Summarize Null Values in Table with Group By - sql

Question

I have two tables:
Person(ID, Name)
Sports(person_ID, Sport)
The Problem: Sport can have NULL values. And if that is the case then if I group by ID the sport should be NULL.
SELECT p.ID, p.Name, s.Sport
FROM Person p
INNER JOIN Sports s ON p.ID=s.person_id
GROUP BY p.ID
Without the Group By the table looks like this:
p.ID p.Name s.Sport
1 tom soccer
1 tom NULL
2 lisa golf
2 lisa soccer
3 tim golf
3 tim NULL
What I want now:
1 tom NULL
2 lisa golf
3 tim NULL
But what I get:
1 tom soccer
2 lisa golf
3 tim golf
I've tried subselects and ifs but I couldn't get anything to work. Thanks in advance!

Answer 1

Here is a query which should generate your expected result set, though as #jarlh has pointed out, it isn't clear why Lisa should play golf over soccer.
SELECT
p.ID,
p.Name,
CASE WHEN COUNT(CASE WHEN s.Sport IS NULL THEN 1 END) > 0
THEN NULL ELSE MIN(s.Sport) END AS Sport
FROM Person p
INNER JOIN Sports s
ON p.ID = s.person_id
GROUP BY
p.ID,
p.name;
Note that I group by both the ID and name, which would be required on many databases (though perhaps not SQLite).

Answer 2

you can't manage the NULL value with aggreagtion function as MIN()
but you could try
SELECT p.ID, p.Name, min(ifnull(s.Sport,''))
FROM Person p
INNER JOIN Sports s ON p.ID=s.person_id
GROUP BY p.ID, p.name

Answer 3

Assuming the version of SQLLite you are using supports row_number(), please try below, you can set a row_number to 1 if you order by s.sport ASC, then select the first row for each category. If there is NULL, it should locate at the top row of each category via this query. You don't need to use group by:
;with cte as (
select p.ID, p.Name, s.Sport,
ROW_NUMBER() OVER (PARTITION BY p.ID ORDER BY s.Sport ASC) AS rn
FROM Person p INNER JOIN Sports s ON p.ID=s.person_id
)
select *
from cte
where rn=1

Answer 4

You can do this with a correlated subquery, avoiding the join in the outer query:
select p.*,
(select s.sport
from sports s
where s.personId = p.id
order by (s.sport is null) desc, s.sport asc
) as min_sport
from person p;
This may prove useful under some circumstances. With an index on sports(personid, sport), it might be faster than the group by, depending on the data (lots of people, few sports per person).
Also, this is slightly different from your query because it returns all people, even those with no sports.