SQL Server DATEDIFF的累積總和成百分比

Question

我正在嘗試從一些基本數據中將DATEDIFF的累積總和轉換為一定百分比，這是一個小快照：

ID     IIn                       IOut
AB123  2015-11-06 15:24:44.057   2015-11-14 01:00:00.000
QA565  2015-10-27 20:12:19.753   2015-11-06 03:00:00.000
UN555  2015-12-29 06:29:23.417   2016-01-03 08:00:00.000
LG602  2015-08-07 16:52:13.573   2015-08-11 03:00:00.000

ETC ETC

然后，我使用DATEDIFF獲得幾天的時間：

SELECT ID, DATEDIFF(hour, IIn, IOut)/24.0 IDays
FROM TimeTable

這給了我：

ID     IDays
AB123  7.416666
QA565  9.291666
UN555  5.083333
LG602  3.458333

我想要的是ID的計數IDay （向下舍入），從最低IDay到最高IDay ％的累積百分比，如下所示：

ID     IDays  IDaysPer
LG602  3      12.5
UN555  5      33.33
AB123  7      62.49
QA565  9      100

Answer 1

您可以使用幾個窗口聚合來完成此操作，為方便起見將原始查詢放在CTE中（子查詢也可以）：

declare @timeTable table (ID char(5) not null, IIn datetime not null,
                          IOut datetime not null)
insert into @timeTable(ID,IIn,IOut) values
('AB123','2015-11-06T15:24:44.057','2015-11-14T01:00:00.000'),
('QA565','2015-10-27T20:12:19.753','2015-11-06T03:00:00.000'),
('UN555','2015-12-29T06:29:23.417','2016-01-03T08:00:00.000'),
('LG602','2015-08-07T16:52:13.573','2015-08-11T03:00:00.000')

;With Diffs as (
    SELECT ID, DATEDIFF(hour, IIn, IOut)/24.0 IDays
    FROM @timeTable
)
select
    *,
    (
      SUM(IDays) OVER (ORDER BY IDays, ID)
      /
      SUM(IDays) OVER ()
    ) * 100 as IDaysPer
from
    Diffs
order by IDays

請注意，我不太了解您的“四舍五入”要求，但您應該能夠使用圍繞適當的計算方法包裝的任何通用四舍五入技術。 所以我的輸出與您的輸出不太匹配：

ID    IDays                                   IDaysPer
----- --------------------------------------- ---------------------------------------
LG602 3.458333                                13.696300
UN555 5.083333                                33.828300
AB123 7.416666                                63.201300
QA565 9.291666                                100.000000

Answer 2

在這里，您可以：輸出與您的匹配...

        create table #TEMp
(ID VARCHAR(100)
,IIn datetime
,IOut datetime
)


insert into #temp(ID,IIn,IOut) values
('AB123','2015-11-06T15:24:44.057','2015-11-14T01:00:00.000'),
('QA565','2015-10-27T20:12:19.753','2015-11-06T03:00:00.000'),
('UN555','2015-12-29T06:29:23.417','2016-01-03T08:00:00.000'),
('LG602','2015-08-07T16:52:13.573','2015-08-11T03:00:00.000')

select ID,IDays AS Idays,ROUND(CAST(SUM(IDays) OVER(ORDER BY IDays) AS FLOAT)/CAST(SUM(IDays)OVER() AS FLOAT) * 100,2) AS IdaysPer
from
(
select *,ROUND(DATEDIFF(hour, IIn, IOut)/24,0) IDays
from #TEMP
)T

Answer 3

考慮時間表已經有數據

WITH t1 (ID, IDays)
AS (
    SELECT ID, DATEDIFF(hour, IIn, IOut) / 24.0 AS IDays
    FROM TimeTable
)
SELECT 
    ID, FLOOR(IDays), 
    (FLOOR(IDays) / (SELECT SUM(FLOOR(IDays)) FROM t1 t2 WHERE t1.IDays <= t2.IDays)) * 100.0 AS IDaysPer
FROM t1
ORDER BY 2 ASC