How to get the last different occurrence in a SCD table

Question

hey,
i have a slowly changing dimension and i want to get the last different occurrence for each row.
Sample Data
the field LAST_VALUE is the one i'm looking for.
ID     START_DATE     END_DATE     VALUE     LAST_VALUE
1     01/01/2000     01/02/2000     X     null
1     01/02/2000     01/03/2000     X     null
1     01/03/2000     01/04/2000     Y     X
1     01/04/2000     01/05/2000     Y     X
1     01/05/2000     01/06/2000     X     Y
1     01/06/2000     01/07/2000     X     Y
1     01/07/2000     01/08/2000     Z     X
1     01/08/2000     01/09/2000     Q     Z i want to try and do that without three sub selects.
i've tried analytic functions but lag brings me the last occurrence which is not necessarily different.
thanks.

Answer 1

Something like this :
SQL> with sample_data as (
  2    select 1 ID, to_date('01/01/2000','DD/MM/YYYY') START_DATE, to_date('01/02/2000','DD/MM/YYYY') END_DATE, 'X' VALUE from dual union all
  3    select 1, to_date('01/02/2000','DD/MM/YYYY'), to_date('01/03/2000','DD/MM/YYYY'), 'X' from dual union all
  4    select 1, to_date('01/03/2000','DD/MM/YYYY'), to_date('01/04/2000','DD/MM/YYYY'), 'Y' from dual union all
  5    select 1, to_date('01/04/2000','DD/MM/YYYY'), to_date('01/05/2000','DD/MM/YYYY'), 'Y' from dual union all
  6    select 1, to_date('01/05/2000','DD/MM/YYYY'), to_date('01/06/2000','DD/MM/YYYY'), 'X' from dual union all
  7    select 1, to_date('01/06/2000','DD/MM/YYYY'), to_date('01/07/2000','DD/MM/YYYY'), 'X' from dual union all
  8    select 1, to_date('01/07/2000','DD/MM/YYYY'), to_date('01/08/2000','DD/MM/YYYY'), 'Z' from dual union all
  9    select 1, to_date('01/08/2000','DD/MM/YYYY'), to_date('01/09/2000','DD/MM/YYYY'), 'Q' from dual
10  )
11  select v.id
12       , v.start_date
13       , v.end_date
14       , v.value
15       , last_value(case when value != prev_value then prev_value end ignore nulls)
16               over(partition by id order by start_date) as "LAST_VALUE"
17  from (
18    select t.*
19         , lag(value) over(partition by id order by start_date) as prev_value
20    from sample_data t
21  ) v
22  ;
        ID START_DATE  END_DATE    VALUE LAST_VALUE
         1 01/01/2000  01/02/2000  X    
         1 01/02/2000  01/03/2000  X    
         1 01/03/2000  01/04/2000  Y     X
         1 01/04/2000  01/05/2000  Y     X
         1 01/05/2000  01/06/2000  X     Y
         1 01/06/2000  01/07/2000  X     Y
         1 01/07/2000  01/08/2000  Z     X
         1 01/08/2000  01/09/2000  Q     Z
8 rows selected