我有一个像Id,DateTime,Device,Value这样的表。 我需要找到Value> gt = = 5的所有句点,并返回此期间的From和To dateTime。 设备可以是温度传感器,我需要一个温度超过5的From / To列表。
如果我们认为它是一个温度传感器,它每5分钟记录一次。表格将是这样的:
Id DateTime Device Value
--------------------------------
1 2015.09.01 09:10 T1 3,2
2 2015.09.01 09:15 T1 5,2
3 2015.09.01 09:20 T1 6,2
4 2015.09.01 09:25 T1 5,8
5 2015.09.01 09:30 T1 3,2
6 2015.09.01 09:35 T1 1,2
7 2015.09.01 09:40 T1 5,6
8 2015.09.01 09:45 T1 6,1
9 2015.09.01 09:50 T1 5,0
10 2015.09.01 09:55 T1 2,0
我试图找到的时间是从09:15到09:25以及09:40到09:50。
在SQL Server 2014中是否有这样做的智能方法?
答案 0 :(得分:2)
这是获得所需结果的一种方式:
SELECT MIN([DateTime]) AS StartPeriod, MAX([DateTime]) AS EndPeriod,
Device
FROM (
SELECT Id, [DateTime], Device, Value,
ROW_NUMBER() OVER (PARTITION BY Device ORDER BY Id) -
COUNT(CASE WHEN Value >= 5 THEN 1 END)
OVER (PARTITION BY Device ORDER BY Id) AS grp
FROM mytable ) AS t
WHERE t.Value >= 5
GROUP BY Device, grp
如果您希望每天获得PARTITION BY Device, [DateTime]
- Start
个字段,则可能需要End
。
答案 1 :(得分:0)
是的,有一种聪明的方法可以做到这一点。
看看'PARTITION BY'和'OVER'条款。
使用这些功能我相信你应该能够对句点中的数据进行分区(分组)并获得每个组的DateTime列的MIN和MAX值。
答案 2 :(得分:0)
这基本上是answered before,诀窍在于你已经连续和连续地订购了id
列。如果那不存在,你需要做更多的前期工作,以使减法正常工作。
select * into #tempTable from (
select Id = 1,[DateTime] = cast('2015.09.01 09:10' as datetime), device = 'T1', value = 3.2
union all select 2,cast('2015.09.01 09:15' as datetime),'T1',5.2
union all select 3,cast('2015.09.01 09:20' as datetime),'T1',6.2
union all select 4,cast('2015.09.01 09:25' as datetime),'T1',5.8
union all select 5,cast('2015.09.01 09:30' as datetime),'T1',3.2
union all select 6,cast('2015.09.01 09:35' as datetime),'T1',1.2
union all select 7,cast('2015.09.01 09:40' as datetime),'T1',5.6
union all select 8,cast('2015.09.01 09:45' as datetime),'T1',6.1
union all select 9,cast('2015.09.01 09:50' as datetime),'T1',5
union all select 10,cast('2015.09.01 09:55' as datetime),'T1',2) A
with T as (select *, row_number() over (order by id) - id as grp
from #tempTable
where value >=5
)
select [to] = min(T.Datetime), [from] = max(T.DateTime)
from T
group by T.grp
order by T.grp
输出为
to from
2015-09-01 09:40:00.000 2015-09-01 09:50:00.000
2015-09-01 09:15:00.000 2015-09-01 09:25:00.000
(我必须做一些演员才能让你的数字/日期格式适合我)
答案 3 :(得分:0)
我一直在使用一个表User Defined Function,它在请求中占用请求的日期,并且函数内的游标允许填充返回的表,如下所示:
CREATE TABLE sensor (id int not null identity(1,1) primary key,
measureDate datetime, sensor nvarchar(10), measure float)
INSERT sensor (measureDate, sensor, measure) VALUES
('2015-09-01 09:10', 'T1', '3.2'), ('2015-09-01 09:15', 'T1', '5.2'),
('2015-09-01 09:20', 'T1', '6.2'), ('2015-09-01 09:25', 'T1', '5.8'),
('2015-09-01 09:30', 'T1', '3.2'), ('2015-09-01 09:35', 'T1', '1.2'),
('2015-09-01 09:40', 'T1', '5.6'), ('2015-09-01 09:45', 'T1', '6.1'),
('2015-09-01 09:50', 'T1', '5.0'), ('2015-09-01 09:55', 'T1', '2.0')
GO
CREATE FUNCTION [dbo].[getTimeSpansBelowMaxTemp] (@measureDate date)
RETURNS @timeSpans TABLE (fromTime time, toTime time) AS BEGIN
DECLARE @measure float, @currentMeasure float = NULL
DECLARE @measureTime time, @fromMeasureTime time, @toMeasureTime time
DECLARE yourCursor CURSOR FOR
SELECT CAST(measureDate AS time), measure
FROM sensor
WHERE CAST(measureDate as date) = @measureDate
OPEN yourCursor
FETCH NEXT FROM yourCursor INTO @measureTime, @measure
WHILE (@@FETCH_STATUS = 0) BEGIN -- Loops on all the measures of the given day
IF @measure >= 5.0 BEGIN
IF @currentMeasure IS NULL BEGIN -- Start of a period
SET @currentMeasure = @measure
SET @fromMeasureTime = @measureTime
END
SET @toMeasureTime = @measureTime
END
ELSE BEGIN
IF @currentMeasure IS NOT NULL BEGIN -- End of a period
INSERT INTO @timeSpans VALUES (@fromMeasureTime, @toMeasureTime)
SET @currentMeasure = NULL
END
END
FETCH NEXT FROM yourCursor INTO @measureTime, @measure
END
CLOSE yourCursor
DEALLOCATE yourCursor
RETURN
END
GO
select * from dbo.[getTimeSpansBelowMaxTemp]('2015-09-01')