Oracle 12C - 使用 Width_Bucket 并通过子句进行过度分区

Question

I have the following dataset (simplified) that constists of a 'WORK_TYPE' and a 'TASKTIME' associated with that work type.

+-----------+----------+--------+
| WORK_TYPE | TASKTIME | OUTPUT |
+-----------+----------+--------+
| TYPE1     |       10 |      1 |
| TYPE1     |       20 |      1 |
| TYPE1     |       30 |      2 |
| TYPE1     |       30 |      2 |
| TYPE2     |       10 |      1 |
| TYPE2     |       10 |      1 |
| TYPE2     |       20 |      2 |
| TYPE2     |       20 |      2 |
+-----------+----------+--------+

I wish to use the width_bucket function on this dataset. However I want to partition the data by the work_types so each type is grouped up irrespective of the entire dataset.

SELECT 
      TASKTIME
     ,WORK_TYPE
     ,WIDTH_BUCKET(TASKTIME,0,100,30) AS TASKTIME_BUCKET
     ,WIDTH_BUCKET(TASKTIME,0,100,30) OVER (PARTITION BY WORK_TYPE) AS TASKTME_BUCKET_WT --This Errors
FROM TABLE1

The first width_bucket works, however buckets the values across the whole dataset. I tried to use the OVER (PARITION BY WORK_TYPE) after the width_bucket, however this is causing the following error: ORA-00923: FROM keyword not found where expected

Any ideas?

Answer 1

If you want equal width buckets for each group, you can calculate separate min and max values for each gruop:

SELECT TASKTIME, WORK_TYPE,
       WIDTH_BUCKET(TASKTIME, 0, 100, 30) AS TASKTIME_BUCKET
       WIDTH_BUCKET(TASKTIME, MIN_TASKTIME, MAX_TASKTIME, 30) AS TASKTME_BUCKET_WT 
FROM (SELECT t1.*,
             MIN(TASKTIME) OVER (PARTITION BY WORK_TYPE) as MIN_TASKTIME,
             MAX(TASKTIME) OVER (PARTITION BY WORK_TYPE) as MAX_TASKTIME
      FROM TABLE1 t1
     ) t1