In the same column, compare each value with previous multiple values with condition - sql

Question

I'm working on a table looks like this. The actual dataset contains thousand of Guest_ID, and I'm just showing few sample lines here.
Guest_ID
Visit_ID
Collection_time
Value
6
a178
2007-11-09 11:28:00
2.6
6
a188
2007-11-10 20:28:00
6.6
12
a278
2008-11-11 10:28:00
2.7
12
a278
2008-11-11 11:38:00
3.2
12
a278
2008-11-12 11:48:00
6.8
12
c348
2009-10-12 11:38:00
3.8
15
e179
2013-01-15 09:25:00
1.8
15
e179
2013-01-15 10:26:00
1.6
15
e179
2013-01-15 12:15:00
3.8
15
e179
2013-01-17 09:25:00
3.6
What I'm trying to do here is to find out the values that had increased by at least 3 within the past 48hr, and these values need to be under the same visit_id. In this case, result should only return
Guest_ID
Visit_ID
Collection_time
Value
12
a278
2008-11-12 11:48:00
6.8
I have some vague ideas of creating islands and gaps in SQL Server, but not sure how to approach it. Conceptually, for each value X, I need to extract all the previous value meets the conditions (within past 48hr AND under the same Visit_ID), then check if X - min(previous value) >= 3. And if yes, keep or label X as 1, and repeat the procedure.
I read a lot of posts like using lag() or row_number() over (partition by ... order by ...), but still unsure about what to do. Any help is appreciated!

Answer 1

This would have been a good spot to use window functions with a date range specification. Alas, SQL Server does not support that (yet?).
The simplest approach might be exists and a correlated subquery:
select t.*
from mytable t
where exists (
select 1
from mytable t1
where
t1.visit_id = t.visit_id
and t1.collection_time >= dateadd(day, -2.collection_time)
and t1.collection_time < t.collection_time
and t1.value < t.value - 3
)
Or you can use cross apply:
select t.*
from mytable t
cross apply (
select min(t1.value) as min_value
from mytable t1
where
t1.visit_id = t.visit_id
and t1.collection_time >= dateadd(day, -2.collection_time)
and t1.collection_time < t.collection_time
) t1
where t1.min_value < t.value - 3

Answer 2

I used a CTE to filter out the qualifying rows first and then just join it back up to the original table to grab those rows:
CREATE TABLE #tmp(Guest_ID int, Visit_ID varchar(10), Collection_time datetime, Value decimal(10,1))
INSERT INTO #tmp VALUES
(6, 'a178', '2007-11-09 11:28:00', 2.6),
(6, 'a188', '2007-11-10 20:28:00', 6.6),
(12, 'a278', '2008-11-11 10:28:00', 2.7),
(12, 'a278', '2008-11-11 11:38:00', 3.2),
(12, 'a278', '2008-11-12 11:48:00', 6.8),
(12, 'c348', '2009-10-12 11:38:00', 3.8),
(15, 'e179', '2013-01-15 09:25:00', 1.8),
(15, 'e179', '2013-01-15 10:26:00', 1.6),
(15, 'e179', '2013-01-15 12:15:00', 3.8),
(15, 'e179', '2013-01-17 09:25:00', 3.6)
;WITH CTE AS(
SELECT MAX(Collection_time) MaxCollection_Time, Max(Value) - Min(Value) DiffInValue ,Visit_ID
FROM #tmp
GROUP BY Visit_ID
HAVING Max(Value) - Min(Value) >= 3
)
SELECT t1.*
FROM #tmp t1
INNER JOIN CTE t2 on t1.Visit_ID = t2.Visit_ID and T1.Collection_time = t2.MaxCollection_Time