SQL query initial and final balance as well sum of events - oracle

Question

I have a fairly big table (about 300gb) as:
event_day event_hour customer_id initial_bal final_bal topups debts
01/01 00 11111 0 50 60 10
01/01 01 11111 50 80 45 15
01/01 02 11111 80 30 0 50
...
I want to summarize it into whole days, e.g.:
event_day customer_id initial_bal final_bal topups debts
01/01 11111 0 30 105 75
...
But I´m having troubles with the analitical functions.. I'm working with something like:
select *
from
(
select
event_day, customer_id, initial_bal, final_bal,
sum(topups) over (partition by event_day, customer_id by event_day, customer_id asc) topups,
row_number() over (partition by event_day, customer_id order by event_day, customer_id asc) as initial_bal,
row_number() over (partition by event_day, customer_id order by event_day, customer_id desc) as final_bal
from MY_300GB_TABLE t
)
where initial_bal = 1 or final_bal = 1
order by customer_id, event_day
Which inst doing what I expected... a hand?
I´m tring to avoid joins, sub-queries and such.. I simplified here but there the actual project involves a few big tables and performance might an issue. I'm using Oracle 12C.
thanks!

Answer 1

Good occasion to aggregate with first (or last) option:
select event_day, customer_id,
max(initial_bal) keep (dense_rank first order by event_hour) initial_bal,
max(final_bal) keep (dense_rank last order by event_hour) final_bal,
sum(topups) topups, sum(debts) debts
from tla_t_balance_summary t
group by event_day, customer_id;
dbfiddle demo
Your query works too, but you made a mistake in order by. And it requires additional aggregation, because we have two rows for customers:
select event_day, customer_id, max(topups), max(debts),
min(case rib when 1 then initial_bal end) ib, min(case rfb when 1 then final_bal end) fb
from (
select event_day, customer_id, initial_bal, final_bal,
sum(topups) over (partition by event_day, customer_id) topups,
sum(debts) over (partition by event_day, customer_id) debts,
row_number() over (partition by event_day, customer_id order by event_hour) as rib,
row_number() over (partition by event_day, customer_id order by event_hour desc) as rfb
from tla_t_balance_summary t)
where rib = 1 or rfb = 1
group by customer_id, event_day;

Answer 2

If you are looking for the first and last entry for each day based on hour, your row_number function should reflect that, with something like:
select *
from
(
select
event_day, customer_id, initial_bal, final_bal,
sum(recharge_amount) over (partition byevent_day, customer_id) topups,
row_number() over (partition by event_day, customer_id order by event_hour asc) as initial_hr,
row_number() over (partition by event_day, customer_id order by event_hour desc) as final_hr
from TLA_T_BALANCE_SUMMARY t
)
where initial_hr = 1 or final_hr = 1
order by customer_id, event_day
Its hard to comment exactly because your query doesn't really match the data in terms of columns etc.

Answer 3

I think you will need to use GROUP BY and analytical function as following:
SELECT
EVENT_DAY,
CUSTOMER_ID,
MAX(INITIAL_BAL) AS INITIAL_BAL,
MAX(FINAL_BAL) AS FINAL_BAL,
SUM(TOPUPS) AS TOPUPS,
SUM(DEBTS) AS DEBTS
FROM
(
SELECT
EVENT_DAY,
CUSTOMER_ID,
FIRST_VALUE(INITIAL_BAL) OVER(
PARTITION BY EVENT_DAY, CUSTOMER_ID
ORDER BY
EVENT_HOUR
) AS INITIAL_BAL,
FIRST_VALUE(FINAL_BAL) OVER(
PARTITION BY EVENT_DAY, CUSTOMER_ID
ORDER BY
EVENT_HOUR DESC
) AS FINAL_BAL,
TOPUPS,
DEBTS
FROM
TLA_T_BALANCE_SUMMARY T
WHERE
INITIAL_BAL = 1
OR FINAL_BAL = 1
)
GROUP BY
EVENT_DAY,
CUSTOMER_ID
ORDER BY
CUSTOMER_ID,
EVENT_DAY;
Cheers!!