PIVOT query on Distinct records

后端 未结 1 1659
执念已碎
执念已碎 2020-11-27 08:27

I have below table:

------------------------------------------------------
| Id    Code  percentage  name  name1   activity     |
---------------------------         


        
相关标签:
1条回答
  • 2020-11-27 08:51

    If you add a ROW_NUMBER() into the mix, your pivot will be able to retain the association between activities and percentages.

    ;with cte as 
    (
        select *, ROW_NUMBER() over (partition by name order by percentage desc) ROWNUM
        from A
    ),
    cte2 as
    (
        SELECT Id,Code,ROWNUM,James,James_,Sam,Sam_,Lisa,Lisa_
        FROM cte
        PIVOT(MAX(activity)
              FOR name IN (James,Sam,Lisa)) AS PVTTable PIVOT
              (
              MAX(percentage)
              FOR name1 IN (James_,Sam_,Lisa_)) AS PVTTable1
    )
    select Id, Code, MAX(James) James, MAX(James_) James_, MAX(Sam) Sam, MAX(Sam_) Sam_, MAX(Lisa) Lisa, MAX(Lisa_) Lisa_
    from cte2
    group by Id, Code, ROWNUM
    

    Returns:

    Id  Code        James       James_  Sam         Sam_    Lisa    Lisa_
    1   Prashant    Running     43.43   Cooking 1   73      Walking 90.34
    1   Prashant    Stealing    0.00    Cooking     3.43    NULL    NULL
    1   Prashant    Lacking     0.00    NULL        NULL    NULL    NULL
    

    The idea is, in thefirst common table expression, you transform the A table to this:

    Id  Code        percentage  name    name1   activity    ROWNUM
    1   Prashant    43.43       James   James_  Running     1
    1   Prashant    0.00        James   James_  Stealing    2
    1   Prashant    0.00        James   James_  Lacking     3
    1   Prashant    90.34       Lisa    Lisa_   Walking     1
    1   Prashant    73          Sam     Sam_    Cooking 1   1
    1   Prashant    3.43        Sam     Sam_    Cooking     2
    

    And throughout the remaining query, the ROWNUM column just acts to bind the percentage value to the activity.

    Making it dynamic is easy once you have a working query. Just replace all the dynamic parts (in this case, comma-delimited lists of names, right?) with variables. Something like this:

    declare @sql nvarchar(max)
    declare @name_concat nvarchar(max)
    declare @name1_concat nvarchar(max)
    declare @select_aggs nvarchar(max)
    select @name_concat = STUFF((select distinct ',' + quotename(name) from A order by 1 for xml path('')), 1, 1, '')
    select @name1_concat = STUFF((select distinct ',' + quotename(name1) from A order by 1 for xml path('')), 1, 1, '')
    
    ;with cte_all_names as (
        select name from A
        union all 
        select name1 from A
    )
    select @select_aggs = STUFF((select distinct ',MAX(' + quotename(name) + ') ' + quotename(name) from cte_all_names order by 1 for xml path('')), 1, 1, '')
    
    select @sql = '
    ;with cte as 
    (
        select *, ROW_NUMBER() over (partition by name order by percentage desc) ROWNUM
        from A
    ),
    cte2 as
    (
        SELECT Id,Code,ROWNUM,' + @name_concat + ',' + @name1_concat + '
        FROM cte
        PIVOT(MAX(activity)
              FOR name IN (' + @name_concat + ')) AS PVTTable PIVOT
              (
              MAX(percentage)
              FOR name1 IN (' + @name1_concat + ')) AS PVTTable1
    )
    select Id, Code, ' + @select_aggs + '
    from cte2
    group by Id, Code, ROWNUM
    '
    
    exec sp_executesql @sql
    
    0 讨论(0)
提交回复
热议问题