如何根据解析日期对事件进行计数-GROUP BY问题

Question

[我正在尝试计算每天发生的FAILURE个事件，并将事件存储在MainEventTable中，其列为EventDateTime，EventId和EventStatus。我正在使用SQL Server Management Studio 2016，但无法识别DATEFROMPARTS函数。到目前为止，这是我编写的代码：

SELECT 
    t.EventDate, 
    SUM(t.EventCount) AS EventCount 
FROM (
    SELECT 
        CAST(
                (
                    CAST(
                        DATEPART(yyyy,s.EventDateTime) 
                        AS VARCHAR(4)
                        ) + '-' + 
                    CAST(
                        DATEPART(mm,s.EventDateTime) 
                        AS VARCHAR(2)
                        ) + '-' + 
                    CAST(
                        DATEPART(dd,s.EventDateTime) 
                        AS VARCHAR(2)
                        )
                ) AS DATE
            ) AS EventDate,
        Count(s.EventId) AS EventCount 
    FROM (
        SELECT 
            EventDateTime, 
            EventId 
        FROM 
            MainEventTable WITH(NOLOCK) 
        WHERE EventDateTime > '2016-12-07 00:00:00' 
        AND EventStatus = 'FAILURE'
        ) AS s GROUP BY CAST(
                (
                    CAST(
                        DATEPART(yyyy,s.EventDateTime) 
                        AS VARCHAR(4)
                        ) + '-' + 
                    CAST(
                        DATEPART(mm,s.EventDateTime) 
                        AS VARCHAR(2)
                        ) +  '-' + 
                    CAST(
                        DATEPART(dd,s.EventDateTime) 
                        AS VARCHAR(2)
                        )
                ) AS VARCHAR(10)
            )
    ) AS t
GROUP BY t.EventDate;

更新：（感谢@wrslphil和@ PM_77-1为我的GROUP BY问题提供帮助）我已经修复了上面的GROUP BY问题，发现这很有效，尽管它很笨重。 @KeithL在下面对其进行了简化...

Answer 1

select CAST(EventDateTime AS DATE),COUNT(*)
FROM MainEventTable
WHERE EventDateTime > '2016-12-07 00:00:00' 
        AND EventStatus = 'FAILURE'
GROUP BY CAST(EventDateTime AS DATE)

Answer 2

这里没什么可做的，但是，如果您有一个不在按列表分组或在select子句中汇总的项目，则查询绝对不会运行

我认为您可能需要对您的count字段求和...就像这样

SELECT
    t.EventDate,
    SUM(t.EventCount) as EventCount
FROM (
    SELECT
        CAST(
                (
                    CAST(
                        DATEPART(yyyy,s.EventDateTime)
                        AS VARCHAR(4)
                        ) +
                    CAST(
                        DATEPART(mm,s.EventDateTime)
                        AS VARCHAR(2)
                        ) +
                    CAST(
                        DATEPART(dd,s.EventDateTime)
                        AS VARCHAR(2)
                        )
                ) AS DATE
            ) AS EventDate,
        Count(s.EventId) As EventCount
    FROM (
        SELECT
            EventDateTime,
            EventId
        FROM
            MainEventTable WITH(NOLOCK)
        WHERE EventDateTime > '2016-12-07 00:00:00'
        AND EventStatus = 'FAILURE'
        ) AS s
    ) AS t
GROUP BY t.EventDate;

该查询仍然不太可能运行，因为内部查询中存在相同的问题。我可能只是在内部查询中对日期进行转换，然后将其封装在包含count和group by的查询中]

Answer 3

您的子查询具有以下结构：

SELECT field1, COUNT(field2) 
FROM theTable

在同一选择中，您使用“原始”字段和聚合（COUNT）。您没有GROUP BY子句。

您正在混淆您的SQL引擎，因为无法确定应该选择field1的值。

如果要计算field2的非NULL值的记录数，则field1的非[NULL]值是：

SELECT field1, COUNT(field2) 
FROM theTable
GROUP BY field1