Query to aggregate data

Question

Hi, I need some help coming up with a query for department analysis. I am providing test cases below. My data is currently being taken from a Oracle 10g materialized view. What I need to do is give a date interval and find out how many people were in each dept (over all depts that exist in the view) at the beginning and end of the interval. If a person was in more than 1 dept during that period, only the last one is considered. Also, how many new people were added to the company
For example, using the data below and 13-AUG-04 through 30-AUG-04, I would get:
Unit #Beg #End #New
18 0 0 1
33 1 1 0
56 0 0 0
70 1 1 0
71 1 1 0
The last two columns in the test table (First/Last) refer to the person's employment in the company. There will be for sure either Y/Y (if the person has one period of employment only) or both Y/N, N/Y.
Where I'm having problems is to keep track of both current departments as well as new to the company. Someone can be new to the company (evidenced by First='Y'), but if he 's only moving to a different department then he's not new to the company.
Thanks,
Rob
create table test_dept
dept number,
head varchar2(20),
wrkid number,
name varchar2(50),
unitid number,
startdate date,
enddate date,
firstentry char(1),
lastentry char(1)
insert into test_dept values(20812,'JONES',27264,'SMITH, Mary',71,to_date('09-AUG-04','DD-MON-YY'),to_date('11-AUG-04','DD-MON-YY'),'Y','N');
insert into test_dept values(20812,'JONES',27264,'SMITH, Mary',71,to_date('11-AUG-04','DD-MON-YY'),to_date('04-OCT-04','DD-MON-YY'),'N','N');
insert into test_dept values(20812,'JONES',27264,'SMITH, Mary',33,to_date('04-OCT-04','DD-MON-YY'),to_date('05-OCT-04','DD-MON-YY'),'N','N');
insert into test_dept values(20812,'JONES',27264,'SMITH, Mary',33,to_date('05-OCT-04','DD-MON-YY'),to_date('19-APR-05','DD-MON-YY'),'N','Y');
insert into test_dept values(20812,'JONES',27265,'SMITH, Jack',71,to_date('09-AUG-04','DD-MON-YY'),to_date('11-AUG-04','DD-MON-YY'),'Y','N');
insert into test_dept values(20812,'JONES',27265,'SMITH, Jack',33,to_date('10-AUG-04','DD-MON-YY'),to_date('05-OCT-04','DD-MON-YY'),'N','N');
insert into test_dept values(20812,'JONES',27265,'SMITH, Jack',33,to_date('05-OCT-04','DD-MON-YY'),to_date('20-APR-05','DD-MON-YY'),'N','Y');
insert into test_dept values(28022,'BABS',39220,'RUPERT, A',70,to_date('11-AUG-04','DD-MON-YY'),to_date('29-OCT-04','DD-MON-YY'),'Y','N');
insert into test_dept values(28022,'BABS',39220,'RUPERT, A',33,to_date('19-OCT-04','DD-MON-YY'),to_date('25-OCT-04','DD-MON-YY'),'N','N');
insert into test_dept values(28022,'BABS',39220,'RUPERT, A',33,to_date('25-OCT-04','DD-MON-YY'),to_date('23-NOV-04','DD-MON-YY'),'N','N');
insert into test_dept values(28022,'BABS',39220,'RUPERT, A',70,to_date('23-NOV-04','DD-MON-YY'),to_date('27-JAN-05','DD-MON-YY'),'N','N');
insert into test_dept values(28022,'BABS',39220,'RUPERT, A',33,to_date('08-FEB-05','DD-MON-YY'),to_date('13-JUL-05','DD-MON-YY'),'N','N');
insert into test_dept values(28022,'BABS',39220,'RUPERT, A',56,to_date('13-JUL-05','DD-MON-YY'),to_date('31-OCT-05','DD-MON-YY'),'N','Y');
insert into test_dept values(20812,'JONES',10000,'B',18,to_date('15-AUG-04','DD-MON-YY'),to_date('29-AUG-04','DD-MON-YY'),'Y','Y');

Answer 1

this?
var d1 varchar2(100)
var d2 varchar2(100)
exec :d1 := '13-AUG-04'
exec :d2 := '30-AUG-04'
select unitid,
sum( case when startdate <= to_date(:d1) and enddate >= to_Date(:d1)
            and nextstart >= to_date(:d2) then 1 else 0 end ) beg#,
sum( case when startdate <= to_date(:d2) and enddate >= to_Date(:d2) then 1 else 0 end ) end#,
sum( case when firstentry='Y' and startdate between to_Date(:d1) and to_date(:d2) then 1 else 0 end ) new#
from (
  select unitid, wrkid, name,
   startdate, enddate, firstentry,
   last_value(startdate) over (partition by wrkid order by startdate desc
      rows between unbounded preceding and 1 preceding) nextstart
  from test_dept
group by unitid
/you should have test data where a dept has different beg and end numbers.