How to find max value in a row with changing - loops

Question

I have an dataset which has 12 different values for an ID, and also start and end Values. What I want to initilize is take the start value as my begining argument of loop and end value as the last argument. Search trough values accourding to them, and finding the maximum of them. After finding maximum search through values again wrt start and end value and find the longest consecutive max value occurance.
Below I posted an example dataset:
create table #sample_data(
ID VARCHAR(10), val1 INT, val2 INT, val3 INT, val4 INT, val5 INT,
val6 INT, val7 INT, val8 INT, val9 INT, val10 INT, val11 INT, val12 INT,
startValue INT, endValue INT );
insert into #sample_data values
(1001,3,2,1,0,1,2,3,0,0,0,0,0,1,7),
(1002,1,2,3,4,0,0,0,1,2,3,0,0,1,12),
(1003,0,3,2,1,0,0,0,3,3,0,0,0,1,12),
(1004,0,1,2,4,4,0,0,0,0,0,0,0,3,9),
(1005,1,2,2,1,0,0,2,2,2,1,0,0,1,8);
The result I expect for Id=1001 start=1, end = 7, max value is 3 and, it occurs 2 times but they arent consecutive, therefore final output I'd like to get is 1.
For ID=1002 start=1, end=12, max is 4 and it only occurs 1 time, so final output shoud be 1.
For ID = 1003 start=1, end=12, max is 3, 3 occurs three times but only 2 of them are consecutive therefore I expect to get 2.
For ID = 1004 start=3, end=9, max is 4 it occurs two times consecutively therefore output should be 2.
For ID = 1005 start=1, end=8, max is 2 it totaly occurs 5 times, 2 and 3 times consecutively, I expect to get 3 as my final output since it is longest.

Answer 1

If I understand the question correctly, the result for the row with Id 1005 should be 2 and not 3, because the max value (which is 2) appears consecutively in places 2,3 and then again in places 7,8,9 - but the endValue of that row is 8, and therefor the larger consecutive should not be counted.
Based on that understanding (which might be incorrect, hence the comment I've written to the question), this can be done with a set based approach (meaning, without any loops), with the help of some nice SQL tricks.
So the first thing you want to do is to use cross apply with a table value constructor to convert the val1...val12 columns to rows. I guess this can also be done using Pivot but I never quite got the hang of pivot so I prefer other solutions to get the same thing.
In my code, this step is done in the first common table expression (called CTEValues).
Next, you use a trick from Itzik Ben-Gan to handle gaps and island problems to identify the groups of consecutive values within each row. This step is done in the second cte (CTEGroups).
The third and final cte called CTEConsecutive use a simple group by and count to get the number of consecutive max values within each row of the original table, providing their column location is between the startValue and EndValue.
The last thing to do is get the max value of that count for each id - and that should give you the desired results.
Here's the full code:
WITH CTEValues AS
(
SELECT ID, startValue, EndValue, Val, ValId, IIF(Val = MAX(Val) OVER(PARTITION BY ID), 1, 0) As IsMax
FROM #sample_data
CROSS APPLY
(
SELECT *
FROM (VALUES
(Val1, 1),
(Val2, 2),
(Val3, 3),
(Val4, 4),
(Val5, 5),
(Val6, 6),
(Val7, 7),
(Val8, 8),
(Val9, 9),
(Val10, 10),
(Val11, 11),
(Val12, 12)
)V(Val, ValId)
) vals
), CTEGroups AS
(
SELECT ID, startValue, EndValue, Val, ValId, IsMax,
ROW_NUMBER() OVER(PARTITION BY ID ORDER BY ValId) -
ROW_NUMBER() OVER(PARTITION BY ID, IsMax ORDER BY ValId) As Grp
FROM CTEValues
), CTEConsecutive AS
(
SELECT ID, COUNT(Val) As NumOfConsecutiveMaxValues --*, OVER(PARTITION BY Id, Grp) As NumOfValues
FROM CTEGroups
WHERE IsMax = 1
AND ValId >= startValue
AND ValId <= EndValue
GROUP BY ID, Grp
)
SELECT ID, MAX(NumOfConsecutiveMaxValues)
FROM CTEConsecutive
GROUP BY ID
ORDER BY Id
You can see a live demo on rextester.
If, however, I'm wrong in my initial assumption and the startvalue and endvalue are only relevant to the range of which to search for the max value, (and that would give you the expected results you've posted in the question), you will need another cte.
WITH CTEValues AS
(
SELECT ID, startValue, EndValue, Val, ValId
FROM #sample_data
CROSS APPLY
(
SELECT *
FROM (VALUES
(Val1, 1),
(Val2, 2),
(Val3, 3),
(Val4, 4),
(Val5, 5),
(Val6, 6),
(Val7, 7),
(Val8, 8),
(Val9, 9),
(Val10, 10),
(Val11, 11),
(Val12, 12)
)V(Val, ValId)
) vals
), CTEValuesWithMax AS
(
SELECT ID, startValue, EndValue, Val, ValId,
IIF(Val = (
SELECT MAX(Val)
FROM CTEValues AS T1
WHERE T0.ID = T1.ID
AND T1.ValId >= T1.startValue
AND T1.ValId <= T1.EndValue
), 1, 0) As IsMax
FROM CTEValues AS T0
)
The rest of the code remains the same, except that CTEGroups now selects from CTEValuesWithMax instead of from CTEValues.
You can see a live demo of this as well.