Analytic functions and plain SQL equivalent - sql

Question

I'm using Oracle and SQL Developer. I have downloaded HR schema and need to do some queries with it. Now I'm working with table Employees. As an user I need to see the list of employees with lowest salary in each department. I need to provide different solutions by means of plain SQL and one of analytic functions. About analytic functions, I have used RANK():
SELECT *
FROM
(SELECT
employee_id,
first_name,
department_id,
salary,
RANK() OVER (PARTITION BY department_id
ORDER BY salary) result
FROM
employees)
WHERE
result = 1
AND department_id IS NOT NULL;
The result seems correct:
but when I try to use plain SQL I actually get all employees with their salaries.
Here is my attempt with GROUP BY:
SELECT
department_id, MIN(salary) AS "Lowest salary"
FROM
employees
GROUP BY
department_id;
This code seems good, but I need to also get columns first_name and employee_id.
I tried to do something like this:
SELECT
employee_id,
first_name,
department_id,
MIN(salary) result
FROM
employees
GROUP BY
employee_id,
first_name,
department_id;
and this:
SELECT
employee_id,
first_name,
salary,
departments.department_id
FROM
employees
LEFT OUTER JOIN
departments ON (employees.department_id = departments.department_id)
WHERE
employees.salary = (SELECT MIN(salary)
FROM departments
WHERE department_id = employees.department_id)
These seem wrong. How can I change or modify my queries to get the same result as when I'm using RANK() by means of plain SQL (two solutions at least)?

Answer 1

One of the options could be like here (with old EMP table)...
SELECT EMPNO, ENAME, DEPTNO, SAL
FROM EMP e
WHERE SAL = (Select MIN_SAL From (SELECT DEPTNO, Min(SAL) "MIN_SAL"
FROM EMP
GROUP BY DEPTNO)
Where DEPTNO = e.DEPTNO)
ORDER BY DEPTNO, SAL;
Second option could be...
SELECT EMPNO, ENAME, DEPTNO, SAL
FROM (SELECT e.EMPNO, e.ENAME, e.DEPTNO, e. SAL, (Select Min(SAL) "MIN_SAL" From EMP Where DEPTNO = e.DEPTNO) "MIN_SAL" From EMP e)
WHERE SAL = MIN_SAL
ORDER BY DEPTNO, SAL;
Regards...

Answer 2

You can use a subquery to find the lowest salary per employee and use the main query to only show the information of those employees that are selected by this subquery:
SELECT
employee_id,
first_name,
department_id,
salary
FROM employees e1
WHERE salary =
(SELECT MIN(e2.salary)
FROM employees e2
WHERE e1.employee_id = e2.employee_id);
This will produce exactly the same outcome as your query with RANK.
I think it would make sense to apply some sorting which is missing in your query. I don't know how you want to sort, but here an example to sort by the employee's name:
SELECT
employee_id,
first_name,
department_id,
salary
FROM employees e1
WHERE salary =
(SELECT MIN(e2.salary)
FROM employees e2
WHERE e1.employee_id = e2.employee_id)
ORDER BY first_name;
Since you asked for at least two solutions, let's have a look on another option:
SELECT
e1.employee_id,
e1.first_name,
e1.department_id,
e1.salary
FROM employees e1
JOIN (
SELECT employee_id, MIN(salary) salary
FROM employees
GROUP BY employee_id ) e2
ON e1.employee_id = e2.employee_id AND e1.salary = e2.salary
ORDER BY first_name;
As you can see, this differs since the sub query will apply a GROUP BY clause and it can be successfully executed as own query which is not possible for the sub query used in the previous query.
The JOIN to the main query will then make sure to get again the desired result.

Answer 3

Here are some options to get the employees with the minimum salary in their department:
With MIN (salary) OVER (...)
select employee_id, first_name, department_id, salary
from
(
select e.*, min(salary) over (partition by department_id) as min_sal
from employees e
)
where sal = min_sal;
With RANK and FETCH FIRST
select *
from employees
order by rank() over (partition by department_id order by salary)
fetch first row with ties;
With IN
select *
from employees
where (department_id, salary) in
(
select department_id, min(salary)
from employees
group by department_id
);
With NOT EXISTS
select *
from employees e
where not exists
(
select null
from employees other
where other.department_id = e.department_id
and other.salary < e.salary
);

Answer 4

If you will only ever have one person with the minimum salary per department then you can use KEEP:
SELECT department_id,
MIN(employee_id) KEEP (DENSE_RANK FIRST ORDER BY salary) AS employee_id,
MIN(first_name) KEEP (DENSE_RANK FIRST ORDER BY salary, employee_id) AS first_name,
MIN(salary) AS min_salary
FROM employees
GROUP BY department_id
Which, for the sample data:
CREATE TABLE employees (employee_id, department_id, first_name, salary) AS
SELECT 1, 1, 'Alice', 1000 FROM DUAL UNION ALL
SELECT 2, 1, 'Betty', 2000 FROM DUAL UNION ALL
SELECT 3, 2, 'Carol', 3000 FROM DUAL UNION ALL
SELECT 4, 2, 'Debra', 3000 FROM DUAL UNION ALL
SELECT 5, 2, 'Emily', 4000 FROM DUAL;
Outputs:
DEPARTMENT_ID
EMPLOYEE_ID
FIRST_NAME
MIN_SALARY
1
1
Alice
1000
2
3
Carol
3000
Note: this will not match Debra, even though she also has the lowest salary in department 2, as it will only find a single employee with the minimum salary and the minimum employee id.
If you can have multiple employees with the same minimum-per-department then you can use a correlated sub-query:
SELECT department_id,
employee_id,
first_name,
salary
FROM employees e
WHERE EXISTS(
SELECT 1
FROM employees x
WHERE e.department_id = x.department_id
HAVING MIN(x.salary) = e.salary
);
Which, for the sample data, outputs:
DEPARTMENT_ID
EMPLOYEE_ID
FIRST_NAME
SALARY
1
1
Alice
1000
2
3
Carol
3000
2
4
Debra
3000
Which does return Debra.
fiddle