Pivotera en tabell och visa n ordnade tidsserier

Prova detta:

SELECT r.id,
       r.`date`,
       GROUP_CONCAT(IF(r.device_id = 1,r.`VALUE`,NULL)) AS device_id_1,
       GROUP_CONCAT(IF(r.device_id = 2,r.`VALUE`,NULL)) AS device_id_2
FROM readings r
GROUP BY r.`DATE`
ORDER BY r.`DATE` ASC;

+----+---------------------+-------------+-------------+
| id | date                | device_id_1 | device_id_2 |
+----+---------------------+-------------+-------------+
|  1 | 2015-10-01 10:00:00 | 2           | 0.5         |
|  3 | 2015-10-01 10:01:00 | 3.1         | 7.5         |
+----+---------------------+-------------+-------------+
2 rows in set (0.00 sec)

Och här en version för dynamiska device_id's

SELECT  group_concat( 
    "GROUP_CONCAT(IF(device_id = ",
    device_id,",`value`,NULL)) AS device_id_",
    device_id
    SEPARATOR ',\n') INTO @sql_mid
FROM (SELECT DISTINCT device_id FROM readings ORDER BY device_id) AS r
ORDER BY device_id;

SELECT CONCAT("SELECT id, `date`, ", @sql_mid, " FROM readings GROUP BY `DATE` ORDER BY `DATE` ASC") INTO @sql;

PREPARE stmt FROM @sql;
EXECUTE stmt;
DEALLOCATE PREPARE stmt;

Den genererar alla kolumner

+----+---------------------+-------------+-------------+-------------+
| id | date                | device_id_1 | device_id_2 | device_id_3 |
+----+---------------------+-------------+-------------+-------------+
|  1 | 2015-10-01 10:00:00 | 2           | 0.5         | NULL        |
|  3 | 2015-10-01 10:01:00 | 3.1         | 7.5         | NULL        |
|  5 | 2015-10-01 10:11:00 | NULL        | NULL        | 9.9         |
+----+---------------------+-------------+-------------+-------------+
3 rows in set (0.00 sec)