Hour-by-hour day usage extraction

Question

Hey SQLers,
I have a large table that shows the total nr. of connections to my_software (let's call it like this) during 1 month. I would like to create an one-day average view of connections from 00:00 to 23:59; and see the connection pattern to my_software during this past month.
EDIT The date column is defined as: 'YYYY:MM:DD H24:MI:SS'
My SQL skills are not sufficient for this, and I wonder if anyone could point me to the right directions? I believe this must not be very complex, but I'm struggling with the implementation of 'average' in SQL.
Thanks!
M.
Edited by: mcasanovas on Sep 2, 2011 6:04 AM

Answer 1

Hi,
mcasanovas wrote:
I believe I didn't explain myself correctly. I am sorry for that. Whenever you post a problem, post a little sample data (CREATE TABLE and INSERT statements) and the results you want from that data.
Always say which version of Oracle you're using.
Simplify as much as possible. Fior example, you might only include times before 03:00 AM, but explain that your real data (and results) will go all the way to midnight.
Each row of the table is one connection, and I've got plenty. One of the columns is the date+hour, and it indicates the exact moment of the acces. In a more practical way, I need to be able to create a graph that shows the average per-hour (from 00 to 23) for the whole month. The result of the select should be something like:
--AVERAGE CONNECTIONS DAY
acces_time average_connections
00:00 550
01:00 340
22:00 1200
23:00 1130
Where the first row is the average nr. of connections at 00:00 during the whole month.Let's say there were 31 days in the month. Are you saying that you'd want output like this is, for example, there were
500 connections between 00:00 on 15 of those days,
550 connections between 00:00 on 1 day, and
600 connections between 00:00 on 15 days?
Here's one way:
WITH     parameters     AS
     SELECT     DATE '2011-08-01'     AS start_date
     ,     31               AS total_days
     FROM     dual
SELECT       TO_CHAR (connection_time, 'HH":00"')     AS hour
,       COUNT (*) / total_days          AS avg_connections
FROM            table_x
CROSS JOIN     parameters
WHERE       connection_time >= start_date
AND       connection_time <  start_date + total_days
GROUP BY  TO_CHAR (connection_time, 'HH":00"')
ORDER BY  hour
;In this query, the starting date and the number of days are hard-coded into the sub-query. You might use variables (such as SQL*Plus substitution variables) instead. With a little more coding, you could enter just the starting date, and have the query derive the total number of days in that month.
This assumes that connection_time is a DATE, not a string. There is no reason to store dates in string columns.