How can I make this sub query more efficient?

Question

I have been learning more about making my sql queries more efficient, but I don't know if I'm doing this query properly.

I'm trying to find a set in one group, but not the other group. This is the original query which works:

SELECT
 cNumber
FROM
  [dataset.table]
WHERE
  DATE(date) >= '2016-01-01'
  AND cNumber NOT IN (
    SELECT
     cNumber
    FROM
     [dataset.table]
    WHERE
     date BETWEEN DATEDIFF(CURRENT_DATE(),-5, "QUARTER")
     AND DATEDIFF(CURRENT_DATE(), -1, "QUARTER"))
GROUP BY
 1

When I change my query to the one below, I get a larger number of results:

SELECT
 cNumber
FROM
 dataset.table
WHERE
 DATE(date) >= '2016-01-01'
 AND date NOT BETWEEN DATEDIFF(CURRENT_DATE(), -5, "QUARTER")
 AND DATEDIFF(CURRENT_DATE(), -1, "QUARTER")
GROUP BY
 1

For this type of query, do I need to include a NOT IN subquery or should I could I just use the second query?

I think I am confused in the second query by trying to include the AND orderCreationDate NOT BETWEEN . I thought that since I'm doing a subquery on the same table, that I would be able to wrap it up into one query.

Answer 1

I think this may do what you want:

SELECT cNumber
FROM [dataset.table]
GROUP BY cnumber
HAVING MAX(DATE(date)) >= '2016-01-01' AND
       SUM(CASE WHEN date BETWEEN DATEDIFF(CURRENT_DATE(), -5, "QUARTER") AND
                                  DATEDIFF(CURRENT_DATE(), -1, "QUARTER"))
                THEN 1 ELSE 0
           END) = 0;

That is, choose cnumber s that have a record sometime since 2016-01-01 and don't have one in the other period specified.

Whether this is faster or not, I do not know. You have to try it out.

Answer 2

for BigQuery (Legacy SQL) try below

SELECT cNumber
FROM [dataset.table]
GROUP BY cNumber
HAVING MAX(DATE(date)) >= '2016-01-01' 
AND SUM(CASE 
    WHEN DATE(date) BETWEEN DATE(DATE_ADD(CURRENT_DATE(), -5, "QUARTER")) 
    AND  DATE(DATE_ADD(CURRENT_DATE(), -1, "QUARTER"))
    THEN 1 ELSE 0
  END) = 0