Using Aggregate and Max functions - sql

Question

I'm having problems with the below query, working in SQL Server.
SELECT
emp_id= CASE employee_id
WHEN ''
THEN RTRIM(last_name) + '_' + RTRIM(first_name)
+ '_' + RTRIM(gender) + '_'
+ RTRIM(race_ethnicity_code) + '_'
+ RTRIM(high_degree_code) + '_' + RTRIM(position_code) + '_'
+ RTRIM(assignment_code)
ELSE employee_id
END ,
last_name, first_name,
assign_perc,
assignment_num,
CAST((total_salary)AS NUMERIC (18,2))* CAST((assign_perc) AS NUMERIC (18,2)) AS salary,
total_salary
FROM employee
ORDER BY last_name, first_name, district_name
My script is a simple extraction of columns, creating a unique key through case statement for emp_id when its null. The problem im having is when multiplying the assign_perc with the total_Salary when that person has multiple assignments and taking the max salary when that sales person is only listed once. For example - my expected results:
John Smith is only a part time worker having one assignment being listed only one time, so his assign_perc will be less than 1 but I still need the max salary rather than the aggregate total (assign_perc*total_salary). Thank you for your help.

Answer 1

As noted above, this is excessively messy since you'll need to use the CASE statement to determine the virtual employee_id. It would be much simpler if you can either refactor this CASE statement into a udf, or store the result in the table for goodness sake!
To clarify what this is going to do - you want to join the employee table on a table containing counts for each employee. The count table would look like this:
SELECT employee_id, COUNT(*) AS employee_count
FROM employee
GROUP BY employee_id
Joining them together would look like this:
SELECT ...
FROM employee
JOIN (SELECT employee_id, COUNT(*) AS employee_count
FROM employee
GROUP BY employee_id) ec
ON employee.employee_id = ec.employee_id
Your calculated salary would now become:
CASE
WHEN ec.employee_count > 1
THEN CAST((total_salary)AS NUMERIC (18,2))* CAST((assign_perc) AS NUMERIC (18,2))
ELSE total_salary
END AS salary,
And here is the complete query substituting the monstrous CASE statement in place of 'employee_id':
SELECT
CASE employee.employee_id
WHEN ''
THEN RTRIM(last_name) + '_' + RTRIM(first_name)
+ '_' + RTRIM(gender) + '_'
+ RTRIM(race_ethnicity_code) + '_'
+ RTRIM(high_degree_code) + '_' + RTRIM(position_code) + '_'
+ RTRIM(assignment_code)
ELSE employee.employee_id
END AS emp_id,
last_name, first_name,
assign_perc,
assignment_num,
CASE
WHEN ec.employee_count > 1
THEN CAST((total_salary)AS NUMERIC (18,2))* CAST((assign_perc) AS NUMERIC (18,2))
ELSE total_salary
END AS salary,
total_salary
FROM employee
JOIN (SELECT CASE employee.employee_id
WHEN ''
THEN RTRIM(last_name) + '_' + RTRIM(first_name)
+ '_' + RTRIM(gender) + '_'
+ RTRIM(race_ethnicity_code) + '_'
+ RTRIM(high_degree_code) + '_' + RTRIM(position_code) + '_'
+ RTRIM(assignment_code)
ELSE employee.employee_id
END AS employee_id, COUNT(*) employee_count
FROM employee
GROUP BY CASE employee.employee_id
WHEN ''
THEN RTRIM(last_name) + '_' + RTRIM(first_name)
+ '_' + RTRIM(gender) + '_'
+ RTRIM(race_ethnicity_code) + '_'
+ RTRIM(high_degree_code) + '_' + RTRIM(position_code) + '_'
+ RTRIM(assignment_code)
ELSE employee.employee_id
END) ec
ON CASE employee.employee_id
WHEN ''
THEN RTRIM(last_name) + '_' + RTRIM(first_name)
+ '_' + RTRIM(gender) + '_'
+ RTRIM(race_ethnicity_code) + '_'
+ RTRIM(high_degree_code) + '_' + RTRIM(position_code) + '_'
+ RTRIM(assignment_code)
ELSE employee.employee_id
END = ec.employee_id
ORDER BY last_name, first_name, district_name