GROUP by data by time range in postgresql - sql

Question

I want to GROUP by data by time range. The example I have start_date and end_date, and I want
the separate range between start_date and end_date on 25 range and get sum value from 1 to 25.
Simple presentation of my table:
select * from t1
where time between start_date and end_date
table t1 have:
time 2019-10-01 value 50
time 2019-10-01 value 50
time 2019-10-02 value 50
time 2019-10-02 value 50
time 2019-10-02 value 50
time 2019-10-02 value 50
time 2019-10-03 value 50
time 2019-10-04 value 50
time 2019-10-05 value 50
time 2019-10-05 value 50
time 2019-10-05 value 50
start_date 2019-10-01
end_date 2019-10-25
generate_series function to separate on
2019-10-01
2019-10-02
2019-10-03
2019-10-04
2019-10-05
2019-10-06
2019-10-07
2019-10-07
2019-10-07
2019-10-08
2019-10-09
2019-10-10
2019-10-11
2019-10-12
2019-10-13
2019-10-14
2019-10-15
2019-10-16
2019-10-17
2019-10-18
2019-10-19
2019-10-20
2019-10-21
2019-10-22
2019-10-23
2019-10-24
2019-10-25
and sum by how this 25
for 2019-10-01 to have value 100
for 2019-10-02 to have value 400

Answer 1

I am going to recommend a lateral join:
select d.dt, t.total_value
from generate_series(date '2019-10-01', date '2019-10-25', interval '1' day
) d(dt) left join lateral
(select coalesce(sum(value), 0) as total_value
from t
where t.time >= d.dt and
t.time < d.dt + interval '1' day
) t
on true;
A lateral join can have better performance than overall aggregation, particularly with an index on (time, value).

Answer 2

I understand that you want to generate a list of days, and compute the sum of a column for each:
select d.dt, coalesce(sum(value), 0) total_value
from
generate_series(date'2019-10-01', date'2019-10-25', interval '1' day) as d(dt)
left join mytable t
on t.time >= d.dt
and t.time < d.dt + interval '1' day
group by d.dt
order by d.dt
On dates for which no record is available in your table, total_value will display 0.

Answer 3

Assuming start_date and end_date are variables, you might wanna try the following CTE. It will group by a sum over value by time. In case you want to replace the null values with a 0, try coalesce as pointed out by #GMB in the other answer.
WITH j AS (
SELECT generate_series(DATE '2019-10-01', DATE '2019-10-25', '1 day') AS day)
SELECT j.day, coalesce(sum(value), 0) FROM t1
RIGHT JOIN j ON j.day = time
GROUP BY j.day ORDER BY j.day;
day | coalesce
------------------------+----------
2019-10-01 00:00:00+02 | 100
2019-10-02 00:00:00+02 | 200
2019-10-03 00:00:00+02 | 50
2019-10-04 00:00:00+02 | 50
2019-10-05 00:00:00+02 | 150
2019-10-06 00:00:00+02 | 0
2019-10-07 00:00:00+02 | 0
2019-10-08 00:00:00+02 | 0
2019-10-09 00:00:00+02 | 0
2019-10-10 00:00:00+02 | 0
2019-10-11 00:00:00+02 | 0
2019-10-12 00:00:00+02 | 0
2019-10-13 00:00:00+02 | 0
2019-10-14 00:00:00+02 | 0
2019-10-15 00:00:00+02 | 0
2019-10-16 00:00:00+02 | 0
2019-10-17 00:00:00+02 | 0
2019-10-18 00:00:00+02 | 0
2019-10-19 00:00:00+02 | 0
2019-10-20 00:00:00+02 | 0
2019-10-21 00:00:00+02 | 0
2019-10-22 00:00:00+02 | 0
2019-10-23 00:00:00+02 | 0
2019-10-24 00:00:00+02 | 0
2019-10-25 00:00:00+02 | 0
(25 rows)
EDIT (see comments below):
Changing the series with a 12 hours interval between the generated elements.
WITH j AS (
SELECT generate_series(DATE '2019-10-01 01:30:00',
DATE '2019-10-03 12:30:00', '12 hours') AS day)
SELECT j.day, coalesce(sum(value),0) FROM t1
RIGHT JOIN j ON j.day = time
GROUP BY j.day ORDER BY j.day;
day | coalesce
------------------------+----------
2019-10-01 00:00:00+02 | 100
2019-10-01 12:00:00+02 | 0
2019-10-02 00:00:00+02 | 200
2019-10-02 12:00:00+02 | 0
2019-10-03 00:00:00+02 | 50
(5 rows)
You can change the parameters inside of the generate_series function as you wish, e.g. 30 minutes, 1 hour, etc.
The same can be done with TIMESTAMP, but the dates you'll join with your table need to be identical!
WITH j AS (
SELECT generate_series(TIMESTAMP '2019-10-01 00:00:00',
TIMESTAMP '2019-10-05 12:30:00', '8 hours') AS day)
SELECT j.day, coalesce(sum(value),0) FROM t1
RIGHT JOIN j ON j.day = time
GROUP BY j.day ORDER BY j.day;
day | coalesce
---------------------+----------
2019-10-01 00:00:00 | 100
2019-10-01 08:00:00 | 0
2019-10-01 16:00:00 | 0
2019-10-02 00:00:00 | 200
2019-10-02 08:00:00 | 0
2019-10-02 16:00:00 | 0
2019-10-03 00:00:00 | 50
2019-10-03 08:00:00 | 0
2019-10-03 16:00:00 | 0
2019-10-04 00:00:00 | 50
2019-10-04 08:00:00 | 0
2019-10-04 16:00:00 | 0
2019-10-05 00:00:00 | 150
2019-10-05 08:00:00 | 0
(14 rows)