按[Status_DateTime] DESC排序后，我尝试按[Status]进行分区。 因此，只有连续的[状态]才会被分区，以便获得最早的日期。
/****** Object: Table [dbo].[Order_Status] Script Date: 1/13/2017 1:06:38 PM ******/
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE TABLE [dbo].[Order_Status](
[Case_No_ID] [int] NULL,
[Status] [nvarchar](60) NULL,
[Status_Datetime] [datetime] NULL,
[Employee_ID] [int] NULL
) ON [PRIMARY]
GO
INSERT [dbo].[Order_Status] ([Case_No_ID], [Status], [Status_Datetime], [Employee_ID]) VALUES
(223079, N'HP - HOLD FOR PARTS', CAST(N'2015-12-21T16:44:50.000' AS DateTime), 740)
,(223079, N'DP - ORDER CREATED', CAST(N'2015-12-21T16:45:11.000' AS DateTime), 740)
,(223079, N'PP - PRINT PL', CAST(N'2015-12-21T16:47:03.000' AS DateTime), 709)
,(223079, N'PS- PARTS SHIPPED', CAST(N'2015-12-21T17:50:39.000' AS DateTime), 824)
,(223079, N'OS - MACHINE RECEIVED', CAST(N'2015-12-29T09:47:34.000' AS DateTime), 2)
,(223079, N'RP - UNDER INSPECTION', CAST(N'2015-12-30T12:14:07.000' AS DateTime), 168)
,(223079, N'RP - UNDER INSPECTION', CAST(N'2015-12-30T13:41:01.000' AS DateTime), 168)
,(223079, N'RP - UNDER INSPECTION', CAST(N'2016-01-05T14:15:46.000' AS DateTime), 241)
,(223079, N'RP - UNDER INSPECTION', CAST(N'2016-01-06T13:44:18.000' AS DateTime), 13)
,(223079, N'HP - HOLD FOR PARTS', CAST(N'2016-01-07T14:18:59.000' AS DateTime), 710)
,(223079, N'RELEASED FROM HOLD FOR PART', CAST(N'2016-01-11T16:37:49.000' AS DateTime), 710)
,(223079, N'RP - UNDER INSPECTION', CAST(N'2016-01-12T08:58:05.000' AS DateTime), 241)
,(223079, N'HP - HOLD FOR PARTS', CAST(N'2016-01-20T09:50:19.000' AS DateTime), 710)
,(223079, N'HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI', CAST(N'2016-01-25T15:32:19.000' AS DateTime), 755)
,(223079, N'HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI', CAST(N'2016-03-04T15:56:52.000' AS DateTime), 755)
,(223079, N'HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI', CAST(N'2016-05-16T14:57:12.000' AS DateTime), 755)
,(223079, N'HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI', CAST(N'2016-06-14T15:35:14.000' AS DateTime), 755)
,(223079, N'HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI', CAST(N'2016-08-15T15:28:32.000' AS DateTime), 755)
GO
更新： 如何在分区BY之前使用Order BY？ 如果某个状态在连续2行中重复，我正在尝试获取第一个日期时间，如果状态未发生变化但Status_DateTime已更改。
例如：
Case_No_ID Status Status_DateTime First_Time Employee_ID
----------- ---------------------------------------------------------------------------------------------------- ----------------------- ----------------------- -----------
223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-08-15 15:28:32.000 2016-01-25 15:32:19.000 755
223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-06-14 15:35:14.000 2016-01-25 15:32:19.000 755
223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-05-16 14:57:12.000 2016-01-25 15:32:19.000 755
223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-03-04 15:56:52.000 2016-01-25 15:32:19.000 755
223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-01-25 15:32:19.000 2016-01-25 15:32:19.000 755
不影响那些不在一行中的那些：＆＃39; HP - HOLD FOR PARTS＆＃39;以下
Case_No_ID Status Status_DateTime First_Time Employee_ID
----------- ---------------------------------------------------------------------------------------------------- ----------------------- ----------------------- -----------
223079 HP - HOLD FOR PARTS 2016-01-20 09:50:19.000 2016-01-20 09:50:19.000 710
