestedColumnOrder = @NestedColumnOrder + 1

    END

    IF @columnOrder = 1 -- start of statement

    SET @sql = 'SELECT ' + @columnValue_prior + ' AS ' + @columnName

    ELSE

    SET @sql = @sql + ', ' + @columnValue_prior + ' AS ' + @columnName

    SET @columnOrder = @columnOrder + 1

    END

    -- complete the sql

    IF EXISTS (SELECT name FROM sysobjects WHERE name = 'finalTable' AND type = 'U')

    DROP TABLE finalTable

    SET @sql = @sql + ' INTO finalTable FROM temp_model'

    EXEC (@sql)

    go

    SELECT *

    FROM finalTable

  • woow... thank you my friend. you saved my day. I could not imagine this.

    thank you for million times.

    have a nice day.

Viewing 11 posts - 1 through 10 (of 10 total)

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