sqHow to get sum values weekly based between two dates? - sql

Question

My table values like ...
Date Amt Cash Money Name
15-Jun 100 10 20 GUL
16-Jun 200 20 40 ABC
20-Jun 300 30 60 GUL
25-Jun 400 40 80 BCA
28-Jun 500 50 10 GUL
3-Jul 600 60 120 ABC
19-Jun 700 70 140 BCA
26-Jun 800 80 160 ABC
7-Jul 900 90 180 GUL
9-Jul 1000 100 200 ABC
I need to return weekly based sum of values between two date in oracle .My expected output.
Date Amt Cash Mony
13 to 19 June 1000 100 200
20 to 26 June 1500 150 300
27 to3 July 1100 110 130
4 to 10 July 1900 190 380

Answer 1

you can achieve this by a case statement:
e.g.
-- test data
with data(dat,
val1,
val2) as
(select sysdate - 7, 12, 13
from dual
union all
select sysdate - 6, 32, 1
from dual
union all
select sysdate - 5, 52, 53
from dual
union all
select sysdate - 4, 2, 16
from dual
union all
select sysdate - 3, 72, 154
from dual)
select -- build up your groups
case
when d.dat < to_date('28.09.2016', 'DD.MM.YYYY') then
'<28.09.'
when d.dat > to_date('30.09.2016', 'DD.MM.YYYY') then
'>30.09.'
else
'28.-30.'
end as grp,
sum(val1),
sum(val2)
from data d
group by case
when d.dat < to_date('28.09.2016', 'DD.MM.YYYY') then
'<28.09.'
when d.dat > to_date('30.09.2016', 'DD.MM.YYYY') then
'>30.09.'
else
'28.-30.'
end;
-- output
grp sum(val1) sum(val2)
28.-30. 84 54
<28.09. 12 13
>30.09. 74 170
To group by calendar week use
-- test data
with data(dat,
val1,
val2) as
(select sysdate - 9, 12, 13
from dual
union all
select sysdate - 6, 32, 1
from dual
union all
select sysdate - 5, 52, 53
from dual
union all
select sysdate - 4, 2, 16
from dual
union all
select sysdate + 3, 72, 154
from dual)
select TRUNC(dat, 'iw') ||'-'|| TRUNC(dat+7, 'iw'),
sum(val1),
sum(val2)
from data
group by TRUNC(dat, 'iw') ||'-'|| TRUNC(dat+7, 'iw');

Answer 2

The query below has the input dates (from and to) in the first factored subquery. Those can be made into bind variables, or whatever mechanism you want to use to pass these inputs to the query. Then I have the test data in the second factored subquery; you don't need that in your final solution. I create all the needed weeks in the "weeks" factored subquery and I use a left outer join, so that weeks with no transactions will show 0 sums. Note that in the main query, where I do a join, the "date" column from the base table is not enclosed within any kind of function; this allows the use of an index on that column, which you should have if the table is very large, or if performance may be a concern for any other reason. Note that the output is different from yours (missing the last row) because I input a to-date before the last row in the table. That is intentional, I wanted to make sure the query works correctly. Also: I didn't use "date" or "week" as column names; that is a very poor practice. Reserved Oracle keywords should not be used as column names. I used "dt" and "wk" instead.
with
user_inputs ( from_dt, to_dt ) as (
select to_date('4-Jun-2016', 'dd-Mon-yyyy'), to_date('3-Jul-2016', 'dd-Mon-yyyy') from dual
),
test_data ( dt, amt, cash, money, name ) as (
select to_date('15-Jun-2016', 'dd-Mon-yyyy'), 100, 10, 20, 'GUL' from dual union all
select to_date('16-Jun-2016', 'dd-Mon-yyyy'), 200, 20, 40, 'ABC' from dual union all
select to_date('20-Jun-2016', 'dd-Mon-yyyy'), 300, 30, 60, 'GUL' from dual union all
select to_date('25-Jun-2016', 'dd-Mon-yyyy'), 400, 40, 80, 'BCA' from dual union all
select to_date('28-Jun-2016', 'dd-Mon-yyyy'), 500, 50, 10, 'GUL' from dual union all
select to_date( '3-Jul-2016', 'dd-Mon-yyyy'), 600, 60, 120, 'ABC' from dual union all
select to_date('19-Jun-2016', 'dd-Mon-yyyy'), 700, 70, 140, 'BCA' from dual union all
select to_date('26-Jun-2016', 'dd-Mon-yyyy'), 800, 80, 160, 'ABC' from dual union all
select to_date( '7-Jul-2016', 'dd-Mon-yyyy'), 900, 90, 180, 'GUL' from dual union all
select to_date( '9-Jul-2016', 'dd-Mon-yyyy'), 1000, 100, 200, 'ABC' from dual
),
weeks ( start_dt ) as (
select trunc(from_dt, 'iw') + 7 * (level - 1)
from user_inputs
connect by level <= 1 + (to_dt - trunc(from_dt, 'iw')) / 7
)
select to_char(w.start_dt, 'dd-Mon-yyyy') || ' - ' ||
to_char(w.start_dt + 6, 'dd-Mon-yyyy') as wk,
nvl(sum(t.amt), 0) as tot_amt, nvl(sum(t.cash), 0) as tot_cash,
nvl(sum(t.money), 0) as tot_money
from weeks w left outer join test_data t
on t.dt >= w.start_dt and t.dt < w.start_dt + 7
group by start_dt
order by start_dt
;
Output:
WK TOT_AMT TOT_CASH TOT_MONEY
-------------------------------------------- ---------- ---------- ----------
30-May-2016 - 05-Jun-2016 0 0 0
06-Jun-2016 - 12-Jun-2016 0 0 0
13-Jun-2016 - 19-Jun-2016 1000 100 200
20-Jun-2016 - 26-Jun-2016 1500 150 300
27-Jun-2016 - 03-Jul-2016 1100 110 130

Answer 3

You can try like below, I chose 13-Jun-2016 as a starting date. You can chose it as per your requirement upto any range of dates.
with t as
(select dt,
min(dt) over (partition by week)||' to '|| max(dt) over (partition by week) week
from (
select to_date('13-Jun-2016','dd-Mon-yyyy')+(level-1) dt,
ceil(level/7) week
from dual
connect by level<=52))
select week,
sum(amt),
sum(cash),
sum(money)
from (
select your_table.*,
t.week
from your_table,t
where trunc(to_date(your_table.dt,'dd-Mon-yyyy'))=trunc(t.dt))
group by week;