Subtracting rows in SQL Server - sql

Question

I have a dataset like this:
ID | IssueDate
194924 | 2013-07-31 00:00:00.000
194924 | 2010-06-15 00:00:00.000
194924 | 2012-07-30 00:00:00.000
194924 | 2012-12-11 00:00:00.000
194924 | 2014-08-04 00:00:00.000
194966 | 2012-06-02 00:00:00.000
194966 | 2011-02-03 00:00:00.000
194966 | 2011-02-01 00:00:00.000
194987 | 2013-04-25 00:00:00.000
194987 | 2010-12-03 00:00:00.000
I want to sort data with ID and IssueDate first, and then subtract IssueDates of two consecutive rows (to find the time between one row and next row), then calculate max, min and average of this times for each unique ID.

Answer 1

If your Sql Server version is 2014 then the below one might be help you.
Schema for your case:
CREATE TABLE #TAB (
ID BIGINT
,IssuDate DATETIME
)
INSERT INTO #TAB
SELECT 194924
,'2013-07-31 00:00:00.000'
UNION ALL
SELECT 194924
,'2010-06-15 00:00:00.000'
UNION ALL
SELECT 194924
,'2012-07-30 00:00:00.000'
UNION ALL
SELECT 194924
,'2012-12-11 00:00:00.000'
UNION ALL
SELECT 194924
,'2014-08-04 00:00:00.000'
UNION ALL
SELECT 194966
,'2012-06-02 00:00:00.000'
UNION ALL
SELECT 194966
,'2011-02-03 00:00:00.000'
UNION ALL
SELECT 194966
,'2011-02-01 00:00:00.000'
UNION ALL
SELECT 194987
,'2013-04-25 00:00:00.000'
UNION ALL
SELECT 194987
,'2010-12-03 00:00:00.000'
Result after sorting and finding the Time difference:
SELECT *, DATEDIFF(DD, ISNULL(LAG(ISSUDATE) OVER(PARTITION BY ID ORDER BY ID,IssuDate ), IssuDate),IssuDate) AS TIME_DIFF_IN_DAYS
FROM #TAB
For aggregation with min Max & avg
SELECT ID, MIN(TIME_DIFF_IN_DAYS) AS MIN_TIME_TAKEN, MAX(TIME_DIFF_IN_DAYS) MAX_TIME_TAKEN, AVG(TIME_DIFF_IN_DAYS) AVG_TIME_TAKEN FROM (
SELECT *, DATEDIFF(DD, ISNULL(LAG(ISSUDATE) OVER(PARTITION BY ID ORDER BY ID,IssuDate ), IssuDate),IssuDate) AS TIME_DIFF_IN_DAYS FROM #TAB
)AS A
WHERE TIME_DIFF_IN_DAYS>0 --This one you can comment if you want to show 0 diffence in time
GROUP BY ID

Answer 2

I am not sure about "and c1.id=c.id" in CTE1 coz I am not sure about your exact requrement. Neverthless you can try some thing like,
declare #t table(ID int,IssuDate datetime)
insert into #t values
(194924,'2013-07-31 00:00:00.000')
,(194924,'2010-06-15 00:00:00.000')
,(194924,'2012-07-30 00:00:00.000')
,(194924,'2012-12-11 00:00:00.000')
,(194924,'2014-08-04 00:00:00.000')
,(194966,'2012-06-02 00:00:00.000')
,(194966,'2011-02-03 00:00:00.000')
,(194966,'2011-02-01 00:00:00.000')
,(194987,'2013-04-25 00:00:00.000')
,(194987,'2010-12-03 00:00:00.000')
;with CTE as
(select *,ROW_NUMBER()over(order by id,IssuDate)rn
from #t
)
,Cte1 as
(
select *
,(select datediff(second,c.IssuDate,c1.IssuDate) from CTE c1 where c1.rn=c.rn+1 and c1.id=c.id)Time_between
from CTE C
)
select sum(Time_between),min(Time_between),avg(Time_between),max(Time_between) from cte1
group by id