查找平均值、最大值和最小值以及值出现的对应日期时间

Question

我正在处理一个 SQL 查询，我必须找出任何日期的每个组的平均值、最大值、最大值-日期时间、最小值、最小值-日期时间。为此，我使用了以下查询

SELECT
    Dl.TagDescID,
    convert( decimal(18,2), AVG( Dl.Value ) ) AS 'TotalAvgValue',
    convert(decimal(18,2), MAX( Dl.Value ) ) AS 'MaxValue',
    MAX( Dl.Date ) MaxDate,
    convert( decimal(18,2), MIN( Dl.Value ) ) AS 'MinValue',
    Min( Dl.Date ) MinDate
FROM
    tblDataLog AS Dl WITH (nolock) 
    INNER JOIN tblTagDescription AS TD WITH (nolock) ON Dl.TagDescID = TD.ID 
WHERE
    ( CONVERT(date, Dl.Date, 103) = CONVERT(date, @StartDate, 103) )
    AND
    ( TD.GroupName = @Group )
GROUP BY
    Dl.TagDescID
`

     TagDescID  TotalAvgValue   MaxValue    MaxDate MinValue    MinDate
       1    36.59   36.59   2020-01-07 10:13:42.293 36.59   2020-01-07 10:13:32.750
       2    20.49   20.49   2020-01-07 10:13:42.293 20.49   2020-01-07 10:13:32.750

In the above Data Set The MaxValue of TagDescID 1 is 36.59 and its Date as 'MaxDate' should be the exact datetime when this maxvalue Arrives in the table tblDataLog. here Max-Date of MaxValue of TagDescID 1 is '2020-01-07 10:13:42.293' but the value should have been '2020-01-07 08:13:42.293'. Same for MinValue and MinDate For each TagDescID.

How to solve it?

Answer 1

问题是您实际上是在同时要求多个聚合和分组集。您观察到的是，每个 column 聚合 表达式都独立于其他列返回该聚合的结果，这是设计使然。

在上述查询中MaxDate 和MinDate 应该是当MaxValue 或MinValue 插入到表tblDataLog 时的日期时间。

在 SQL Server 中，我们有多种查询方法。

一种方法是查询集合中的最大值和最小值，然后重新查询集合中的最大值和最小值日期。

在此示例中，我使用 CTE 而不是简单的嵌套查询，因为我发现语法更具可读性。

两个CROSS APPLY 查询将为Daily CTE 中的每个结果评估一次。 此解决方案可以更有效地对相应日志记录的 DateTime 使用内联查找，因为它不必为 tblDataLog 中的每一行查找一次，此解决方案将导致对每一行进行 2 次额外查找由 group by 返回。

;WITH Daily as
(
    SELECT  
        Dl.TagDescID, 
        convert(decimal(18,2),AVG(Dl.Value)) AS [TotalAvgValue],
        convert(decimal(18,2),MAX(Dl.Value)) AS [MaxValue], 
        convert(decimal(18,2),MIN(Dl.Value)) AS [MinValue] 
    FROM @tblDataLog AS Dl  
    INNER JOIN @tblTagDescription AS TD ON Dl.TagDescID = TD.ID 
    WHERE (CONVERT(date, Dl.Date, 103) = CONVERT(date, @StartDate, 103)) AND (TD.GroupName=@Group) 
    GROUP BY Dl.TagDescID  
)
SELECT Daily.TagDescID, TotalAvgValue, MaxValue, [Max].Date MaxDate, MinValue, [Min].Date MinDate
FROM Daily
CROSS APPLY (SELECT MAX(DlMax.Date) Date FROM @tblDataLog DlMax WHERE DlMax.TagDescID = Daily.TagDescId AND (CONVERT(date, DlMax.Date, 103) = CONVERT(date, @StartDate, 103)) AND DlMax.Value = Daily.MaxValue) as [Max]
CROSS APPLY (SELECT MIN(DlMin.Date) Date FROM @tblDataLog DlMin WHERE DlMin.TagDescID = Daily.TagDescId AND (CONVERT(date, DlMin.Date, 103) = CONVERT(date, @StartDate, 103)) AND DlMin.Value = Daily.MinValue) as [Min]

