I need to write a query that selects a minimum value and it\'s second most minimum value from a list of integers.
Grabbing the smallest value is obvious:
<
SELECT MIN(value)
FROM TABLE
WHERE Value > (SELECT MIN(value) FROM TABLE)
select
value
from
(select
value,
dense_rank() over (order by value) rank
from
table)
where
rank = 2
Advantage: You can get the third value just as easy, or the bottom 10 rows (rank <= 10).
Note that the performance of this query will benefit from a proper index on 'value'.
Use an analytic function
SELECT value
FROM (SELECT value,
dense_rank() over (order by value asc) rnk
FROM table)
WHERE rnk = 2
The analytic functions RANK
, DENSE_RANK
, and ROW_NUMBER
are identical except for how they handle ties. RANK
uses a sports-style process of breaking ties so if two rows tie for a rank of 1, the next row has a rank of 3. DENSE_RANK
gives both of the rows tied for first place a rank of 1 and then assigns the next row a rank of 2. ROW_NUMBER
arbitrarily breaks the tie and gives one of the two rows with the lowest value a rank of 1 and the other a rank of 2.