【数据库】【算法】从一串数据中获取符合某个条件的时间区间

算法背景

在项目中,经常要看到这么一个要求:从一组根据时间排列的数据,要求计算出其中满足某个条件的一串数据,求这串数据的开始与结束时间.

比如说,用小米手环采集到一组用户的心率数据,要求算出其中有多长时间用户的心率>100.

比如说,通过Gps定位获取到车辆的行驶数据,按要求速度超过80就算超速,那么在这组数据中,车辆超速了多长时间.

在数据采集非常方便的时代,像以上的场景及需求比比皆是.

要算出所要求的结果,当然可以写个计算器,一条一条分析过去,但是这样的话效率非常慢.

更多时候还是要求在数据库中直接通过SQL算出结果。

因此我觉得这个算法非常有价值。

 假设具体场景

具体一点,比如下面这串数据:

按照下面的条件将以上数据进行分段切割,

Speed >= 0 且 Speed < 30

Speed >= 30 且 Speed < 60

Speed >= 60 且 Speed < 80

Speed >= 80

并且算出每段数据的开始时间与结束时间,开始SOC与结束SOC.

思路

1.先取到原始数据

;WITH temp AS (
	SELECT ROW_NUMBER() OVER (ORDER BY Car,Time ASC)AS ID,  Time,Car,SOC
		,SpeedFlg = CASE  WHEN  Speed >= 0 AND Speed < 30 THEN 1
						  WHEN  Speed >= 30 AND Speed < 60 THEN 2
						  WHEN  Speed >= 60 AND Speed < 80 THEN 3
						  WHEN  Speed >= 80 THEN 4
						  END
	FROM OrigData
	WHERE Car = ‘ABCDEFK10NZ000001‘
	AND Time >= ‘2016-06-01‘
	AND Time <= GETDATE()
)

先把上面的整串数据取到.

2.对数据进行排序,算出每个分段的开始于结束时间,并过滤掉分段中部的数据,保留分段的开始结束数据

final as(
	SELECT ROW_NUMBER()OVER(ORDER BY a.Car ) as tid,*
	FROM (
		SELECT a.Car,a.SpeedFlg,A.SOC
			,begintime = CASE WHEN A.ID = 1 OR b.SpeedFlg IS NULL THEN A.Time END
			,endtime = CASE WHEN c.SpeedFlg IS NULL THEN A.Time END
		FROM  temp a
		LEFT JOIN temp b ON a.ID = b.ID +1 AND B.SpeedFlg = A.SpeedFlg
		LEFT JOIN temp c ON a.ID = c.ID - 1  AND C.SpeedFlg = A.SpeedFlg
	)A
	where begintime is not null
	or endtime is not null
)

根据上面的算法,可以得出以下数据

3.将每个分段的开始结束数据进行合并,得到一个完整的数据

select a.Car
	,SpeedFlg = case a.SpeedFlg when 1 then ‘0-30‘
								when 2 then ‘30-60‘
								when 3 then ‘60-80‘
								when 4 then ‘80以上‘
								end
	,a.begintime,b.endtime,a.SOC as beginsoc ,b.SOC as endsoc
from final a
INNER JOIN final b on a.Car = b.Car and a.SpeedFlg = b.SpeedFlg and a.tid = b.tid - 1

最终得到下面这串数据

点开可查看完整代码:

;WITH temp AS (
	SELECT ROW_NUMBER() OVER (ORDER BY Car,Time ASC)AS ID,  Time,Car,SOC
		,SpeedFlg = CASE  WHEN  Speed >= 0 AND Speed < 30 THEN 1
						  WHEN  Speed >= 30 AND Speed < 60 THEN 2
						  WHEN  Speed >= 60 AND Speed < 80 THEN 3
						  WHEN  Speed >= 80 THEN 4
						  END
	FROM OrigData
	WHERE Car = ‘ABCDEFK10NZ000001‘
	AND Time >= ‘2016-06-01‘
	AND Time <= GETDATE()
),
final as(
	SELECT ROW_NUMBER()OVER(ORDER BY a.Car ) as tid,*
	FROM (
		SELECT a.Car,a.SpeedFlg,A.SOC
			,begintime = CASE WHEN A.ID = 1 OR b.SpeedFlg IS NULL THEN A.Time END
			,endtime = CASE WHEN c.SpeedFlg IS NULL THEN A.Time END
		FROM  temp a
		LEFT JOIN temp b ON a.ID = b.ID +1 AND B.SpeedFlg = A.SpeedFlg
		LEFT JOIN temp c ON a.ID = c.ID - 1  AND C.SpeedFlg = A.SpeedFlg
	)A
	where begintime is not null
	or endtime is not null
) 

select a.Car
	,SpeedFlg = case a.SpeedFlg when 1 then ‘0-30‘
								when 2 then ‘30-60‘
								when 3 then ‘60-80‘
								when 4 then ‘80以上‘
								end
	,a.begintime,b.endtime,a.SOC as beginsoc ,b.SOC as endsoc
from final a
INNER JOIN final b on a.Car = b.Car and a.SpeedFlg = b.SpeedFlg and a.tid = b.tid - 1