CODEWITHSUNDEEP
mysql
Red Baron
Red Baron

Reputation: 63

Select n random rows per specified group

I am struggling to find an optimal solution for the following problem.

Suppose I have a table 'Table' like this:

id    name    report_id
 1    name1    1
 2    name2    3
 3    name3    5
 4    name1    7
 5    name3    8
 ....................

I want to select for each value in a set: ('name1', 'name2') 10 random unique rows.

Of course it is possible to do with union like:

(SELECT * FROM Table
WHERE
    name='name1'
ORDER BY RAND() LIMIT 10)
UNION
(SELECT * FROM Table
WHERE
    name='name2'
ORDER BY RAND() LIMIT 10)

But if I have 100 unique names for which I have to select 10 random records - this query is going to be a bit large.

Upvotes: 4

Views: 4696

Answers (3)

hertopnerd
hertopnerd

Reputation: 648

This doesn't work in MySQL, but in PostgreSQL you can use partition by

select name,report_id from
(select name,report_id,row_number()
over
(partition by name order by random())
as rn from Table) a
where rn<=10

I had this same question and found this answer from a colleague.

Upvotes: 1

valex
valex

Reputation: 24144

SQLFiddle demo

select ID,NAME,REPORT_ID
from
(
select *, @row:=if(name=@name,@row,0)+1 as rn, @name:=name from 
(select *,RAND() as trand from t) t1,
(select @row:=0,@name:='') tm2 
order by name,trand
) t2
where rn<=10

Upvotes: 5

Hamlet Hakobyan
Hamlet Hakobyan

Reputation: 33391

Try this:

  SELECT
    id, 
    name,
    report_id
  FROM
  (
    SELECT id,
        report_id,
        name,
        CASE WHEN @name != name THEN @rn := 1 ELSE @rn := @rn + 1 END rn,
        @name:=name
     FROM (SELECT * FROM tbl ORDER BY RAND()) a,
        (SELECT @rn:=0, @name := NULL) r
     ORDER BY name
  ) s
  WHERE rn <= 10;

SQL FIDDLE DEMO

Upvotes: 3

Related Questions