Author : Mike
Date: 2003/05/23 16:27
Thanks. For comparative analysis with large logfiles, I tend to do one of the following :
1) filter out everything I'm not interested in beforehand (e.g. with grep whatever
access_log | perl awebvisit.pl), or
2) adapt the script itself (e.g. to ignore all sessions except those with the entry point
somepage.html), and in general
3) run the script once on the old logfile, save the results, and run it again on the new
logfile (in non-cumulative mode). That way, you'll have the overall statistics for both
periods, and you can e.g. run awv-map1.cgi with the old stats file, and awv-map2.cgi with
the new stats file, and
put them side by side.
For longer periods of time, you can run the script daily in cumulative mode, and the day
you change your navigation, save the current stats and data somewhere, and start a new
series - that way you'll be able to compare them too. But this will take more disk space
and memory when it runs of course, because it has to remember past visit tracks through
your website.
For the comparison itself, the "best" way I found (or at least the one most likely to give
you somewhat meaningful results) is still a simple side-by-side comparison, I'm afraid. A
more automated comparison is more likely to mislead you rather than help you - at least
that's my experience... :)
|