SQL - Restrict the number of rows returned based on a count of rows

Question

Here is what my data looks like:

item_ID | group_ID | count_of_items_in_group

2|ABC|3

5|ABC|3

9|ABC|3

29|DEF|3

3|DEF|3

4|DEF|3

200|XYZ|2

300|XYZ|2

6         


        

Answer 1

Use the ROW_NUMBER() function for this:

SELECT *
FROM (select *,ROW_NUMBER() OVER (PARTITION BY group_id ORDER BY item_id) as RowRank 
      from items_in_groups
      )sub
WHERE RowRank <=2

Demo: SQL Fiddle

The ROW_NUMBER() function assigns a number to each row. PARTITION BY is optional, but used to start the numbering over for each value in that group, ie: if you PARTITION BY group_id then for each unique group_id value the numbering would start over at 1. ORDER BY of course is used to define how the counting should go, and is required in the ROW_NUMBER() function.

Answer 2

Can do this as a min and a max select statement, union them together and group by to eliminate any duplicates that one row groups would provide

 select item_id, group_id
from(
(select max(item_id) as item_id, group_id from table group by group_id) a
 union all
(select min(item_id) as item_id, group_id from table group by group_id) b) qry
group by item_id,group_id

Not the prettiest I'm sure, but it works. It's usually best to have some type of logic built in, not "I do not care which 2 are returned."

(Edit - I put the groupID in the groupo by and not item_ID. Ha, silly mistake I'll blame on dyslexia. Corrected now)

Answer 3

What about using ROW_NUMBER() to enum the number of fields of your data set, partitioned by each group_id.

And after that just return when is just less or equal to 2? or any number that you want

SELECT * FROM 
(
  select item_id, group_id, count_of_items, 
  ROW_NUMBER() OVER(PARTITION BY group_id ORDER BY count_of_items DESC) 
  AS RN
  from items_in_groups
) A
WHERE RN <= 2

Here is the Sql Fiddle