SQL: Find top n in groupby data - mysql

Question

I have a table like this:
id | name | surname | city|
-------------------------------
'1', 'mohit', 'garg', 'delhi'
'2', 'mohit', 'gupta', 'delhi'
'3', 'ankita', 'gupta', 'jaipur'
'4', 'ankita', 'garg', 'jaipur'
'5', 'vivek', 'garg', 'delhi'
I am looking for a query that returns (id,city) grouped by city, with at most two (id) per city, but without using nested queries.
Expected output:
'1', 'delhi'
'2', 'delhi'
'3', 'jaipur'
'4', 'jaipur'

Answer 1

Perhaps the only way without subqueries is to use a trick with substring_index() and group_concat():
select city, substring_index(group_concat(id), ',', 2)
from t
group by city;
This puts the ids in a comma-delimited list, rather than in separate rows. Also, you have to be careful about the size of the intermediate results.
Of course, the accepted practice would use either a subquery in the where clause or a subquery using variables.
EDIT:
Here is a method for getting two ids per city without listing the cities:
select city, min(id) as id
from t
group by city
union
select city, max(id)
from t
group by city;

Answer 2

You can do this with a LEFT OUTER JOIN, although using a subquery will probably be clearer and might be faster. Here's a method using the JOIN:
SELECT
T1.id,
T1.city
FROM
My_Table T1
LEFT OUTER JOIN My_Table T2 ON T2.city = T1.city AND T2.id <= T1.id
GROUP BY
T1.id,
T1.city
HAVING
COUNT(*) <= 2
You're effectively finding all rows in T1 where the number of rows with the same name and a lower id is <= 2, which means that it must be one of the top two rows by id.

Answer 3

Try like below
create table
#test (id int, name varchar(10), name2 varchar(10),place varchar(10))
insert into #test
select
'1', 'mohit', 'garg', 'delhi'
union
select
'2', 'mohit', 'gupta', 'delhi'
union
select
'3', 'ankita', 'gupta', 'jaipur'
union
select
'4', 'ankita', 'garg', 'jaipur'
union
select
'5', 'vivek', 'garg', 'delhi'
with data
as
(
select ROW_NUMBER() OVER(PARTITION BY place ORDER BY id) RN,id,name,name2,place
from #test
),
data1
as(
select id, place
from data
where rn <=2
)
select *from data1