将行转换为列-面临问题。

Question

I'm having the following data set (note that my database/schemata is somewhat complex so it's not feasible to include all the details here)

This is a simplified version of my data which I extracted from my base tables and created a view .

workout_activity_record_id  attribute_metric_id attribute_metric_value  workout_exercise_set_created_on
234                                17              10                          2012-02-06 00:00:00
234                                18              30                          2012-02-06 00:00:00
234                                17              20                          2012-03-03 00:00:00
234                                18              12.9                        2012-03-03 00:00:00
234                                17              20                          2012-04-02 00:00:00
234                                18              40                          2012-04-02 00:00:00
.
.
. (So on for further workout_activity_record_ids)

I need to compute (look up in the data to understand better)

Grand Total = (10 * 30) + (20 * 12.9) + (20 * 40)

grouped by month.

A basic approach here that would aid in multiplying is transpose the rows into columns. So the above structure would become -

SELECT
       CASE attribute_metric_id
           WHEN '17' THEN attribute_metric_value END AS 'ex17',
       CASE attribute_metric_id WHEN '18' THEN attribute_metric_value END AS 'ex18', 
       workout_exercise_set_created_on
FROM exercise_attribute
WHERE workout_activity_record_id =234

For this, I went through several SO posts and after trying out various viable/non-viable (:D) options, I came up with the following query.

workout_activity_record_id  ex17    ex18    workout_exercise_set_created_on 
234                          10      NULL   2012-02-06 00:00:00
234                          NULL    30     2012-02-06 00:00:00
234                          20      NULL   2012-03-03 00:00:00
234                          NULL    12.9   2012-03-03 00:00:00
234                          20      NULL   2012-04-06 00:00:00
234                          NULL    40     2012-04-02 00:00:00

And the actual output I got was

\nworkout_activity_record_id ex17 ex18 workout_exercise_set_created_on  \n234 10 NULL 2012-02-06 00:00:00 \n234 NULL 30 2012-02-06 00:00:00 \n234 20 NULL 2012-03-03 00:00:00 \n234 NULL 12.9 2012-03-03 00:00:00 \n234 20 NULL 2012-04-06 00:00:00 \n234 NULL 40 2012-04-02 00:00:00 \n

Can anyone shed some light over this? I'd be grateful. Thanks!

Answer 1

This computes the grand total for you, grouped by year and month:

SELECT YEAR(workout_exercise_set_created_on) AS YEAR,
       MONTH(workout_exercise_set_created_on) AS MONTH,
       sum(SubTotal) AS GrandTotal
FROM
  (SELECT workout_exercise_set_created_on,
          max(CASE WHEN attribute_metric_id = 17 THEN attribute_metric_value END) * max(CASE WHEN attribute_metric_id = 18 THEN attribute_metric_value END) AS SubTotal
   FROM MyTable
   GROUP BY workout_exercise_set_created_on) a
GROUP BY YEAR(workout_exercise_set_created_on),
         MONTH(workout_exercise_set_created_on)

SQL Fiddle Example

Output

YEAR    MONTH   GRANDTOTAL
2012    2        300
2012    3        258
2012    4        800

Answer 2

Just add aggregate function :

SELECT
   MAX(CASE attribute_metric_id
        WHEN '17' THEN attribute_metric_value 
      END) AS 'ex17',
   MAX(CASE attribute_metric_id 
        WHEN '18' THEN attribute_metric_value 
       END) AS 'ex18', 
   workout_exercise_set_created_on
   FROM exercise_attribute
   WHERE workout_activity_record_id =234
   GROUP BY workout_exercise_set_created_on

If you need to show results for different workout_activity_record_id you should modify group by workout_activity_record_id,workout_exercise_set_created_on