透视 sql server 表

Question

如何使用 SQL 查询将 SQL Server 表 meterdata 转换为 Result_Table？

MeterData 表存储功率和频率，主键(meterno, date, timeblock)，meterno 列有变量号。的价值观。

MeterData 表：

meterno date      timeblock power   frequency       
--------------------------------------------------
89      1-Apr-16    1       500       50.02     
89      1-Apr-16    2       100       49.99     
90      1-Apr-16    1       200       50.02     
90      1-Apr-16    2       300       49.89     

Result_Table

date    timeblock   89_power 90_power   89_frequency 90_frequency   
1-Apr-16    1       500      200        50.02       50.02   
1-Apr-16    2       100      300        49.99       49.89

为简单起见，我们假设meterno 列已固定编号。值说（89,90）。我正在尝试以下查询，但没有得到结果。有人可以编辑查询，以便输出是上面的 Results_Table。另外最好请使用数据透视查询来回答。查询是

SELECT
*
 FROM
 (
    SELECT
 *
    FROM meterdata  
 ) AS P
PIVOT
(
  sum (power ) FOR meterno IN (89, 90)
) AS pv1 
PIVOT
(
  sum (frequency ) FOR meterno IN (89, 90)
) AS pv2 

GO

Answer 1

动态 SQL 将为您提供帮助（我使用了 ##MeterData 和 tempdb，您必须使用您的实际表和数据库名称）。将使用源表中的任意数量的列。几个cmets：

• 您需要将CAST 或CONVERT 的所有列归为一种类型（我在@p 变量中使用了nvarchar(10)）；
• 我排除列meterno、date、timeblock，因为您的数据集中没有用作参数；
• 您可以在执行前PRINT @sql查看已形成执行的内容。

这里是查询：

DECLARE @p nvarchar(max), @c nvarchar(max), @s nvarchar(max), @sql nvarchar(max)

--That will get you this:
--,CAST([power] as nvarchar(10)) [power],CAST([frequency] as nvarchar(10)) [frequency]
SELECT @p = (
SELECT ',CAST(' + QUOTENAME(COLUMN_NAME) + ' as nvarchar(10)) '+ QUOTENAME(COLUMN_NAME)
FROM tempdb.INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME = N'##MeterData' AND COLUMN_NAME NOT IN ('meterno','date','timeblock')
FOR XML PATH('')
)
--[power],[frequency]
SELECT @c = STUFF((
SELECT ',' + QUOTENAME(COLUMN_NAME) 
FROM tempdb.INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME = N'##MeterData' AND COLUMN_NAME NOT IN ('meterno','date','timeblock')
FOR XML PATH('')),1,1,'')
--[frequency89],[frequency90],[power89],[power90]
SELECT @s = STUFF((
SELECT DISTINCT ','+QUOTENAME(COLUMN_NAME + CAST(meterno as nvarchar(10)))
FROM ##MeterData m
CROSS JOIN (
    SELECT COLUMN_NAME
    FROM tempdb.INFORMATION_SCHEMA.COLUMNS
    WHERE TABLE_NAME = N'##MeterData' AND COLUMN_NAME NOT IN ('meterno','date','timeblock')
) as p 
FOR XML PATH('')),1,1,'')
--Put it all in dynamic query
SELECT @sql = '
SELECT *
FROM (
    SELECT  [date],
            timeblock,
            params + CAST(meterno as nvarchar(10)) as params,
            [values]
    FROM (
        SELECT  meterno, 
                [date], 
                timeblock'
                +@p+
                '
        FROM ##MeterData
        ) as p
    UNPIVOT (
        [values] FOR params IN ('+@c+')
    ) as unpvt
) as p2
PIVOT (
    MAX([Values]) FOR params IN ('+@s+')
) as pvt'
--And execute
EXEC sp_executesql @sql

输出：

date        timeblock   frequency89 frequency90 power89 power90
2016-04-01  1           50.02       50.02       500     200
2016-04-01  2           49.99       49.89       100     300

Answer 2

这是一个您可以尝试的不使用 SQL Server PIVOT 操作的数据透视查询：

SELECT date, timeblock
    SUM(CASE WHEN meterno=89 THEN power     ELSE 0 END) AS meterno-89_power,
    SUM(CASE WHEN meterno=90 THEN power     ELSE 0 END) AS meterno-90_power,
    SUM(CASE WHEN meterno=89 THEN frequency ELSE 0 END) AS meterno-89_frequency,
    SUM(CASE WHEN meterno=90 THEN frequency ELSE 0 END) AS meterno-89_frequency
FROM MeterData
GROUP BY date, timeblock