Rank() Over with multiple Partition

Question

我正在尝试查找票证更改所有者组的次数。我正在使用这个查询：

SELECT        
TICKETID, CHANGEDATE, OWNERGROUP, 
RANK() OVER (Partition by TICKETID, OWNERGROUP ORDER BY CHANGEDATE) rk
FROM TKSTATUS where TICKETID= '12345' order by CHANGEDATE

并得到了这个结果集： enter image description here

数完 1... 得到 4。

我期待这个结果，但找不到我该怎么做：

结果为 6。

我知道这不是办法，因为它将 TICKETID 和 OWNERGROUP 分组在排名之前。

我会继续寻找一种方法来做到这一点，但我想看看是否有人有一些想法。

Answer 1

    SELECT        
    TICKETID, CHANGEDATE, OWNERGROUP, 
    DENSE_RANK() OVER (Partition by TICKETID ORDER BY CHANGEDATE) as RK
    FROM TKSTATUS where TICKETID= '12345' order by CHANGEDATE

Answer 2

在 SQL Server 2012 中，您可以使用 lag() 通过比较当前行值与 OwnerGroup 的前一行值来查看 OwnerGroup 更改的数量。

select 
    TicketId
  , ChangeDate
  , OwnerGroup
  , OwnerGroupChange = case 
      when coalesce(OwnerGroup,'') 
          = lag(coalesce(OwnerGroup,'')) over (partition by TicketId order by ChangeDate)
        then 0 
    else 1 
    end
from TkStatus
where TicketId = '12345'
order by ChangeDate

要获取更改的数量，您可以将上面的内容用作子查询或通用 table 表达式：

select 
    TicketId
  , OwnerGroupChange = sum(OwnerGroupChange) 
from (
  select 
      TicketId
    , ChangeDate
    , OwnerGroup
    , OwnerGroupChange = case 
        when coalesce(OwnerGroup,'') 
            = lag(coalesce(OwnerGroup,'')) over (partition by TicketId order by ChangeDate)
          then 0 
      else 1 
      end
  from TkStatus
  where TicketId = '12345'
  ) s

在 SQL Server 2012 之前，您可以使用 outer apply() 获取之前的 OwnerGroup，如下所示：

select 
    t.TicketId
  , t.ChangeDate
  , t.OwnerGroup
  , OwnerGroupChange = case 
      when coalesce(t.OwnerGroup,'') = coalesce(x.OwnerGroup,'')
        then 0 
    else 1 
    end
from TkStatus t
  outer apply (
    select top 1
      i.OwnerGroup
    from TkStatus i
    where i.TicketId = t.TicketId
      and i.ChangeDate < t.ChangeDate
    order by i.ChangeDate desc
    ) x
where t.TicketId = '12345'
order by t.ChangeDate

Answer 3

您可以像这样使用 CTE 和 LAG 函数：

WITH CTE_Tickets as 
(  SELECT        
        TICKETID, CHANGEDATE, OWNERGROUP, 
        LAG(OWNERGROUP, 1) OVER (Partition by OWNERGROUP ORDER BY CHANGEDATE, OWNERGROUP ) as Prev
        FROM TKSTATUS where TICKETID= '12345')


    SELECT        
       TICKETID, CHANGEDATE, OWNERGROUP, Prev ,
       IIF(OWNERGROUP=Prev, 0, 1)  as changestatus
       FROM CTE_Tickets
       ORDER BY CHANGEDATE  , OWNERGROUP

Rank() Over with multiple Partition

Rank() Over with multiple Partition

sql

tsql

rank

partition