Group by days interval SQL

Question

I have a table with 4 columns:

ID, Date, Office, Reports

1. "Date" shows when the reports were created.
2. "Office" in which the office was created.
3. "Reports" the number of reports created.

Original table

  CREATE TABLE REPORTS (
    `ID` INTEGER,
    `Date` DATE,
    `Office` VARCHAR(2),
    `Reports` INTEGER
  );

  INSERT INTO REPORTS
    (`ID`, `Date`, `Office`, `Reports`)
  VALUES
    ('1', '2020-09-01', 'AR', '3'),
    ('2', '2020-09-01', 'ES', '2'),
    ('3', '2020-09-03', 'ES', '4'),
    ('4', '2020-09-05', 'AR', '1'),
    ('5', '2020-09-05', 'OG', '1'),
    ('6', '2020-09-05', 'ES', '4'),
    ('7', '2020-09-08', 'ES', '5'),
    ('8', '2020-09-08', 'AR', '6'),
    ('9', '2020-09-15', 'OG', '7'),
    ('10', '2020-09-16', 'ES', '9');

I need to use a sql line code to get a table that shows all the dates (not showing duplicated dates), and two new columns "Reports_7_days_prev" and "Reports_7_days_after". The result should be the following table:

date       | Reports 7 prev | Reports 7 after
:--------- | -------------: | --------------:
2020-09-01 |              5 |              26
2020-09-03 |              9 |              21
2020-09-05 |             15 |              17
2020-09-08 |             26 |              18
2020-09-15 |             18 |              16
2020-09-16 |             16 |               9

The way of calculating each column is the following:

• "Reports 7 days prev": it should sum all the reports created on the 7 days previous to the date of each row. For example, in the row with date "2020-09-15" from the output table, the result 18 is the sum of all the values from column "Reports" within the interval of dates [2020-09-08 , 2020-09-15], corresponding to the rows with "ID" 7, 8 and 9 => (5 + 6 + 7 = 18).

• " Reports 7 days after": it should sum all the reports created on the 7 days after the date of each row. For example, in the row with date "2020-09-05" from the output table, the result 17 is the sum of all the values from column "Reports" within the interval of dates [2020-09-05 , 2020-09-12], corresponding to the rows with "ID" 4, 5, 6, 7 and 8 => (1 + 1 + 4 + 5 + 6 = 17)

I think than my SQL query should be something like:

SELECT Date, 
       sum([QUERY 1]) as "Reports  7 days prev", 
       sum ([QUERY 2]) as "Reports 7 days after" 
FROM REPORTS 
GROUP BY Date

But I cannot figure out what should I include in [QUERY 1] and [QUERY 2] to get the 7 days previous and 7 days after.

Thank you very much,

Answer 1

You can use Scalar Correlated Subqueries , but don't expect good performance:

select date
  ,(select sum(reports) from REPORTS as r2 where r2.date between r.date - 7 and r.date) as `Reports 7 prev`
  ,(select sum(reports) from REPORTS as r2 where r2.date between r.date and r.date + 7) as `Reports 7 after`
from REPORTS as r
group by date

See fiddle

Answer 2

If there's no need to do it with just one select statement you can use something like this code below.

INSERT INTO OUTPUT (Date) SELECT Date FROM REPORTS GROUP BY 1; 
UPDATE OUTPUT SET PREV_7 = (SELECT SUM(Reports) FROM REPORTS WHERE Date BETWEEN OUTPUT.Date -7 AND OUTPUT.Date);
UPDATE OUTPUT SET AFTER_7 = (SELECT SUM(Reports) FROM REPORTS WHERE Date BETWEEN OUTPUT.Date AND OUTPUT.Date + 7);