Question

I have a table in my postgresql 8.4 database like this:

id(serial), event_type_id(id, foreign key), ts(timestamp)

I have two event type id’s that I need to see the age between.

Event_type_id 1 = Arrive
Event_type_id 2 = Leaves

example data

 id | event_type_id |             ts             
----+---------------+----------------------------
 21 |             6 | 2012-04-03 16:02:18.739274
 20 |             5 | 2012-04-03 08:44:13.244287

I want to group these by day and see how many hours per day it is.

can someone point me in the direction how to group and calculate the age?

So far I have this, but I need to generate a series to join on or something similar

Solution
Don’t know if it’s best practice but it works for me.

SELECT dx, age(max(ts), min(ts)) 
from generate_series('2012-04-01', '2012-04-14', interval '1 day') as dx 
left join events on (ts >= dx and ts < dx + '23 hours 59 minutes' and event_type_id in (5,6)) 
group by dx 
order by dx;

Answer 1

I think you need:

select age(max(ts), min(ts)), ts::date as date
from events
where event_type_id in (5,6)
group by date;

But your question is not clear. I assume you mean that you have a table (not a database) like this:

create table events (
  id serial,
  event_type_id int references event_types,
  ts timestamp
);

And you need to calculate an interval between first and last event of type 5 or 6 for every date.