在同一维度表上从具有多个条件的事实表中选择记录

Question

Let's say I have a survey data schema, my fact table represents each person responding this survey, one of the dimension tables has all the answers to the survey, what's the most efficient way to get all the records from the fact table that match multiple conditionals on the answers table?

FACT TABLE (records):

| id | name | date | gender |

DIMENSION TABLE (answers):

| record_id | question_id | value |

I could do a new left join for every question, but this seems very inneficient if I want to find a record that answered multiple questions

SELECT * FROM records r 
    left join answers a on r.id = a.record_id 
    left join answers a2 on r.id = a2.record_id 
    where (a.question_id = 1 and a.value = 2) 
    and (a2.question_id = 3 and a2.value = 1);

Any alternative to this?

Answer 1

If I understand your question correctly, you need:

   SELECT 
       *
     FROM records r 
LEFT JOIN answers a ON r.id            = a.record_id
                   AND a.question_id   = 1 
                   AND a.value         = 2 
LEFT JOIN answers a2 ON r.id           = a2.record_id 
                    AND a2.question_id = 3 
                    AND a2.value       = 1;

Answer 2

After some more digging it looks like the most scalable and dynamic way is using EXISTS

SELECT * FROM records r  WHERE 
  EXISTS (SELECT * FROM answers WHERE record_id = r.id AND question_id = 1 AND value = 1) AND
  EXISTS (SELECT * FROM answers WHERE record_id = r.id AND question_id = 3 AND value = 1)

Answer 3

I would hardcode the list of questions/values you want and use join:

with
 question_filter as (
    select 1 as question_id, 2 as value
    union select 3, 1
)
select *
from records r
join answers a
on r.id=a.record_id
join question_filter
using (question_id,value)
;