223079 RP - UNDER INSPECTION 2016-01-12 08:58:05.000 2016-01-12 08:58:05.000 241
223079 RELEASED FROM HOLD FOR PART 2016-01-11 16:37:49.000 2016-01-11 16:37:49.000 710
223079 HP - HOLD FOR PARTS 2016-01-07 14:18:59.000 2016-01-07 14:18:59.000 710
223079 RP - UNDER INSPECTION 2016-01-06 13:44:18.000 2016-01-06 13:44:18.000 13
我尝试了以下内容：
Select
ROW_NUMBER() Over (order by status_Datetime Desc) rownum
, Case_No_ID
, Status
, Status_DateTime
, First_value(Status_DateTime) over (partition by [status] Order by [status] Desc) First_Time
, Employee_ID
from Order_Status
where Case_No_ID = 223079
order by Status_DateTime desc
如果状态更改是连续的顺序（第1行到第5行），它工作正常第一次进入状态是＆＃39; 2016-01-25 15：32：19.000＆＃39;并且由于状态没有改变，但如果你在第6行检查First_Time列，它会显示第9行的Status_DateTime，但它们之间的状态发生了变化。
rownum Case_No_ID Status Status_DateTime First_Time Employee_ID
-------------------- ----------- ---------------------------------------------------------------------------------------------------- ----------------------- ----------------------- -----------
1 223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-08-15 15:28:32.000 2016-08-15 15:28:32.000 755
2 223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-06-14 15:35:14.000 2016-08-15 15:28:32.000 755
3 223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-05-16 14:57:12.000 2016-08-15 15:28:32.000 755
4 223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-03-04 15:56:52.000 2016-08-15 15:28:32.000 755
5 223079 HP06 - PARTS SHORTAGE, PLANNER CONFIRM ETA AND TRIGGER EDI 2016-01-25 15:32:19.000 2016-08-15 15:28:32.000 755
6 223079 HP - HOLD FOR PARTS 2016-01-20 09:50:19.000 2015-12-21 16:44:50.000 710
7 223079 RP - UNDER INSPECTION 2016-01-12 08:58:05.000 2016-01-12 08:58:05.000 241
8 223079 RELEASED FROM HOLD FOR PART 2016-01-11 16:37:49.000 2016-01-11 16:37:49.000 710
9 223079 HP - HOLD FOR PARTS 2016-01-07 14:18:59.000 2015-12-21 16:44:50.000 710
10 223079 RP - UNDER INSPECTION 2016-01-06 13:44:18.000 2016-01-12 08:58:05.000 13
11 223079 RP - UNDER INSPECTION 2016-01-05 14:15:46.000 2016-01-12 08:58:05.000 241
12 223079 RP - UNDER INSPECTION 2015-12-30 13:41:01.000 2016-01-12 08:58:05.000 168
13 223079 RP - UNDER INSPECTION 2015-12-30 12:14:07.000 2016-01-12 08:58:05.000 168
14 223079 OS - MACHINE RECEIVED 2015-12-29 09:47:34.000 2015-12-29 09:47:34.000 2
15 223079 PS- PARTS SHIPPED 2015-12-21 17:50:39.000 2015-12-21 17:50:39.000 824
16 223079 PP - PRINT PL 2015-12-21 16:47:03.000 2015-12-21 16:47:03.000 709
17 223079 DP - ORDER CREATED 2015-12-21 16:45:11.000 2015-12-21 16:45:11.000 740
18 223079 HP - HOLD FOR PARTS 2015-12-21 16:44:50.000 2015-12-21 16:44:50.000 740
感谢您的帮助
答案 0 :(得分：1)
一种方法只是使用lag()查看以前的状态，并仅保留已更改的状态：
select os.*
from (select os.*,
lag(status) over (partition by Case_No_ID order by Status_Datetime) as prev_status
from Order_Status os
) os
where prev_status is null or prev_status <> status;
答案 1 :(得分：1)
如果我理解正确，你需要这样的东西。这使用运行总延迟来将连续状态分组到同一组中，然后使用first_value分区：
select case_no_id, status, status_datetime, first_value(status_datetime) over (partition by GRP order by status_datetime desc), employee_id
from (
select sum(case when Status = prev_status then 0 else 1 end) over (order by Status_datetime desc) as GRP, *
from (
select *, lag(status) over (order by Status_Datetime desc) as prev_status from Order_Status
) X
) Y
order by Status_Datetime desc