选择实时值的时间间隔 - 缺少第一个和最后一个间隔

时间:2011-03-09 08:20:46

标签: sql tsql sql-server-2008

我有一张包含以下结构的表格

| ChangedDate | IDParameter | ChangedTo(位列)|

所以我需要在参数为True或False时获取时间间隔,如下面的

| IDParameter | ChangedToDate1 | ChangedToDate2 | ChangedTo(true to false || false to true)

我做了

With RankedDates As
(
    Select T1.[ChangedDate], T1.ID, T1.[ChangedToValue]
    , Row_Number() Over( Partition By T1.ID Order By T1.[ChangedDate] ) As Num
    From [Changes] As T1
 )

  SELECT T1.[ID]
      ,T2.[ChangedToValue]
      ,T1.[ChangedDate] AS startDate
      ,T2.[ChangedDate] AS endDate
  FROM [RankedDates] AS T1
    Join RankedDates As T2
                    On T2.ID = T1.ID
                          And T2.Num = T1.Num + 1
                          And T2.[ChangedToValue] <> T1.[ChangedToValue]
                    Order By T2.[ChangedDate]

问题是我在这里缺少第一个和最后一个间隔。对于开始日期,它必须为NULL,如果是第一个,则为每个参数ID的最后一个间隔的endDate为NULL。我想我需要将它添加到UNION但我的麻烦我无法理解如何为每个IDParameter添加它

我不知道什么时候第一次更改了值,我不知道该值是否会在任何时候更改,所以我需要NULL或者为第一个间隔需要一些注意事项,对于最后一个间隔需要NULL或者一些maxdate。

ms sql server 2008

抱歉这么复杂的问题。

示例:

08.03.2011 ID1 0 -> 1 
09.03.2011 ID1 1 -> 0
09.03.2011 ID2 0 -> 1 
10.03.2011 ID1 0 -> 1 
10.03.2011 ID2 1 -> 0

---&GT;

NULL , 08.03.2011 ID1 is 0
NULL , 09.03.2011 ID2 is 0
08.03.2011, 09.03.2011 ID1 is 1
09.03.2011, 10.03.2011 ID2 is 1
09.03.2011, 10.03.2011 ID1 is 0
10.03.2011, NULL ID1 is 1
10.03.2011, NULL ID2 is 0

2 个答案:

答案 0 :(得分:2)

如何使用FULL JOIN代替JOIN

它能解决你的问题吗?

修改

我认为这应该可以随心所欲。

select  isnull(T1.ID, T2.ID) as ID
        ,isnull(T2.[ChangedToValue], case when T1.[ChangedToValue] = 1 then 0 else 1 end) as [ChangedToValue]
        ,T1.[ChangedDate] as startdate
        ,T2.[ChangedDate] as enddate
from    [RankedDates] T1
full join [RankedDates] T2
    on T2.num = T1.num +1
    and T2.ID = T1.ID
    and T1.[ChangedToValue] <> T2.[ChangedToValue]

order by 
    case when T2.[ChangedDate] is null then 1 else 0 end
    ,T2.[ChangedDate]

关于ChangedToValue你是对的,如果T2为空,我现在修改它以显示相反的情况。

答案 1 :(得分:1)

假设你的基表看起来如何:
ChangeDate IDParameter ChangedTo 2011-03-08 ID1 True
2011-03-09 ID1错误
2011-03-09 ID2真实 2011-03-10 ID1 True
2011-03-10 ID2错误

SELECT  (SELECT TOP 1 t0.[ChangeDate] FROM [calendardb].[dbo].[Table_1] t0 
WHERE t0.IDParameter = t1.IDParameter AND t0.ChangeDate < t1.ChangeDate ORDER 
BY t0.ChangeDate DESC),
   [ChangeDate]
  ,[IDParameter]
  ,[ChangedTo]
FROM [calendardb].[dbo].[Table_1] t1
UNION 
SELECT MAX(ChangeDate) as maxd ,NULL,[IDParameter],
(SELECT ChangedTo FROM [calendardb].[dbo].[Table_1] t0 WHERE t0.ChangeDate =    (SELECT MAX(ChangeDate) FROM [calendardb].[dbo].[Table_1]
GROUP BY [IDParameter] HAVING IDParameter = t1.IDParameter) AND t1.IDParameter = t0.IDParameter)

FROM [calendardb].[dbo].[Table_1] t1
GROUP BY [IDParameter]

会给你这样的结果:

NULL 2011-03-08 ID1 1
2011-03-08 2011-03-09 ID1 0
NULL 2011-03-09 ID2 1
2011-03-09 2011-03-10 ID1 1
2011-03-09 2011-03-10 ID2 0
2011-03-10 NULL ID1 1
2011-03-10 NULL ID2 0