我正在尝试编写一个查询包含id，date和value的数据点的查询。这些数据点代表员工的休假时间。
以下是一些示例数据点（请假时间）的示例：
+------------+-------+-------+-------------+
| Date | Value | Total | Status |
+------------+-------+-------+-------------+
| 8/20/2015 | 0.25 | 0.25 | |
| 1/11/2016 | 1 | 1.25 | |
| 5/27/2016 | 0.5 | 1.75 | |
| 8/5/2016 | 0.25 | 2 | |
| 9/13/2016 | 0 | 1.75 | |
| 9/28/2016 | 0.25 | 2 | |
| 9/29/2016 | 0.25 | 2.25 | |
| 10/17/2016 | 0 | 2.25 | |
| 10/17/2016 | 1 | 3.25 | Verbal |
| 11/17/2016 | 0.25 | 3.5 | |
| 12/1/2016 | 1 | 4.5 | Written |
| 12/2/2016 | 1 | 5.5 | Final |
| 12/6/2016 | 0.25 | 5.75 | |
| 12/14/2016 | 0.25 | 6 | Termination |
| 12/19/2016 | 0.25 | 6.25 | |
| 1/6/2017 | 0.25 | 6.5 | |
| 2/15/2017 | 0.25 | 5.75 | Final |
| 3/15/2017 | 0.25 | 6 | Termination |
| 4/17/2017 | 0.25 | 6.25 | |
| 5/8/2017 | 0 | 6.25 | |
| 5/9/2017 | 1 | 7.25 | |
| 5/21/2017 | 0.25 | 7.5 | |
| | | | |
+------------+-------+-------+-------------+
我想要一个计算上述结果的查询。总列由当前数据点日期的1年内先前数据点的总和计算。因此，如果当前数据点的日期为2016年9月29日，我想将所有之前的数据点加在2015年9月29日至2016年9月29日之间。
此查询的最终目标是根据他们的休假时间（数据点）查看一个人在去年内是否有两个最终警告。根据我提供的示例，我们在2016年12月12日和2017年2月15日发布了最终警告，这些警告发生在去年（2016年8月21日 - 2017年8月21日）。
这样做的最佳方式是什么？
以下是我目前的情况：
SELECT
a.dtmDATEAPP AS [Date],
MAX(CASE WHEN a.dtmDATEAPP = b.dtmDATEAPP THEN a.strOCCUR END) AS [Value],
SUM(b.strOCCUR) AS Total,
CASE
WHEN SUM(b.strOCCUR) >= 5 AND SUM(b.strOCCUR) < 6
THEN 'Final'
END AS [Status]
FROM
[Attendance].[dbo].[TIMEUSE] a
JOIN
[Attendance].[dbo].[TIMEUSE] b ON b.dtmDATEAPP BETWEEN DATEADD(year, -1, a.dtmDATEAPP) AND a.dtmDATEAPP
WHERE
a.lngEMPID = 1162002
AND a.Absence <> 'Scheduled'
AND b.lngEMPID = 1162002
AND b.Absence <> 'Scheduled'
GROUP BY
a.dtmDATEAPP
ORDER BY
a.dtmDATEAPP
我不确定在过去一年中员工是否有2个最终警告的最佳途径。
非常感谢所有帮助。
答案 0 :(得分：0)
这样的事可能会对你有所帮助：
declare @TimeUsage table
(
missingDate date,
absence varchar(20),
pctDay decimal(6,2),
status varchar(20),
employeeid int
)
INSERT INTO @TimeUsage VALUES
('8/20/2015', 'Unscheduled', 0.25, '', 1162002),
('1/11/2016', 'Unscheduled', 1, '', 1162002),
('5/27/2016', 'Unscheduled', 0.5, '', 1162002),
('8/5/2016', 'Unscheduled', 0.25, '', 1162002),
('9/13/2016', 'Unscheduled', 0, '', 1162002),
('9/28/2016', 'Unscheduled', 0.25, '', 1162002),
('9/29/2016', 'Unscheduled', 0.25, '', 1162002),
('10/17/2016', 'Unscheduled', 0, '', 1162002),
('10/17/2016', 'Unscheduled', 1, 'Verbal', 1162002),
('11/17/2016', 'Unscheduled', 0.25, '', 1162002),
('12/1/2016', 'Unscheduled', 1, 'Written', 1162002),
('12/2/2016', 'Unscheduled', 1, 'Final', 1162002),
('12/6/2016', 'Unscheduled', 0.25, '', 1162002),
('12/14/2016', 'Unscheduled', 0.25, 'Termination', 1162002),
('12/19/2016', 'Unscheduled', 0.25, '', 1162002),
