create view selecting the max in one column related with another column - sql

Question

I need to create a view in PostgreSQL 9.4 about this table:
CREATE TABLE DOCTOR (
Doc_Number INTEGER,
Name VARCHAR(50) NOT NULL,
Specialty VARCHAR(50) NOT NULL,
Address VARCHAR(50) NOT NULL,
City VARCHAR(30) NOT NULL,
Phone VARCHAR(10) NOT NULL,
Salary DECIMAL(8,2) NOT NULL,
DNI VARCHAR(10) NOT NULL,
CONSTRAINT pk_Doctor PRIMARY KEY (Doc_Number)
);
The view will show the rank of the doctors with highest salary for each specialty, I tried this code but it shows all of the doctors fro each specialty:
CREATE VIEW top_specialty_doctors
AS (Select MAX(Salary), name, specialty from DOCTOR
where specialty = 'family and community'
or specialty = 'psychiatry'
or specialty = 'Rheumatology'
group by name, salary, specialty);
How can I do for the view shows only the doctor with highest salary for each specialty.

Answer 1

DISTINCT ON is a simple Postgres specific technique to get one winner per group. Details:
Select first row in each GROUP BY group?
CREATE VIEW top_specialty_doctors AS
SELECT DISTINCT ON (specialty)
salary, name, specialty
FROM doctor
WHERE specialty IN ('family and community', 'psychiatry', 'Rheumatology')
ORDER BY specialty, salary DESC, doc_number -- as tiebreaker
And you do not need parentheses around the query for CREATE VIEW.
If multiple docs tie for the highest salary, the one with the smallest doc_number is selected.
If salary can be NULL, use DESC NULLS LAST:
PostgreSQL sort by datetime asc, null first?
For big tables and certain data distributions other query techniques are superior:
Optimize GROUP BY query to retrieve latest record per user

Answer 2

Here's a query that shows the best doctor by salary for each of the specialties:
with specialty_ranks as (
select
Salary, name, specialty,
rank() over (
partition by specialty
order by salary desc
) as rank
from DOCTOR
where specialty in ('family and community', 'psychiatry', 'Rheumatology')
)
select specialty, name, salary
from specialty_ranks
where rank = 1;
The query uses a CTE and the RANK() window function to do the job. You might want to read their docs if you haven't used them before.

Answer 3

Without using Common Table Expressions or analytics, you can use an inline view/virtual table:
Create View top_specialty_doctors as
Select m.MaxSalary, d.Name, d.Specialty
From Doctor d
Join( -- Expose largest salary of each specialty
Select Specialty, Max( Salary) as MaxSalary
From Doctor
Group by Specialty
) as m
on m.Specialty = d.Specialty
and m.MaxSalary = d.Salary
Where specialty in( 'family and community', 'psychiatry', 'Rheumatology' );
Using a CTE instead of the inline view makes the query more readable and allows the query optimizer to turn out better performance (usually). They are really easy to learn.