How to replace a NOT IN Query to a join?

Question

Hi,
I have heard join queries are faster than NOT IN. Could anyone please help in converting the below query to a join query? Since, there is only one table, am not sure, if self- reference needs to be made.
Select id, name from department
where id NOT IN
*(Select id from department group by id having count(id)>1)*
order by id;Any help in this regard will be highly appreciated.
Thanks.

Answer 1

A normal NOT IN subquery can usually be rewritten with a left join and a check for null.
example
/* subquery */
Select id, name from department
where id NOT IN (Select dept_id from employees);
/* joined query */
Select id, name from department
left join employees on department.id = employees.dept_id
where employees.dept_id is null;You have a group by logic in there. THis is different. it could be made into a join like this.
/* original */
Select id, name from department
where id NOT IN (Select id from department group by id having count(id)>1);
/* join */
Select d1.id, d1.name
from department d1
join department d2 on d1.id = d2.id
group by d1.id, d1.name
having count(*) > 1;However your statement does not make any sense. The ID column should be unique. So there can't be any duplicates.
If the ID is not unique and if you just want to see which departments do not have duplicates then a logic is better (faster) where you use the table departments only one time.
This code would be almost equivalent.
Select id, min(name) as first_department_name, count(*) departments_duplicate_ID#
from department
group by id
having count(*) > 1;Edited by: Sven W. on Jun 22, 2011 4:51 PM - typos