Only show one row per title in SQL query - sql

Question

I have the following query:
select distinct p.title, e.first_name, e.last_name, max(e.salary)
from employees as e
inner join employees_projects as ep
on e.id = ep.employee_id
inner join projects as p
on p.id = ep.project_id
group by 1,2,3
order by p.title
Which returns multiple rows per title. I only want the max salary for each title.
title | first_name | last_name | max
--------------------------+------------+-----------+-------
Build a cool site | Cailin | Ninson | 30000
Build a cool site | Ian | Peterson | 80000
Build a cool site | Mike | Peterson | 20000
Design 3 New Silly Walks | Ava | Muffinson | 10000
Update TPS Reports | John | Smith | 20000
Tweaked #zealous code and this works:
select
title,
first_name,
last_name,
salary
from
(select
distinct p.title,
e.first_name,
e.last_name,
e.salary,
dense_rank() over (partition by p.title order by e.salary desc) as rnk
from employees as e
inner join employees_projects as ep
on e.id = ep.employee_id
inner join projects as p
on p.id = ep.project_id
group by 1,2,3, 4
) t
where rnk = 1
order by title

Answer 1

Try this window function dense_rank(). If there is tie in salary then it will return both records with max salary.
If you just want one record with max salary then use row_number().
select
title,
first_name,
last_name,
salary
from
(select
distinct p.title,
e.first_name,
e.last_name,
salary,
dense_rank() over (partition by title order by salary desc) as rnk
from employees as e
inner join employees_projects as ep
on e.id = ep.employee_id
inner join projects as p
on p.id = ep.project_id
group by 1,2,3
) t
where rnk = 1
order by title

Answer 2

Assuming you want one row per title, then use distinct on:
select distinct on (p.title) p.title, e.first_name, e.last_name, e.salary
from employees e join
employees_projects ep
on e.id = ep.employee_id join
projects p
on p.id = ep.project_id
order by title, salary desc;
If you can have multiple titles, then you can use rank() or dense_rank() in a subquery:
select title, first_name, last_name, salary
from (select p.title, e.first_name, e.last_name, e.salary,
rank() over (partition by p.title order by e.salary desc) as seqnum
from employees e join
employees_projects ep
on e.id = ep.employee_id join
projects p
on p.id = ep.project_id
) p
where seqnum = 1
order by title;

Answer 3

This is basically the case when group by cant be used meaning you wanna group with no loss of all columns too.
You could simply use below
Select * from (
select distinct p.title, e.first_name,
e.last_name, e.salary
, rank()
Over(partition by 1 order by
e.salary desc)
) rn
from employees as e
inner join employees_projects as ep
on e.id = ep.employee_id
inner join projects as p
on p.id = ep.project_id)
Where rn=1