Question

我有一个数据表，我想导出到CSV。理想情况下，我想切换行和列，以便将数据分组得更好。

进一步解释，目前，数据库看起来像这样..

data_id     data_timestamp         data_value    
--------------------------------------------
1           2011-07-07 00:01:00    0.400  
1           2011-07-07 00:02:00    0.500
1           2011-07-07 00:03:00    0.600
1           2011-07-07 00:04:00    0.700
2           2011-07-07 00:01:00    0.100  
2           2011-07-07 00:02:00    0.200
2           2011-07-07 00:03:00    0.250
2           2011-07-07 00:04:00    2.300

我想要做的是使用data_timestamp值对data_value进行分组，以便将时间戳分组，并且每个data_id的每个data_value都显示在一列中，而不是一行。

data_timestamp         input_1    input_2     
--------------------------------------------
2011-07-07 00:01:00    0.400      0.100  
2011-07-07 00:02:00    0.500      0.200
2011-07-07 00:03:00    0.600      0.250
2011-07-07 00:04:00    0.700      2.300

以下是我正在使用的查询...

SELECT d.data_timestamp, d.input_1, d.input_2
FROM (
    SELECT data_timestamp,
    IF(data_id=1,data_value,NULL) AS 'input_1',
    IF(data_id=2,data_value,NULL) AS 'input_2' FROM data
) AS d ORDER BY data_timestamp ASC

但这并不是我想要的，因为只要一个data_id没有值，就会出现NULL值。 GROUP BY似乎也对data_value进行分组，这不是我想要的。

有什么建议吗？

修改

我已经尝试在外部查询中使用WHERE d.input_1 IS NOT NULL，但无法获得结果..

在WHERE之前...

data_timestamp         input_1    input_2     
--------------------------------------------
2011-07-07 00:01:00    0.400      NULL
2011-07-07 00:01:00    NULL       0.100  
2011-07-07 00:02:00    0.500      NULL
2011-07-07 00:02:00    NULL       0.200
2011-07-07 00:03:00    0.600      NULL
2011-07-07 00:03:00    NULL       0.250
2011-07-07 00:04:00    0.700      NULL
2011-07-07 00:04:00    NULL       2.300

添加WHERE d.input_1 IS NOT NULL将删除input_2值..

data_timestamp         input_1    input_2     
--------------------------------------------
2011-07-07 00:01:00    0.400      NULL
2011-07-07 00:02:00    0.500      NULL
2011-07-07 00:03:00    0.600      NULL
2011-07-07 00:04:00    0.700      NULL

另外，实际上，我有大约20个id来分组，所以不是最好的想法或者所有这些...

Answer 1

PIVOT既不容易（也不好）。我更喜欢使用CASE：

SELECT d.data_timestamp
     , SUM( CASE WHEN data_id =  1 THEN data_value ELSE 0 END ) AS 'input_1'
     , SUM( CASE WHEN data_id =  2 THEN data_value ELSE 0 END ) AS 'input_2'
     ...
     , SUM( CASE WHEN data_id = 20 THEN data_value ELSE 0 END ) AS 'input_20'
FROM data 
GROUP BY data_timestamp
ORDER BY data_timestamp ASC

但IF也适用于MySQL：

SELECT d.data_timestamp
     , SUM( IF(data_id =  1, data_value, 0) ) AS 'input_1'
     , SUM( IF(data_id =  2, data_value, 0) ) AS 'input_2'
     ...
     , SUM( IF(data_id = 20, data_value, 0) ) AS 'input_20'
FROM data 
GROUP BY data_timestamp
ORDER BY data_timestamp ASC

或者，您可以使用20级JOIN：

SELECT d.data_timestamp
     , d01.data_value AS 'input_1'
     , d02.data_value AS 'input_2'
     ...
     , d20.data_value AS 'input_20'
FROM
  ( SELECT DISTINCT d.data_timestamp
    FROM data
  ) AS d 
  LEFT JOIN data AS d01
    ON d01.data_timestamp = d.data_timestamp
    AND d01.data_id =  1
  LEFT JOIN data AS d02
    ON d02.data_timestamp = d.data_timestamp
    AND d02.data_id =  2
  ...                                   --- 20 JOINs
  LEFT JOIN data AS d20
    ON d20.data_timestamp = d.data_timestamp
    AND d20.data_id = 20
ORDER BY d.data_timestamp ASC

Answer 2

只需加入桌子！

SELECT dt1.data_timestamp, dt1.input_1, dt2.input_2
FROM data_timestamp dt1
JOIN data_timestamp dt2 
    on dt1.data_timestamp = dt2.data_timestamp 
    and dt2.input_1 is null
WHERE dt1.input_2 is null;

请注意，此查询假定每个input_1值都存在input_2的值。如果不是这种情况，请使用LEFT JOIN或CROSS JOIN等

使用GROUP BY的mysql数据透视查询结果

2 个答案: