Rolliing backword window. How to calculate and get op like this

Question

I have date as below
with abc as (select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 216 ilo_id, to_date('06/01/2009', 'dd/mm/yyyy') i_date from dual union all
             select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 215 ilo_id, to_date('05/01/2009', 'dd/mm/yyyy') i_date from dual union all
             select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 214 ilo_id, to_date('04/01/2009', 'dd/mm/yyyy') i_date from dual union all
             select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 213 ilo_id, to_date('04/01/2009', 'dd/mm/yyyy') i_date from dual union all
             select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 212 ilo_id, to_date('03/01/2009', 'dd/mm/yyyy') i_date from dual union all
             select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 211 ilo_id, to_date('02/01/2009', 'dd/mm/yyyy') i_date from dual)
select *
from   abc
USERID     CLO_ID      C_DATE                       ILO_ID  I_DATE           
         1         11 05/01/2009 00:00:00        216 06/01/2009 00:00:00
         1         11 05/01/2009 00:00:00        215 05/01/2009 00:00:00
         1         11 05/01/2009 00:00:00        214 04/01/2009 00:00:00
         1         11 05/01/2009 00:00:00        213 04/01/2009 00:00:00
         1         11 05/01/2009 00:00:00        212 03/01/2009 00:00:00
         1         11 05/01/2009 00:00:00        211 02/01/2009 00:00:00
Now i want to calculate rolling backword frequency of each ilo id and output should come like as based Frequency winodw
e.g. if frequency window is 2 then
USERID   CLO_ID C_DATE                     ILO_ID  I_DATE                        FREQUENCY
         1         11 05/01/2009 00:00:00        216 06/01/2009 00:00:00          2
         1         11 05/01/2009 00:00:00        215 05/01/2009 00:00:00          3
         1         11 05/01/2009 00:00:00        214 04/01/2009 00:00:00          3
         1         11 05/01/2009 00:00:00        213 04/01/2009 00:00:00          2
         1         11 05/01/2009 00:00:00        212 03/01/2009 00:00:00          2
         1         11 05/01/2009 00:00:00        211 02/01/2009 00:00:00          1
1. For each ILO ID, go back number od days given as Frequency window (in our case its 2)
2. calculate the number of ilo_id in that range
3. number of ilo_id is the frequency of that ilo_idHow can i achieve this.
Edited by: Kuldeep2 on Sep 29, 2010 2:30 AM
Edited by: Kuldeep2 on Sep 29, 2010 2:33 AM

Answer 1

Use analytic COUNT:
variable frequency number
exec :frequency := 2;
with abc as (select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 216 ilo_id, to_date('06/01/2009', 'dd/mm/yyyy') i_date from dual union all
select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 215 ilo_id, to_date('05/01/2009', 'dd/mm/yyyy') i_date from dual union all
select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 214 ilo_id, to_date('04/01/2009', 'dd/mm/yyyy') i_date from dual union all
select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 213 ilo_id, to_date('04/01/2009', 'dd/mm/yyyy') i_date from dual union all
select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 212 ilo_id, to_date('03/01/2009', 'dd/mm/yyyy') i_date from dual union all
select 1 userid, 11 clo_id, to_date('05/01/2009', 'dd/mm/yyyy') c_date, 211 ilo_id, to_date('02/01/2009', 'dd/mm/yyyy') i_date from dual)
select  abc.*,
        count(ilo_id) over(order by i_date range between :frequency - 1 preceding and current row) cnt
  from  abc
  order by i_date desc
    USERID     CLO_ID C_DATE        ILO_ID I_DATE           CNT
         1         11 05-JAN-09        216 06-JAN-09          2
         1         11 05-JAN-09        215 05-JAN-09          3
         1         11 05-JAN-09        214 04-JAN-09          3
         1         11 05-JAN-09        213 04-JAN-09          3
         1         11 05-JAN-09        212 03-JAN-09          2
         1         11 05-JAN-09        211 02-JAN-09          1
6 rows selected.
SQL> SY.