SQL Calculate a Total Time in a specific State - sql-server

Question

I have these datas in a Table
_____DateTime____|Variable__|Value
2017/03/29 23:00:00 | Variable1 | 1
2017/03/31 01:00:00 | Variable1 | 0
2017/03/31 02:00:00 | Variable1 | 1
2017/03/31 03:00:00 | Variable1 | 0
2017/03/31 04:00:00 | Variable2 | 1
2017/03/31 23:00:00 | Variable1 | 1
2017/04/01 01:00:00 | Variable1 | 0
And I would like to calculate the total duration where each variable was in state 1 between two date
For example between for Var1 2017/03/31 00:00:00 and 2017/04/01 00:00:00
The result is :
1 hour between 2017/03/31 00:00:00 and 2017/03/31 01:00:00
1 hour between 2017/03/31 02:00:00 and 2017/03/31 03:00:00
1 hour between 2017/03/31 23:00:00 and 2017/04/01 00:00:00
So the result I want for Var1 should be 3 hours
For example between for Var2 2017/03/31 00:00:00 and 2017/04/01 00:00:00
The result is :
1 hour between 2017/03/31 04:00:00 and 2017/04/01 00:00:00 (no value before but because it change to 1 I suppose that it was 0 before)
So the result I want for Var2 should be 20 hours
Variable|__Time in Value (second)
Variable1 | 180
Variable2 | 1200
If someone can help me.
Thanks in advance

Answer 1

For SQL Server 2012+ (because of lead() and concat())
Using a stacked cte to generate an hours table to inner join a subquery that uses the lead() window function to get the next date for status change partitioned by Variable.
To adapt for prior versions, use an outer apply() to get the next dt for each variable instead of lead(); and regular string concatenation with proper conversions instead of concat().
declare #fromdate datetime = '20170331 00:00:00';
declare #thrudate datetime = '20170401 00:00:00';
;with n as (select n from (values(0),(1),(2),(3),(4),(5),(6),(7),(8),(9)) t(n))
, hours as (
select top ((datediff(hour, #fromdate, #thrudate)+1))
[DateHour]=dateadd(hour,(row_number() over (order by (select 1)) -1),#fromdate)
from n as deka cross join n as hecto cross join n as kilo
cross join n as tenK cross join n as hundredK
order by 1
)
select variable, value
, hours = count(h.datehour)
, start_dt = convert(varchar(20),min(h.datehour),120)
, end_dt = convert(varchar(20),end_dt,120)
, txt = concat(
count(h.datehour),' '
, case when count(h.datehour) < 2 then 'hour' else 'hours' end
, ' between '
, convert(varchar(20),min(h.datehour),120)
, ' and '
, convert(varchar(20),end_dt,120)
)
from hours h
inner join (
select
variable
, value
, start_dt = dt
, end_dt = case when coalesce(lead(dt) over (partition by variable order by dt),#thrudate) > #thrudate
then #thrudate
else coalesce(lead(dt) over (partition by variable order by dt),#thrudate)
end
from t
) s
on h.datehour >= s.start_dt
and h.datehour < s.end_dt
where h.datehour >= #fromdate
and h.datehour < #thrudate
and s.value = 1
group by variable, value, start_dt, end_dt
rextester demo: http://rextester.com/ZBWP22523
returns:
+-----------+-------+-------+---------------------+---------------------+--------------------------------------------------------------+
| variable | value | hours | start_dt | end_dt | txt |
+-----------+-------+-------+---------------------+---------------------+--------------------------------------------------------------+
| Variable1 | 1 | 1 | 2017-03-31 00:00:00 | 2017-03-31 01:00:00 | 1 hour between 2017-03-31 00:00:00 and 2017-03-31 01:00:00 |
| Variable1 | 1 | 1 | 2017-03-31 02:00:00 | 2017-03-31 03:00:00 | 1 hour between 2017-03-31 02:00:00 and 2017-03-31 03:00:00 |
| Variable1 | 1 | 1 | 2017-03-31 23:00:00 | 2017-04-01 01:00:00 | 1 hour between 2017-03-31 23:00:00 and 2017-04-01 01:00:00 |
| Variable2 | 1 | 20 | 2017-03-31 04:00:00 | 2017-04-01 00:00:00 | 20 hours between 2017-03-31 04:00:00 and 2017-04-01 00:00:00 |
+-----------+-------+-------+---------------------+---------------------+--------------------------------------------------------------+
If you need to do this often, you might consider creating an actual table for hours. Otherwise, using the stacked cte is as fast as most other options, and is much faster than a recursive cte as the number of values generated increases.
Number and Calendar table reference:
Generate a set or sequence without loops - 1 - Aaron Bertrand
Generate a set or sequence without loops - 2 - Aaron Bertrand
Generate a set or sequence without loops - 3 - Aaron Bertrand
The "Numbers" or "Tally" Table: What it is and how it replaces a loop - Jeff Moden
Creating a Date Table/Dimension in SQL Server 2008 - David Stein
Calendar Tables - Why You Need One - David Stein
Creating a date dimension or calendar table in SQL Server - Aaron Bertrand
TSQL Function to Determine Holidays in SQL Server - Tim Cullen
F_TABLE_DATE - Michael Valentine Jones