Using the LAG function

Question

Hi all,
using: Oracle9i Enterprise Edition Release 9.0.1.3.
I have a table that looks basically like this:
SQL> select labour_rate_id, organisation_id, commence_date_time, termination_date_time, labour_rate
  2    from labour_rate
  3  where organisation_id = 3265022
  4  order by commence_date_time;
LABOUR_RATE_ID ORGANISATION_ID COMMENCE_ TERMINATI LABOUR_RATE
          7095         3265022 20-APR-05 30-OCT-07        43.5
          9428         3265022 01-JAN-06 31-DEC-07        43.5
         10762         3265022 01-NOV-07 31-DEC-08       55.52As you can see with the first organisation, some of the dates overlap - i.e. the second last termination date is between the start and end dates of the last record.
I'm writing an update statement to make the previous termination date, one day less than the current commence date.
I cannot rely on the labour rate Id to provide any sequence, it's just a coincidence that these are in order. I can rely on the commence date to be accurate.
here's what I have so far:
select * from (select organisation_id,
                      labour_rate_id,
                      commence_date_time,
                      termination_date_time,              
                      lag(labour_rate_id) over (order by organisation_id, commence_date_time) as prev_labour_rate_id,
                      lag(termination_date_time) over(order by organisation_id, commence_date_time) as prev_term_date_time
                 from labour_rate)      
where prev_term_date_time between commence_date_time and termination_date_timethis should select all those where the previous termination date is between the start and end date of the current one. the only obvious problem with this is:
LABOUR_RATE_ID ORGANISATION_ID COMMENCE_ TERMINATI LABOUR_RATE
         10742         4406709 01-NOV-06 31-DEC-07          40
         10743         4406711 18-DEC-06 31-DEC-07          46
         10750         4415820 31-OCT-07 31-DEC-08        4.75most of the data is like this. I only want to get the non-contiguous dates for each organisation ID. the above query will pick up these rows due to the lag function not taking into account distinct organisation ID's.
this works:
select  lrp.labour_rate_id,
        lrp.organisation_id,       
        lr.commence_date_time,
        lr.termination_date_time,
        lrp.labour_rate_id as prev_labour_rate_id,
        lrp.termination_date_time as prev_term_date_time
   from labour_rate lr,
        labour_rate lrp
  where lrp.organisation_id = lr.organisation_id
    and lrp.commence_date_time = (select max(commence_date_time)
                                from labour_rate lrs
                               where lrs.organisation_id = lr.organisation_id
                                 and lrs.commence_date_time < lr.commence_date_time
    and lrp.termination_date_time > lr.commence_date_timebut it makes me cringe. there surely is a more elegant way of doing this?
ultimately I want to be able to set the termination date equal to the day before the subsequent commencement date for the same organisation where the termination date is between the subsequent commencement and end dates.
would someone be able to point me in the right direction?

Answer 1

Well, for a start
lag(labour_rate_id) over (order by organisation_id, commence_date_time)should be
lag(labour_rate_id) over (partition by organisation_id order by commence_date_time)