T-SQL - How to use a recursive calculation in a CTE? - tsql

Question

I want to calculate the avarage earnings per month of an employee during his time he worked for a company.
I build a recursive CTE based on a similar thread and I get the results only for one employee. What kind of modification do I need to do to make the employee ID variable and to get the same kind of results for every employee in the table?
Here is my code:
DECLARE #tbl AS TABLE
(
ID VARCHAR(50) ,
Earning FLOAT,
StartDate DATE ,
EndDate DATE
)INSERT INTO #tbl
( ID, Earning, StartDate, EndDate
SELECT employee_ID AS ID
,([Total Earning]/ (SELECT datediff(MONTH,[EndDate],
[StartDate])+1
FROM [employee_table] WHERE employee_ID = 'EKA-0004562'))
,[StarTDate]
,[EndDate]
FROM [employee_table]
WHERE employee_ID = 'EKA-0004562'
--final query using recursive cte
;
WITH cte
AS ( SELECT T.ID ,
T.Earning ,
T.StartDate ,
T.EndDatum ,
CONVERT(DATE, NULL) AS Dt ,
n = 0
FROM #tbl AS T
UNION ALL
SELECT cte.ID ,
cte.Earning ,
cte.StartDate ,
cte.EndDatum ,
DATEADD(MONTH, n, cte.StartDate) ,
cte.n + 1
FROM cte
WHERE n <= DATEDIFF(MONTH, cte.StartDate, cte.EndDatum)
)
SELECT cte.ID ,
cte.Earning,
dt AS Months
FROM cte
WHERE cte.Dt IS NOT NULL
This is my goal:
Well, I have a table with a lot of data with information concerning employees.
Every employee has a "startdate" (the time when he startet to work for the company) and and "enddate" (the time when he quit the job).
I would like to write into a table the same count of rows as the employee worked for the company in month. For example:
My basic table:
Employee Number | StartDate | EndDate | Total Earnings (Total Earnings/(EndDate-StartDate)
4711 20150101 20150523 24110
This example shows that the employee worked for the company for 5 Month and earned 24110€. In Avarage he earned 4822€ per month.
So I want to insert in the new table 5 rows with the following information:
New Table:
Employee Number | StartDate | EndDate | AVG Earnings
row1: 4711 20150101 20150523 4822
row2: 4711 20150201 20150523 4822
row3: 4711 20150301 20150523 4822
row4: 4711 20150401 20150523 4822
row5: 4711 20150501 20150523 4822

Answer 1

Here you go:
SAMPLE DATA
create table #test
(
ID INT,
StartDate DATE,
EndDatum DATE,
Earning INT
)
insert into #test values
(4711,'20150101','20150523',24110),
(4712,'20150101','20150625',32550)
QUERY
--final query using recursive cte
;
WITH cte
AS ( SELECT T.ID ,
T.Earning ,
T.StartDate ,
T.EndDatum ,
CONVERT(DATE, NULL) AS Dt ,
n = 0
FROM #test AS T
UNION ALL
SELECT cte.ID ,
cte.Earning ,
cte.StartDate ,
cte.EndDatum ,
DATEADD(MONTH, n, cte.StartDate) ,
cte.n + 1
FROM cte
WHERE n <= DATEDIFF(MONTH, cte.StartDate, cte.EndDatum)
)
SELECT cte.ID ,
dt StartDate,
cte.EndDatum EndDate,
cte.Earning / ((DATEDIFF(MONTH, cte.StartDate, cte.EndDatum)) +1) [AVG Earnings]
FROM cte
WHERE cte.Dt IS NOT NULL
OUTPUT
ID StartDate EndDate AVG Earnings
4712 2015-01-01 2015-06-25 5425
4712 2015-02-01 2015-06-25 5425
4712 2015-03-01 2015-06-25 5425
4712 2015-04-01 2015-06-25 5425
4712 2015-05-01 2015-06-25 5425
4712 2015-06-01 2015-06-25 5425
4711 2015-01-01 2015-05-23 4822
4711 2015-02-01 2015-05-23 4822
4711 2015-03-01 2015-05-23 4822
4711 2015-04-01 2015-05-23 4822
4711 2015-05-01 2015-05-23 4822

Answer 2

I may be wrong but I think your query seems overly complicated and the recursive CTE is not needed. It may be become inneficient when dealing with a lot of employees and consecutive years.
This query give the correct output for the 2 sample users:
Sample
declare #employee_table table(Employee_Number int, StartDate date, EndDate date, Total_Earnings bigint);
Insert Into #employee_table(Employee_Number, StartDate, EndDate, Total_Earnings) values
(4711, '20150101', '20150523', 24110)
, (4712, '20150101', '20150423', 24110);
Query
with list(n) as (
Select ROW_NUMBER() over(Order By n)
From (
Select 1 From (values(1), (1), (1), (1), (1), (1), (1), (1), (1), (1)) as l1(n)
Cross Join (values(1), (1), (1), (1), (1), (1), (1), (1), (1), (1)) as l2(n)
Cross Join (values(1), (1), (1), (1), (1), (1), (1), (1), (1), (1)) as l3(n)
) as l(n)
)
Select l.n, e.Employee_Number
, [StartDate] = DATEADD(MONTH, l.n-1, e.StartDate)
, [EndDate] = e.EndDate
, [AVG Earnings] = e.Total_Earnings/(DATEDIFF(MONTH, e.StartDate, e.EndDate)+1)
From #employee_table as e
Inner Join list as l on l.n <= DATEDIFF(MONTH, e.StartDate, e.EndDate)+1
Order By e.Employee_Number, l.n
Output
n Employee_Number StartDate EndDate AVG Earnings
1 4711 2015-01-01 2015-05-23 4822
2 4711 2015-02-01 2015-05-23 4822
3 4711 2015-03-01 2015-05-23 4822
4 4711 2015-04-01 2015-05-23 4822
5 4711 2015-05-01 2015-05-23 4822
1 4712 2015-01-01 2015-04-23 6027
2 4712 2015-02-01 2015-04-23 6027
3 4712 2015-03-01 2015-04-23 6027
4 4712 2015-04-01 2015-04-23 6027
notes:
#employee_table contains 2 employee. 4711 is similar to your sample
The list CTE efficiently creates a list from 1 to 1000 which should be good enough for 1000 months (> 83 years)
The main query uses it to create the list of consecutive months using just a simple INNER JOIN.