29 Correlated Subqueries

1.  Explain the main difference between correlated and non-correlated subqueries ? 

As explained in Multiple-Row Subqueries Practice > Problem 3:

Correlated subquery is executed multiple times once for each intermediate result row from outer query And for each row of the intermediate results inner correlated subquery is executed, if the inner correlated subquery is satisfied, that row becomes part of final results from the whole query. A correlated subquery will get a candidate row from an outer query, execute the inner query using candidate row value, and use values from the inner query to qualify or disqualify the candidate row.

Non-correlated subquery: result/s of inner query are calculated first and reused throughout the execution of outer query.

2.  Write a query that lists the highest earners for each department. Include the last_name, department_id, and the salary for each employee.

WITH highs AS ( SELECT NVL(department_id,-1), MAX(salary) FROM employees GROUP BY NVL(department_id,-1))

SELECT last_name, department_id, salary

FROM employees

WHERE (NVL(department_id,-1), salary) IN (SELECT * FROM highs );

or

SELECT oe.last_name, oe.department_id, oe.salary

FROM employees oe

WHERE oe.salary = (SELECT MAX(ie.salary) FROM employees ie WHERE NVL(ie.department_id,-1) = NVL(oe.department_id,-1));

LAST_NAME	DEPARTMENT_ID	SALARY
King	90	24000
Whalen	10	4400
Higgins	110	12000
Abel	80	11000
Grant		7000
Mourgos	50	5800
Hunold	60	9000
Hartstein	20	13000

3.  Examine the following select statement and finish it so that it will return the last_name, department_id, and salary of employees who have at least one person reporting to them. So we are effectively looking for managers only. In the partially written SELECT statement, the WHERE clause will work as it is. It is simply testing for the existence of a row in the subquery.

SELECT (enter columns here)

FROM (enter table name here) outer

WHERE 'x' IN (SELECT 'x'

FROM (enter table name here) inner

WHERE inner(enter column name here) = inner(enter column name here)

Finish off the statement by sorting the rows on the department_id column.

SELECT outer.last_name, outer.department_id, outer.salary

FROM employees outer

WHERE outer.employee_id  IN (SELECT DISTINCT inner.manager_id

FROM employees  inner

WHERE inner.manager_id = outer.employee_id)

ORDER BY outer.department_id;

OR

SELECT outer.last_name, outer.department_id, outer.salary

FROM employees outer

WHERE outer.employee_id IN (SELECT DISTINCT inner.manager_id

FROM employees  inner

WHERE inner.manager_id IS NOT NULL)

ORDER BY outer.department_id;

4.  Using a WITH clause, write a SELECT statement to list the job_title of those jobs whose maximum salary is more than half the maximum salary of the entire company. Name your subquery MAX_CALC_SAL. Name the columns in the result JOB_TITLE and JOB_TOTAL, and sort the result on JOB_TOTAL in descending order.

Hint: Examine the jobs table. You will need to join JOBS and EMPLOYEES to display the job_title.

If I ignore the hint and job_total, which is a very ambiguous requirement here:

WITH max_calc_sal as (SELECT MAX(max_salary)/2 FROM jobs)

SELECT job_title

FROM jobs

WHERE jobs.max_salary > (SELECT * FROM max_calc_sal );

JOB_TITLE

President

Administration Vice President

But if I consider hints, I think might be job_total is not a total, but should have been called job_actual_max. Now there is a possibility that a job is there in job table but no employee is there for that job. For such case actual max could be 0.

WITH max_calc_sal AS (SELECT jobs.job_id , jobs.job_title, MAX(NVL(employees.salary,0)) AS job_actual_max FROM employees RIGHT OUTER JOIN   jobs  ON  employees.job_id =  jobs.job_id    GROUP BY jobs.job_id,jobs.job_title)

SELECT job_title, job_actual_max AS  job_total

FROM max_calc_sal

WHERE job_actual_max > (SELECT MAX(job_actual_max)/2 FROM max_calc_sal)

ORDER BY job_total  DESC;

JOB_TITLE	JOB_TOTAL
President	24000
Administration Vice President	17000
Marketing Manager	13000