Exclude duplicates and capture only changes - sql

Question

I have a scenario, where I have to exclude duplicates and capture only the changes. Also calculate the valid_from and valid_to on the fly. I have tried a query and it works but it is very slow in performance and it is failing with memory error .
Input : Only capture Entries where there is a change either in Amount/Check-In-Out.
Calculate Valid_from and Valid_to based on Date Changed.
Output:
SQL I tried.
select * from (select
lead(start_date, "window_offset" - rn + 1, '9999-12-31') over (order by "grp" ) as valid_to,
case when rn = max(rn) over (partition by "grp") then 1 else 0 end as "isLastUpdate",
start_date as valid_from,*
from (
select
min("DateChanged") over (partition by "grp") as start_date,
count(*) over (partition by "grp") as "window_offset",
row_number() over (partition by "grp" order by "DateChanged") as rn,
*
from (
select sum("isChanged") over (partition by OrderId order by "DateChanged") as "grp",*
from (
select
case when "Amount" = lag( "Amount" ) over (partition by OrderId order by "DateChanged") and
"Check-In" = lag( "Check-In" ) over (partition by OrderId order by "DateChanged") and
"Check-Out" = lag( "Check-Out" ) over (partition by OrderId order by "DateChanged")
then 0
else 1
end "isChanged",
*
FROM :in_table
)
))
where "isLastUpdate" = 1;

Answer 1

The logic of your expected answer is unclear as to why you get valid_from as 8-mar-21 for the first order_id and 9-apr-21 for the second order_id as both order_ids have overlapping ranges but you take the least of the previous check_out and the next check_in for the first order_id and the greatest of those two for the second and it is inconsistent.
If you want to get valid_from as the greatest of either the current check_in or the previous check_outs and valid_to as the greatest of either the current check_out or the next check_in or, if there are no more rows, 9999-12-31 then:
SELECT orderid,
amount,
check_in,
check_out,
GREATEST(
check_in,
COALESCE(
MAX(check_out) OVER (
PARTITION BY orderid
ORDER BY check_in, check_out
ROWS BETWEEN UNBOUNDED PRECEDING AND 1 PRECEDING
),
check_in
)
) AS valid_from,
GREATEST(
check_out,
LEAD(check_in, 1, DATE '9999-12-31') OVER (
PARTITION BY orderid ORDER BY check_in, check_out
)
) AS valid_to
FROM (
SELECT DISTINCT *
FROM table_name
)
Which, for the sample data:
CREATE TABLE table_name (orderid, datechanged, amount, check_in, check_out) AS
SELECT 1, DATE '2021-03-3', 12.12, DATE '2021-03-03', DATE '2021-03-10' FROM DUAL UNION ALL
SELECT 1, DATE '2021-03-3', 12.12, DATE '2021-03-03', DATE '2021-03-10' FROM DUAL UNION ALL
SELECT 1, DATE '2021-03-3', 12.12, DATE '2021-03-03', DATE '2021-03-10' FROM DUAL UNION ALL
SELECT 1, DATE '2021-03-8', 21.12, DATE '2021-03-08', DATE '2021-03-18' FROM DUAL UNION ALL
SELECT 1, DATE '2021-03-8', 21.12, DATE '2021-03-08', DATE '2021-03-18' FROM DUAL UNION ALL
SELECT 2, DATE '2021-04-4', 9.10, DATE '2021-04-04', DATE '2021-04-09' FROM DUAL UNION ALL
SELECT 2, DATE '2021-04-4', 9.10, DATE '2021-04-04', DATE '2021-04-09' FROM DUAL UNION ALL
SELECT 2, DATE '2021-04-4', 10.20, DATE '2021-04-04', DATE '2021-04-12' FROM DUAL;
Outputs:
ORDERID
AMOUNT
CHECK_IN
CHECK_OUT
VALID_FROM
VALID_TO
1
12.12
2021-03-03 00:00:00
2021-03-10 00:00:00
2021-03-03 00:00:00
2021-03-10 00:00:00
1
21.12
2021-03-08 00:00:00
2021-03-18 00:00:00
2021-03-10 00:00:00
9999-12-31 00:00:00
2
9.1
2021-04-04 00:00:00
2021-04-09 00:00:00
2021-04-04 00:00:00
2021-04-09 00:00:00
2
10.2
2021-04-04 00:00:00
2021-04-12 00:00:00
2021-04-09 00:00:00
9999-12-31 00:00:00
db<>fiddle here