SQL count exposure of life time by age - sql

Question

(Using SQL Server 2008)
I need some help visualizing a solution. Let's say I have the following simple table for members of a pension scheme:
[Date of Birth] [Date Joined] [Date Left]
1970/06/1 2003/01/01 2007/03/01
I need to calculate the number of lives in each age group from 2000 to 2009.
NOTE: "Age" is defined as "age last birthday" (or "ALB") on 1 January of each of those yeasrs. e.g. if you are exactly 41.35 or 41.77 etc. years old on 1/1/2009 then you would be ALB 41.
So if the record above were the only entry in the database, then the output would be something like:
[Year] [Age ] [Number of Lives]
2003 32 1
2004 33 1
2005 34 1
2006 35 1
2007 36 1
(For 2000, 2001, 2002, 2008 and 2009 there are no lives on file since the sole member only joined on 1/1/2003 and left on 1/3/2007)
I hope I am making myself clear enough.
Anyone have any suggestions?
Thanks, Karl
[EDIT]
Adding another layer to the problem:
What if I had:
[Date of Birth] [Date Joined] [Date Left] [Gender] [Pension Value]
1970/06/1 2003/01/01 2007/03/01 'M' 100,000
and I want the output to be:
[Year] [Age ] [Gender] sum([Pension Value]) [Number of Lives]
2003 32 M 100,000 1
2004 33 M 100,000 1
2005 34 M 100,000 1
2006 35 M 100,000 1
2007 36 M 100,000 1
Any ideas?

Answer 1

WITH years AS
(
SELECT 1900 AS y
UNION ALL
SELECT y + 1
FROM years
WHERE y < YEAR(GETDATE())
),
agg AS
(
SELECT YEAR(Dob) AS Yob, YEAR(DJoined) AS YJoined, YEAR(DLeft) AS YLeft
FROM mytable
)
SELECT y, y - Yob, COUNT(*)
FROM agg
JOIN years
ON y BETWEEN YJoined AND YLeft
GROUP BY
y, y - Yob
OPTION (MAXRECURSION 0)
People born on same year always have the same age in your model
That's why if they go at all, they always go into one group and you just need to generate one row per year for the period they stay in the program.

Answer 2

You can try something like this
DECLARE #Table TABLE(
[Date of Birth] DATETIME,
[Date Joined] DATETIME,
[Date Left] DATETIME
)
INSERT INTO #Table ([Date of Birth],[Date Joined],[Date Left]) SELECT '01 Jun 1970', '01 Jan 2003', '01 Mar 2007'
INSERT INTO #Table ([Date of Birth],[Date Joined],[Date Left]) SELECT '01 Jun 1979', '01 Jan 2002', '01 Mar 2008'
DECLARE #StartYear INT,
#EndYear INT
SELECT #StartYear = 2000,
#EndYear = 2009
;WITH sel AS(
SELECT #StartYear YearVal
UNION ALL
SELECT YearVal + 1
FROM sel
WHERE YearVal < #EndYear
)
SELECT YearVal AS [Year],
COUNT(Age) [Number of Lives]
FROM (
SELECT YearVal,
YearVal - DATEPART(yy, [Date of Birth]) - 1 Age
FROM sel LEFT JOIN
#Table ON DATEPART(yy, [Date Joined]) <= sel.YearVal
AND DATEPART(yy, [Date Left]) >= sel.YearVal
) Sub
GROUP BY YearVal

Answer 3

Try the following sample query
SET NOCOUNT ON
Declare #PersonTable as Table
(
PersonId Integer,
DateofBirth DateTime,
DateJoined DateTime,
DateLeft DateTime
)
INSERT INTO #PersonTable Values
(1, '1970/06/10', '2003/01/01', '2007/03/01'),
(1, '1970/07/11', '2003/01/01', '2007/03/01'),
(1, '1970/03/12', '2003/01/01', '2007/03/01'),
(1, '1973/07/13', '2003/01/01', '2007/03/01'),
(1, '1972/06/14', '2003/01/01', '2007/03/01')
Declare #YearTable as Table
(
YearId Integer,
StartOfYear DateTime
)
insert into #YearTable Values
(1, '1/1/2000'),
(1, '1/1/2001'),
(1, '1/1/2002'),
(1, '1/1/2003'),
(1, '1/1/2004'),
(1, '1/1/2005'),
(1, '1/1/2006'),
(1, '1/1/2007'),
(1, '1/1/2008'),
(1, '1/1/2009')
;WITH AgeTable AS
(
select StartOfYear, DATEDIFF (YYYY, DateOfBirth, StartOfYear) Age
from #PersonTable
Cross join #YearTable
)
SELECT StartOfYear, Age, COUNT (1) NumIndividuals
FROM AgeTable
GROUP BY StartOfYear, Age
ORDER BY StartOfYear, Age

Answer 4

First some preparation to have something to test with:
CREATE TABLE People (
ID int PRIMARY KEY
,[Name] varchar(50)
,DateOfBirth datetime
,DateJoined datetime
,DateLeft datetime
)
go
-- some data to test with
INSERT INTO dbo.People
VALUES
(1, 'Bob', '1961-04-02', '1999-01-01', '2007-05-07')
,(2, 'Sadra', '1960-07-11', '1999-01-01', '2008-05-07')
,(3, 'Joe', '1961-09-25', '1999-01-01', '2009-02-11')
go
-- helper table to hold years
CREATE TABLE dimYear (
CalendarYear int PRIMARY KEY
)
go
-- fill-in years for report
DECLARE
#yr int
,#StartYear int
,#EndYear int
SET #StartYear = 2000
SET #EndYear = 2009
SET #yr = #StartYear
WHILE #yr <= #EndYear
BEGIN
INSERT INTO dimYear (CalendarYear) values(#yr)
SET #yr =#yr+1
END
-- show test data and year tables
select * from dbo.People
select * from dbo.dimYear
go
Then a function to return person's age for each year, if the person is still an active member.
-- returns [CalendarYear], [Age] for a member, if still active member in that year
CREATE FUNCTION dbo.MemberAge(#DateOfBirth datetime, #DateLeft datetime)
RETURNS TABLE
AS
RETURN (
SELECT
CalendarYear,
CASE
WHEN DATEDIFF(dd, cast(CalendarYear AS varchar(4)) + '-01-01',#DateLeft) > 0
THEN DATEDIFF(yy, #DateOfBirth, cast(CalendarYear AS varchar(4)) + '-01-01')
ELSE -1
END AS Age
FROM dimYear
);
go
And the final query:
SELECT
a.CalendarYear AS "Year"
,a.Age AS "Age"
,count(*) AS "Number Of Lives"
FROM
dbo.People AS p
CROSS APPLY dbo.MemberAge(p.DateOfBirth, p.DateLeft) AS a
WHERE a.Age > 0
GROUP BY a.CalendarYear, a.Age

Answer 5

Deal with this in pieces (some random thoughts) - create views to test you dev steps if you can:
ALB - do a query that, for a given year, gives you your memeber's ALB
Member in year - another bit of query that tell you whether a member was a member in a given year
Put those two together and you should be able to create a query that says whether a person was a member in a given year and what their ALB was for that year.
Hmm, tricky - following this chain of thought what you'd then want to do is generate a table that has all the years the person was a member and their ALB in that year (and a unique id)
From 4. select year, alb, count(id) group by year, alb
I'm not sure I'm going in the right direction from about 3 though it should work.
You may find a (temporary) table of years helpful - joining things to a table of dates makes all kinds of things possible.
Not really an answer, but certainly some direction...