Question

Posting this again as I’ve made some updates.

I’m trying to group a set of values, but I only want to display the values if there was a change.

Imagine that the value is fluctuating, ranging from 1, to 4, to 3, to 1, to 1 and to 1 again.

Some example data

value | date_loaded | 
  1   |  2012-03-07 |
  1   |  2012-03-06 |
  1   |  2012-03-05 |
  3   |  2012-03-04 |
  4   |  2012-03-03 |
  1   |  2012-03-02 |

I want to display each of the values in order of which they fluctuated, but group them together. So you would only see in order, 1, 4, 3, 1, rather then the three 1’s most recently.

So I would like to display the latest value, with it’s earliest date, e.g.

value | date_loaded | 
  1   |  2012-03-05 |
  3   |  2012-03-04 |
  4   |  2012-03-03 |
  1   |  2012-03-02 |

What is the best way to go about this? Would it be possible to do an if statement? If value 1 is different to value 2, +1 to “change” ? Therefore I’d be able to group the values by “change1”, “change2” etc.?

Answer 1

How about:

;WITH data AS
(
SELECT 1 as [value],'2012-03-07' as date
UNION ALL SELECT 1,'2012-03-06'
UNION ALL SELECT 1,'2012-03-05'
UNION ALL SELECT 3,'2012-03-04'
UNION ALL SELECT 4,'2012-03-03'
UNION ALL SELECT 1,'2012-03-02'
)
,data2 AS
(
SELECT 
[Value]
,date
,row_number() OVER (ORDER BY (SELECT NULL)) as row1
FROM data
)
,data3 AS
(
SELECT * 
, row1 - row_number() OVER (ORDER BY [Value]) as row2
from data2
)
SELECT 
MIN([Value]) AS [Value]
,MIN(Date) as Date
FROM data3
GROUP BY [Value] - row2
ORDER BY MIN(row1)