Hackerrank Winners chart SQL advance level question - sql

Question

This is a new question Hackerrank has added for the advance level SQL certification. I was not able to solve it at all. Can anyone help?
There were a number of contests where participants each made number of attempts. The attempt with the highest score is only one considered. Write a query to list the contestants ranked in the top 3 for each contest. If multiple contestants have the same score in each contest, they are at the same rank.
Report event_id, rank 1 name(s), rank 2 name(s), rank 3 name(s). Order the contest by event_id. Name that share a rank should be ordered alphabetically and separated by a comma.
Order the report by event_id

Answer 1

The following is some sample data for your scenario. A table of contestants, and the attempts made. Made each person's attempts on their own line so you can see obvious different attempts per person.
create table contestants
( id int identity(1,1) not null,
personName nvarchar(10) )
insert into contestants ( personName )
values ( 'Bill' ), ('Mary'), ('Jane' ), ('Mark')
create table attempts
( id int identity(1,1) not null,
contestantid int not null,
score int not null )
insert into attempts ( contestantid, score )
values
( 1, 72 ), ( 1, 88 ), (1, 81 ),
( 2, 83 ), ( 2, 88 ), (2, 79), (2,86),
( 3, 94 ),
( 4, 79 ), (4, 87)
Now, the simple premise is each contestants best score which is a simple MAX() of the score per contestant.
select
contestantid,
max( score ) highestScore
from
attempts
group by
contestantid
The result of the above query is the BASIS of the final ranking. So I have put that query as the FROM source. So instead of a table, the from is the result of the above query which I have aliased "PreAgg" for the pre-aggregation per contestant.
select
ContestantID,
c.personName,
DENSE_RANK() OVER ( order by HighestScore DESC ) as FinalRank
from
(select
contestantid,
max( score ) highestScore
from
attempts
group by
contestantid ) preAgg
JOIN Contestants c
on preAgg.contestantid = c.id
The join to the contestant is easy enough to pull the name, but now look at the DENSE_RANK() clause. Since there is no grouping per score, such as say the Olympics where there is a specific sport, and each sport has highest ranks, we do not need the "PARTITION" clause.
The ORDER BY clause is what you want. In this case, the HighestScore column from the pre-aggregation query and want that in DESCENDING order so the HIGHEST score is at the top and going down. The "as" gives it the final column name.
DENSE_RANK() OVER ( order by HighestScore DESC ) as FinalRank
Results
ContestantID personName FinalRank
3 Jane 1
1 Bill 2
2 Mary 2
4 Mark 3
Now, if you only wanted a limit such as the top 3 ranks and you actually had 20+ competitors, just wrap this up one more time where a where clase
select * from
(
select
ContestantID,
c.personName,
DENSE_RANK() OVER ( order by HighestScore DESC ) as FinalRank
from
(select
contestantid,
max( score ) highestScore
from
attempts
group by
contestantid ) preAgg
JOIN Contestants c
on preAgg.contestantid = c.id ) dr
where
dr.FinalRank < 3

Answer 2

MY Reference table
MYSQL SOLUTION
using multiple common table expressions , dense_rank , Join and group_concat
WITH t1 AS
(SELECT *,DENSE_RANK() OVER(PARTITION BY event_id ORDER BY score DESC) AS 'rk' FROM Scoretable),
t2 AS
(SELECT * FROM t1 WHERE rk<=3),
t3 AS
(SELECT event_id , CASE WHEN rk=1 THEN p_Name ELSE NULL END AS 'first' FROM t2 WHERE rk=1 ),
t4 AS
(SELECT event_id , CASE WHEN rk=2 THEN p_Name ELSE NULL END AS 'second' FROM t2 WHERE rk=2 ),
t5 AS
(SELECT event_id , CASE WHEN rk=3 THEN p_Name ELSE NULL END AS 'third' FROM t2 WHERE rk=3 ),
t6 AS
(SELECT t3.event_id , t3.first , t4.second , t5.third FROM t3 JOIN t4 ON t3.event_id = t4.event_id JOIN t5 ON t4.event_id=t5.event_id ORDER BY 1,2,3,4)
SELECT event_id , GROUP_CONCAT(DISTINCT first) AS 'rank 1' , GROUP_CONCAT(DISTINCT second) AS 'rank 2' , GROUP_CONCAT(DISTINCT third) AS 'rank 3'
FROM t6 GROUP BY 1 ORDER BY 1;