与多列不同，然后按一栏分组，mysql

Question

I have a dataset that gives the underlying output to the underlying query:

Analyze
  .where(some_conditions)
  .joins(category: [category_subs: :sub])
  .select('user_id, product_id, category_subs.sub_id as sub_id, is_successful, category_subs.level as level')

+---------+-------------+---------------+--------+-------------+
| user_id | product_id  | is_successful | sub_id | level       |
+---------+-------------+---------------+--------+-------------+
| 3223    | 20441       | 1             | 56     | 14          |
| 3223    | 20441       | 1             | 72     | 22          |
| 3223    | 20441       | 1             | 20     | 22          |
| 3223    | 20443       | 1             | 20     | 14          |
| 3223    | 20443       | 1             | 20     | 22          |
| 3223    | 20443       | 1             | 25     | 6           |
| 3223    | 20443       | 1             | 7      | 22          |
| 3223    | 20444       | 1             | 20     | 14          |
| 3223    | 20444       | 1             | 20     | 22          |
| 3223    | 20444       | 1             | 25     | 6           |
| 3223    | 20444       | 1             | 7      | 22          |
| 3223    | 20445       | 0             | 20     | 14          |
| 3223    | 20445       | 0             | 20     | 22          |
| 3223    | 20445       | 0             | 25     | 6           |
| 3223    | 20445       | 0             | 7      | 22          |
| 3223    | 20448       | 1             | 20     | 14          |
| 3223    | 20448       | 1             | 20     | 22          |
| 3223    | 20448       | 1             | 25     | 6           |
| 3223    | 20448       | 1             | 7      | 22          |
| 3223    | 20449       | 1             | 20     | 14          |
| 3223    | 20449       | 1             | 4      | 2           |
| 3223    | 20449       | 1             | 24     | 14          |
+---------+-------------+---------------+--------+-------------+

All i want from that data is success by sub category with ignoring the level attribute.

I have tried this, but it doesn't work as i expected:

Analyze
  .where(some_conditions)
  .joins(category: [category_subs: :sub])
  .group('sub_id')
  .select('user_id, product_id, category_subs.sub_id as sub_id, sum(is_successful=0)')

Output the output that did not give the result I wanted:

+--------+----------------------+
| sub_id | sum(is_successful=1) |
+--------+----------------------+
| 4      | 1                    |
| 7      | 3                    |
| 20     | 8                    |
| 24     | 1                    |
| 25     | 3                    |
| 56     | 1                    |
| 72     | 1                    |
+--------+----------------------+

Expected output should be:

+--------+----------------------+
| sub_id | sum(is_successful=1) |
+--------+----------------------+
| 4      | 1                    |
| 7      | 3                    |
| 20     | 5                    |
| 24     | 1                    |
| 25     | 3                    |
| 56     | 1                    |
| 72     | 1                    |
+--------+----------------------+

I can get the data in ruby way, but i need it in the ActiveRecord::Relation instance.

Answer 1

select
  t.sub_id,
  count(*)
from (
  select sub_id
  from f
  where is_successful = 1
  group by
    product_id,
    sub_id
  order by sub_id
) t
group by t.sub_id;

That might work. You can use ActiveRecord from to create the inner query for the from statement and then use that result in the outer select.

Here a fiddle.