I've got a table with purchase orders stored in it. Each row has a timestamp indicating when the order was placed. I'd like to be able to create a report indicating the number of purchases each day, month, or year. I figured I would do a simple SELECT COUNT(xxx) FROM tbl_orders GROUP BY tbl_orders.purchase_time and get the value, but it turns out I can't GROUP BY a timestamp column.
Is there another way to accomplish this? I'd ideally like a flexible solution so I could use whatever timeframe I needed (hourly, monthly, weekly, etc.) Thanks for any suggestions you can give!
Grouping by a timestamp column works fine for me here, keeping in mind that even a 1-microsecond difference will prevent two rows from being grouped together.
To group by larger time periods, group by an expression on the timestamp column that returns an appropriately truncated value. date_trunc can be useful here, as can to_char.
This does the trick without the date_trunc function (easier to read).
// 2014
select created_on::DATE from users group by created_on::DATE
// updated September 2018 (thanks to @wegry)
select created_on::DATE as co from users group by co
What we're doing here is casting the original value into a DATE rendering the time data in this value inconsequential.