('1/6/2017', 'Unscheduled', 0.25, '', 1162002),
('2/15/2017', 'Unscheduled', 0.25, 'Final', 1162002),
('3/15/2017', 'Unscheduled', 0.25, 'Termination', 1162002),
('4/17/2017', 'Unscheduled', 0.25, '', 1162002),
('5/8/2017', 'Unscheduled', 0, '', 1162002),
('5/9/2017', 'Unscheduled', 1, '', 1162002),
('5/21/2017', 'Unscheduled', 0.25, '', 1162002)
;WITH cte AS
(SELECT missingDate, pctDay,
(SELECT SUM(pctDay) FROM @TimeUsage
WHERE missingDate <= d.missingDate
AND missingDate >= DATEADD(yy,-1,d.missingDate)
AND employeeid = d.employeeid
AND absence <> 'Scheduled') as prevYrTot,
(SELECT COUNT(*) FROM @TimeUsage
WHERE missingDate <= d.missingDate
AND missingDate >= DATEADD(yy,-1,d.missingDate)
AND employeeid = d.employeeid
AND status = 'Final') AS FinalCount,
status,employeeid
from @TimeUsage d)
SELECT * FROM cte where employeeid IN
(SELECT DISTINCT(employeeid) FROM cte WHERE FinalCount > 1)
修改
如果最终状态由prevyrtot决定，那么试试这个：
declare @TimeUsage table
(
missingDate date,
absence varchar(20),
pctDay decimal(6,2),
employeeid int
)
INSERT INTO @TimeUsage VALUES
('8/20/2015', 'Unscheduled', 0.25, 1162002),
('1/11/2016', 'Unscheduled', 1, 1162002),
('5/27/2016', 'Unscheduled', 0.5, 1162002),
('8/5/2016', 'Unscheduled', 0.25, 1162002),
('9/13/2016', 'Unscheduled', 0, 1162002),
('9/28/2016', 'Unscheduled', 0.25, 1162002),
('9/29/2016', 'Unscheduled', 0.25, 1162002),
('10/17/2016', 'Unscheduled', 0, 1162002),
('10/17/2016', 'Unscheduled', 1, 1162002),
('11/17/2016', 'Unscheduled', 0.25, 1162002),
('12/1/2016', 'Unscheduled', 1, 1162002),
('12/2/2016', 'Unscheduled', 1, 1162002),
('12/6/2016', 'Unscheduled', 0.25, 1162002),
('12/14/2016', 'Unscheduled', 0.25, 1162002),
('12/19/2016', 'Unscheduled', 0.25, 1162002),
('1/6/2017', 'Unscheduled', 0.25, 1162002),
('2/15/2017', 'Unscheduled', 0.25, 1162002),
('3/15/2017', 'Unscheduled', 0.25, 1162002),
('4/17/2017', 'Unscheduled', 0.25, 1162002),
('5/8/2017', 'Unscheduled', 0, 1162002),
('5/9/2017', 'Unscheduled', 1, 1162002),
('5/21/2017', 'Unscheduled', 0.25, 1162002)
;WITH cte as
(SELECT missingDate, pctDay, prevYrTot, employeeid,
CASE WHEN prevYrTot >= 5 AND prevYrTot < 6
THEN 'Final' ELSE '' END AS Status FROM
(SELECT missingDate, pctDay,
(SELECT SUM(pctDay) FROM @TimeUsage
WHERE missingDate <= d.missingDate
AND missingDate >= DATEADD(yy,-1,d.missingDate)
AND employeeid = d.employeeid
AND absence <> 'Scheduled') as prevYrTot,
employeeid
from @TimeUsage d) tu)
SELECT missingDate, pctDay, prevYrTot, employeeid, Status FROM cte
WHERE employeeid IN
(SELECT DISTINCT employeeid FROM (SELECT employeeid, COUNT(*) as FinalCount FROM cte
WHERE Status = 'Final'
GROUP BY employeeid
HAVING Count(*) > 1) de)
我认为这样做你想要的，但你需要测试！它适用于给定的数据。