Group By element ID and continuous date range as a single row - URGENT!!!!

Question

Hi All,
I have a source table and a target table.
Source Table Target Table
Element ID Period_dt Element ID Effective date End date
DD001 200901 DD001 200901 200903
DD001 200902 DD001 200906 200908
DD001 200903 DD002 200801 200803
DD001 200906
DD001 200907
DD001 200908
DD002 200801
DD002 200802
DD002 200803
I want the result as in the target table. Basically, continuous date range should be grouped and shown as single row even it falls under the same elment_id.
I have tried the LAG and LEAD function and RANK function as well but unsuccessful.
I was able to get like this in the target table using MIN and MAX function.
DD001 200901 200908
DD002 200801 200803
For DD001, you can see there is a break in the months. 200901 - 200903 and 200906 - 200908. we are missing 4th,5th month. 1 to 3rd month and 6th to 8th month should be grouped and shown as separate rows in the target table for the same DD001 element_ID
I will post the SQL query tommorrow. Please give your suggestions.
Regards
Balaji

Answer 1

Thanks guys. It worked perfectly. I apologize for using the 'U' word. This is my first post here.
select prod_element_cd,
min(period_dt) effective_date,
max(Last_day(period_dt)) end_date,
SUM(Fixed_factor),
SUM(var_factor),
val1
from (
select prod_element_cd, period_dt,Fixed_factor,var_factor, val, last_value(val ignore nulls) over(partition by prod_element_cd order by period_dt) val1
from (
select prod_element_cd,
period_dt,
NVL(Fixed,0) Fixed_factor,
NVL(variable,0) var_factor,
lag(period_dt) over(partition by prod_element_cd order by period_dt) dt,
case when add_months(period_dt,-1) = lag(period_dt) over(partition by prod_element_cd order by period_dt)
then null
else rownum end val
from pmax_land.TMP_COST_CASH_STD_INPUT)
group by prod_element_cd, val1
order by prod_element_cd
The above query pulls the below result
PROD_ELEMENT_CD EFFECTIVE_DATE END_DATE FIXED VARIABLE VAL1
DDA001 01/01/2009 03/31/2009 4.20 7.62 1.00
DDA001 06/01/2009 11/30/2009 4.80 0.72 10.00
DDA001 01/01/2010 01/31/2010 0.75 0.50 13.00
DDA002 07/01/2008 09/30/2008 2.40 0.36 11.00
DDA002 02/01/2008 03/31/2008 1.50 1.00 14.00
one more logic added to the requirement
for each occurance, for eg: DDA001, the last row of DDA001 should be taken and the end_date of that should be hardcoded to 12/31/9999
here we have two cases
last row for DDA001 end_date
DDA001 01/01/2010 01/31/2010 0.75 0.50 13.00
end date is 01/31/2010 for this above row of DDA001. It should be hardcoded to 12/31/9999
similarly
last row for DDA002 end_date
DDA002 02/01/2008 03/31/2008 1.50 1.00 14.00
end date is 03/31/2008 for this above row of DDA002. It should be hardcoded to 12/31/9999
Similarly for DDA003,DDA004.......... etc
Thanks for your previous replies. Please give your suggestions.
Regards
Balaji
Edited by: user12119826 on Oct 27, 2009 11:49 PM