Question

I who like to find a solution in T-SQL that could find a a way to detect a change in a given list or records.

The physical table is like this:

| id |val |
|----|----|
| 1  | A  |
|----|----|
| 2  | A  |
|----|----|
| 3  | B  |
|----|----|
| 4  | B  |
|----|----|
| 5  | A  |
|----|----|
| 6  | A  |
|----|----|

id is a sequencial integer

val is an arbitrary value

I would like to add an calculated field that could somehow denote a change of val

Desired result:

| id |val | segment |
|----|----|---------|
| 1  | A  | 1       |
|----|----|---------|
| 2  | A  | 1       |
|----|----|---------|
| 3  | B  | 2       |
|----|----|---------|
| 4  | B  | 2       |
|----|----|---------|
| 5  | A  | 3       |
|----|----|---------|
| 6  | A  | 3       |
|----|----|---------|

What I’m trying to do is the possibility to group by “segments” like this:

| from_id | to_id | val |
|---------|-------|-----|
|     1   |    2  |  A  |
|     3   |    4  |  B  |
|     5   |    6  |  A  |
|---------|-------|-----|

Answer 1

Assuming SQL Server 2005+

DECLARE @T TABLE (
id INT PRIMARY KEY,
val CHAR(1))

INSERT INTO @T
SELECT 1,'A' UNION ALL SELECT 2,'A' UNION ALL
SELECT 3,'B' UNION ALL SELECT 4,'B' UNION ALL
SELECT 5,'A' UNION ALL SELECT 6,'A'

;WITH cte1 AS(
SELECT 
      id,
      val, 
      ROW_NUMBER() OVER (ORDER BY id) - ROW_NUMBER() OVER (PARTITION BY val ORDER BY id) AS Grp
FROM @T
),
cte2 AS(
SELECT 
      id,
      val, 
      MIN(id) OVER (PARTITION BY Grp, val) AS GrpStart
FROM cte1
)
SELECT 
      id,
      val, 
      DENSE_RANK() OVER (ORDER BY GrpStart) AS segment 
FROM cte2

Or the updated requirement is a bit simpler

;WITH cte AS(
SELECT 
      id,
      val, 
      ROW_NUMBER() OVER (ORDER BY id) - ROW_NUMBER() OVER (PARTITION BY val ORDER BY id) AS Grp
FROM @T
)
SELECT 
      val, 
      MIN(id) AS from_id,
      MAX(id) AS to_id
FROM cte
GROUP BY Grp, val
ORDER BY from_id