Make sure that min() only retrieves one value - sql

Question

I have a SQL query that retrieves data from a table that lists athletes' 100- and 200-meter race times. The query only retrieves the best race time of each athlete based on the athlete_id, it also wants to know if the race-time is a 100 or 200 meter-time (event_code).
So a runner can have several race-times but the query only gets the best race-time from each runner at each event.
The problem is that if an athlete have done exactly the same best race time two or more times, the query retrieves all those race times. How can I make sure the query only retrives one value?
Here is the code:
select r.*
from result r
inner join (
select athelete_id, min(result_time) as FastestTime
from result
where event_code = 1
group by athelete_id
) rm on r.athelete_id = rm.athelete_id and r.result_time = rm.FastestTime

Answer 1

It is a pain in SQL Server 2000. This would be much easier using row_number, but that requires 2005.
However, the idea is simple, you just need one more layer of subqueries:
select r.*
from result r join
(select r.athelete_id, MIN(result_id) as minresult_id
from result r inner join
(select athelete_id, min(result_time) as FastestTime
from result
where event_code = 1
group by athelete_id
) rm
on r.athelete_id = rm.athelete_id and r.result_time = rm.FastestTime
group by r.athelete_id
) aft
on r.result_id = minresult_id
The innermost subquery is basically your subquery. Then, this is aggregated by athelete_id, to get the minimum result_id, which is used for the final join.

Answer 2

Sorry, I misunderstood the question a bit, but you can manage it somehow like this for all competitor like this:
select MIN(r.id), r.athelete_id, r.result_time, r....
from result r
inner join (
select athelete_id, min(result_time) as FastestTime
from result
where event_code = 1
group by athelete_id
) rm on r.athelete_id = rm.athelete_id and r.result_time = rm.FastestTime
GROUP BY r.athelete_id, r.result_time, r....
--(as how many column you have, except r.ID)
Try this:
select top 1 r.*
from result r
inner join (
select athelete_id, min(result_time) as FastestTime
from result
where event_code = 1
group by athelete_id
) rm on r.athelete_id = rm.athelete_id and r.result_time = rm.FastestTime
OR
SET ROWCOUNT 1
select r.*
from result r
inner join (
select athelete_id, min(result_time) as FastestTime
from result
where event_code = 1
group by athelete_id
) rm on r.athelete_id = rm.athelete_id and r.result_time = rm.FastestTime

Answer 3

Largely a similar approach to Gordon's, but breaks ties using the earliest event (which should be the identity column, if that's what result_id is, but that's not always guaranteed).
DECLARE #event_code INT;
SET #event_code = 1;
SELECT r.*
FROM dbo.result AS r
INNER JOIN
(
SELECT r2.athelete_id, x.mintime, mindate = MIN(r2.result_date)
FROM
(
SELECT athelete_id, mintime = MIN(result_time)
FROM dbo.result
WHERE event_code = #event_code
GROUP BY athelete_id
) AS x
INNER JOIN dbo.result AS r2
ON x.athelete_id = r2.athelete_id
AND x.mintime = r2.result_time
WHERE r2.event_code = #event_code
GROUP BY r2.athelete_id, x.mintime
) AS y
ON r.athelete_id = y.athelete_id
AND r.result_time = y.mintime
AND r.result_date = y.mindate
WHERE r.event_code = #event_code;
Much easier in SQL Server 2005, of course:
;WITH x AS
(
SELECT athelete_id, --... other columns,
rn = ROW_NUMBER() OVER (PARTITION BY athelete_id ORDER BY result_time, result_date)
FROM dbo.result
WHERE event_code = 1
)
SELECT athelete_id, --... other columns
FROM x WHERE rn = 1;