另一种方法是按日期分组，而不是过滤明确的日期值，因此这将按日期和标签进行汇总，如果需要，可为您提供跨多个日期和多个标签的报告：

;WITH Daily as
(
    SELECT  
        Dl.TagDescID, 
        TD.GroupName,
        CONVERT(date, Dl.Date, 103) as Date,
        MAX(Dl.Date) as Max_Date,
        MIN(Dl.Date) as Min_Date,
        convert(decimal(18,2),AVG(Dl.Value)) AS [TotalAvgValue],
        convert(decimal(18,2),MAX(Dl.Value)) AS [MaxValue], 
        convert(decimal(18,2),MIN(Dl.Value)) AS [MinValue] 
    FROM @tblDataLog AS Dl  
    INNER JOIN @tblTagDescription AS TD ON Dl.TagDescID = TD.ID 
    GROUP BY Dl.TagDescID, TD.GroupName, CONVERT(date, Dl.Date, 103)
)
SELECT Daily.Date, Daily.GroupName, Daily.TagDescID, TotalAvgValue, MaxValue, [Max].Date MaxDate, MinValue, [Min].Date MinDate
FROM Daily
CROSS APPLY (SELECT MAX(DlMax.Date) Date FROM @tblDataLog DlMax WHERE DlMax.TagDescID = Daily.TagDescId AND DlMax.Date >= Daily.Min_Date AND DlMax.Date <= Daily.Max_Date) [Max]
CROSS APPLY (SELECT MIN(DlMin.Date) Date FROM @tblDataLog DlMin WHERE DlMin.TagDescID = Daily.TagDescId AND DlMin.Date >= Daily.Min_Date AND DlMin.Date <= Daily.Max_Date) [Min]
-- Optional Date Filter
-- WHERE TD.GroupName=@Group AND Daily.Date = @StartDate

在我的测试集中，返回：

Date        GroupName   TagDescID   TotalAvgValue   MaxValue    MaxDate                      MinValue   MinDate
2020-03-11  Test        1           5.00            8.00        2020-03-11  22:00:00.000     2.00       2020-03-11 10:00:00.000
2020-03-15  Test        1           5.00            8.00        2020-03-15  13:00:00.000     2.00       2020-03-15 06:00:00.000

取消注释where 子句以启用您的日期过滤器，或者您可以将其更改为一个日期范围，您仍然会看到按天显示的结果。

Answer 2

我想这就是你所追求的。 Value时对应的Date是最小值和最大值。您可以使用sub-query 来执行此操作

SELECT
    Dl.TagDescID,
    convert( decimal(18,2), AVG( Dl.Value ) ) AS 'TotalAvgValue',
    convert(decimal(18,2), MAX( Dl.Value ) ) AS 'MaxValue',
    --MAX( Dl.Date ) MaxDate,
    (SELECT TOP 1 x.[Date] FROM  tblDataLog x WHERE x.TagDescID = Dl.TagDescID ORDER BY Value DESC) As MaxDate,
    convert( decimal(18,2), MIN( Dl.Value ) ) AS 'MinValue',
    --Min( Dl.Date ) MinDate
    (SELECT TOP 1 x.[Date] FROM  tblDataLog x WHERE x.TagDescID = Dl.TagDescID ORDER BY Value) As MinDate
FROM
    tblDataLog AS Dl WITH (nolock) 
    INNER JOIN tblTagDescription AS TD WITH (nolock) ON Dl.TagDescID = TD.ID 
WHERE
    ( CONVERT(date, Dl.Date, 103) = CONVERT(date, @StartDate, 103) )
    AND
    ( TD.GroupName = @Group )
GROUP BY
    Dl.TagDescID

编辑：正如其他人评论的那样，避免CONVERT(date, Dl.Date, 103) = CONVERT(date, @StartDate, 103)

您可以将其替换为

WHERE Dl.Date >= @StartDate
AND   Dl.Date <  DATEADD (DAY, 1, @StartDate)