discuss: webcounter (was: Re: the tldp foundation)
Subject:
Re: webcounter (was: Re: the tldp foundation)
From:
Stein Gjoen ####@####.####
Date:
10 Jul 2004 12:09:06 -0000
Message-Id: <40EFDCDA.4000407@mail.nyx.net>
Machtelt Garrels wrote:
> On Sat, 26 Jun 2004, Stein Gjoen wrote:
>> http://www.nyx.net/~sgjoen/sample_proc2.html
> To be honnest, this needs a bit of explanation because I don't understand
> this graphic... And how will this check the amount of hits on mirror
> sites?
Sorry to be a bad correspondent here, work has kept me too busy
lately and now I am starting a 2 week holiday...
Anyway:
- the first right column is the web page measured
- ...except from first line "00" that is a page outside scan (elsewhere)
- blue column is number of hits aggregated from that web page
- column left of the blue is hits going FROM that web page elsewhere
- ...except from numbers in beige field which are reloads
- each position right of beige field is to pages down from current line
So to take an example:
line "~sgjoen/disk1.html" has
- no reloads
- 4 hits to the next page "sgjoen/disk2.html"
- 0 hits to the next page "sgjoen/disk3.html"
- 1 hit to the next page "sgjoen/disk4.html"
- 1 hit to page "sgjoen/disk.html"
This then clearly (well, to me at least) shows how people
progress through my web pages.
This is local hit statistics and does not show anything for
mirrors. The principle here is correlation from raw data that
can be from local logs (as in this case) or from web bug stats.
I hope this makes things clearer. Obviously much more can be
done, including some needed debugging. As a prototype it
should at least illustrate my intention.
The most important starting point in the analysis is asking
what kind of question we want to ask. As always premature data
reduction should be avoided.
Regards,
Stein Gjoen