asked on

SQL MAX Help

Hi,
I have a very simple select statement that returns this dataset

SELECT parent_id, component_id, control_id FROM item where parent_id = '14923'

parent_id      component_id      control_id
14923      12054      53715
14923      12054      53721
14923      12054      53715
14923      12054      53721
14923      12423      53715
14923      12423      53721
14923      14176      53715
14923      14176      53721
14923      14783      53715
14923      14783      53721
14923      14795      53715
14923      14795      53721
14923      14840      53715
14923      14840      53721
14923      14880      53715
14923      14880      53721
14923      14924      53715
14923      14924      53721
14923      14926      53715
14923      14926      53721

I need to add an additional where clause that only returns the highest control values, so in this case the returned dataset would look like this:

parent_id      component_id      control_id
14923      12054      53721
14923      12054      53721
14923      12423      53721
14923      14176      53721
14923      14783      53721
14923      14795      53721
14923      14840      53721
14923      14880      53721
14923      14924      53721
14923      14926      53721

Any suggestions as to how I can accomplish this please

ASKER CERTIFIED SOLUTION

Dan Craciun

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

Vikas Garg

HI,

Select Parent_id, Component_id, (select max(control_id) from table) ControlID from Table

Open in new window

Vitor Montalvão

Use the MAX() function:

SELECT parent_id, component_id, MAX(control_id)
FROM item 
WHERE parent_id = '14923'
GROUP BY parent_id, component_id

Open in new window

Jim Horn

The above answer is correct.

If you want to eliminate duplicates, use DISTINCT.

SELECT DISTINCT parent_id, component_id, MAX(control_id)
FROM item 
WHERE parent_id = '14923'
GROUP BY parent_id, component_id

Open in new window

Also, if you'd like further reading on aggregates such as MAX then I have an article out there called SQL Server GROUP BY Solutions that would be an excellent reading.

Vitor Montalvão

Jim, you won't need the DISTINCT since it's already aggregate by parent_id and component_id so duplicates will be automatically dropped.

andyw27

ASKER

Thanks, worked like a charm.

Vitor Montalvão

Andy, any of these solutions works (Vikas forgot the WHERE clause) but there's no need to use subselects.
You can do it in a more direct and efficient way.

Cheers

Dan Craciun

Glad I could help!

@Vitor and Jim: on your queries, if a component_id does not have a row where control_id is 53721, it will be listed with the highest control_id value if finds. The OP wanted only rows where control_id is 53721. At least that's the way I understood the question.

Vitor Montalvão

Dan, I didn't understand the question like that:

where clause that only returns the highest control values

Vikas Garg

Hello,

SELECT parent_id, component_id, control_id FROM item a
INNER JOIN
(Select Parent_id, Component_id, (select max(control_id) from item) ControlID from item) B
ON A.parent_id = B.Parent_id AND A.component_id = B.component_id AND A.control_id = B.control_id

Jim Horn

Vitor - You are correct. What was I thinking...