Question

我正在尝试按查询显示来自分组结果的最多错误

我认为我的初始组是正确的，但无法找出子查询来按模块对错误数进行计数并滤除最高计数。

SELECT Code AS Module
   , ErrCode As ErrorCode
   , Count(ErrCode) AS ErrorCount
   , CAST([Msg-EN] as varchar(max)) as ErrorMsg
FROM ViewAllAlarm
WHERE RiseTime >= DateAdd(Hour, -12, GetDate())
GROUP BY Code, ErrCode,  CAST([Msg-EN] as varchar(max))
Order by Module ASC

结果：

Module | ErrorCode | ErrorCount | ErrorMsg
-------+-----------+------------+---------
F108   | 194       |        127 | Error2
F108   | 358       |        209 | Error1
F109   | 129       |         11 | Error3
F110   | 129       |        200 | Error3
F110   | 358       |        300 | Error1
F111   | 358       |          2 | Error1
F111   | 129       |          5 | Error3
F112   | 129       |         12 | Error3
F113   | 358       |        200 | Error1
F114   | 194       |        300 | Error2

我现在需要按模块总计错误数量（ErrorCount），并显示计数最高的模块（前5名）。

这给出了F108的336个错误，F109的11个错误，F110的500个错误，等等。我想展示五个具有最高错误计数的模块，它们是F110（500），F108（336），F114（300）， F113（200），F112（12）。不会显示模块F129（11）和F111（7）。

Module | ErrorCode | ErrorCount | ErrorMsg
-------+-----------+------------+---------
F108   | 194       |        127 | Error2
F108   | 358       |        209 | Error1
F110   | 129       |        200 | Error3
F110   | 358       |        300 | Error1
F112   | 129       |         12 | Error3
F113   | 358       |        200 | Error1
F114   | 194       |        300 | Error2

这是在MS 2016 SQL服务器上。

Answer 1

使用IN子句仅选择错误次数最多的五个模块。

WITH current AS
(
  SELECT *
  FROM viewallalarm
  WHERE risetime >= DATEADD(HOUR, -12, GETDATE())
)
SELECT
    code AS module
  , errcode As errorcode
  , COUNT(*) AS errorcount
  , MAX(CAST([Msg-EN] as VARCHAR(MAX))) AS errormsg
FROM current
WHERE module IN
(
  SELECT TOP(5) WITH TIES code
  FROM current
  GROUP BY code
  ORDER BY COUNT(*) DESC
)
GROUP BY code, errcode
ORDER BY module;

或使用窗口功能：

SELECT *
FROM
(
  SELECT TOP(5) WITH TIES *
  FROM
  (
    SELECT
        code AS module
      , errcode As errorcode
      , COUNT(*) AS errorcount
      , CAST([Msg-EN] as VARCHAR(MAX)) AS errormsg
      , SUM(COUNT(*)) OVER(PARTITION BY code) AS total
    FROM viewallalarm
    WHERE risetime >= DATEADD(HOUR, -12, GETDATE())
    GROUP BY code, errcode, CAST([Msg-EN] AS VARCHAR(MAX))
  ) summed
  ORDER BY DENSE_RANK() OVER (ORDER BY total DESC)
) top_five
ORDER BY module;

Answer 2

嗯。如果只需要模块，则使用group by和top：

SELECT TOP (5) Code AS Module, COUNT(*) as cnt
FROM ViewAllAlarm
WHERE RiseTime >= DateAdd(Hour, -12, GetDate())
GROUP BY Code
ORDER BY COUNT(*) DESC;

如果您也想要这些错误，则可以使用窗口函数来解决，但我认为简单的连接可能是最简单的：

select vaa.*
from ViewAllAlarm vaa join
     (SELECT TOP (5) Code, COUNT(*) as cnt
      FROM ViewAllAlarm
      WHERE RiseTime >= DateAdd(Hour, -12, GetDate())
      GROUP BY Code
      ORDER BY COUNT(*) DESC
     ) t5
     on t5.code = vaa.code;

Answer 3

如果我的SQL则使用限制

select Code AS Module, count(*) as cnt
FROM ViewAllAlarm
WHERE RiseTime >= DateAdd(Hour, -12, GetDate())
GROUP BY Code
ORDER BY COUNT(*) DESC limit 5

如何计算分组查询中的项目数并显示Top（x）？

3 个答案: