Return Data based Relevant Dates - sql

Question

I have an occupancy table and a pay history table. I want to return the state that the employee is in from the occupancy at the time of the relevant pay.
Occupancy Table
Emp#|Commence Date|State
-----|-------------|----
101 | 1/01/2016 | VIC
101 | 1/04/2016 | NSW
101 | 1/08/2016 | ACT
Pay History Table
Emp#|Pay Date
----|--------
101 |15/01/2016
101 |15/02/2016
101 |15/03/2016
101 |15/04/2016
101 |15/05/2016
101 |15/06/2016
101 |15/07/2016
101 |15/08/2016
101 |15/09/2016
I'm wanting to return the following
Emp#|:Pay Date:|State
----|----------|-----
101 |15/01/2016|VIC
101 |15/02/2016|VIC
101 |15/03/2016|VIC
101 |15/04/2016|NSW
101 |15/05/2016|NSW
101 |15/06/2016|NSW
101 |15/07/2016|NSW
101 |15/08/2016|ACT
101 |15/09/2016|ACT
Can someone assist, please

Answer 1

You need to generate the end_date in the occupancy table in a subquery; the lead() function is perfect for this purpose. I use it with all three arguments - the third argument gives a "default" date which I chose arbitrarily as 15 December 2099 for the "current" status. Then its a simple join on empno and a between condition on dates.
I assume you have more than one empno in your data, so I accommodated that. Then: I don't know if # is legal in Oracle column names, but I didn't want to try; I changed to empno. And names definitely can't have spaces in them unless you quote the names, which has many disadvantages; I worked around that too.
with
occupancy ( empno, commence_date, state ) as (
select 101, to_date('1/01/2016', 'dd/mm/yyyy'), 'VIC' from dual union all
select 101, to_date('1/04/2016', 'dd/mm/yyyy'), 'NSW' from dual union all
select 101, to_date('1/08/2016', 'dd/mm/yyyy'), 'ACT' from dual
),
pay_history ( empno, pay_date ) as (
select 101, to_date('15/01/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/02/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/03/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/04/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/05/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/06/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/07/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/08/2016', 'dd/mm/yyyy') from dual union all
select 101, to_date('15/09/2016', 'dd/mm/yyyy') from dual
)
-- end of test data (not part of the SQL query); query begins below this line
select p.empno, p.pay_date, o.state
from pay_history p inner join (
select empno, commence_date,
lead(commence_date, 1, date '2099-12-15')
over (partition by empno order by commence_date) as end_date,
state
from occupancy ) o
on p.empno = o.empno
and p.pay_date between o.commence_date and o.end_date
order by empno, pay_date -- if needed
;
Output:
EMPNO PAY_DATE STATE
----- ---------- -----
101 15/01/2016 VIC
101 15/02/2016 VIC
101 15/03/2016 VIC
101 15/04/2016 NSW
101 15/05/2016 NSW
101 15/06/2016 NSW
101 15/07/2016 NSW
101 15/08/2016 ACT
101 15/09/2016 ACT
9 rows selected.

Answer 2

Negotiating against myself here :-) I am posting this as a separate Answer, rather than editing my earlier one, because this is indeed a different answer.
Please refer to my other Answer for the input data and sample output - they are the same. Only the query is different. Instead of a join, we can UNION ALL the two tables (with some necessary adjustments: add a null column for state to the pay_history table, and a flag of 0 for the occupancy table and 1 for the pay_history table); then use the last_value() analytic function on the resulting union, and filter out the rows from the occupancy table in the outermost query. This may be quite a bit faster than the join-based solution.
select empno, dt as pay_date, state
from (
select empno, dt, flag,
last_value(state ignore nulls)
over (partition by empno order by dt, flag) as state
from (
select empno, commence_date as dt, state, 0 as flag
from occupancy
union all
select empno, pay_date, null, 1
from pay_history
)
)
where flag = 1
order by empno, pay_date -- if needed
;