Login Chart

artist has asked for the wisdom of the Perl Monks concerning the following question:

Replies are listed 'Best First'.
Re: Login Chart by Joost (Canon) on Jun 13, 2005 at 16:50 UTC
Dear artist. I'm confused. What do you mean with the "nature" of the user, and what does the "Login Chart" have to do with it? Do you mean anything like awstats? "What should it profit a man, if he should win a flame war, yet lose his cool?"	[reply]
Re^2: Login Chart by artist (Parson) on Jun 13, 2005 at 18:02 UTC
It's definitely possible to be confused. I have a log file with 2 columns: time\|login Only from this information, I like to see the login behavior of particular users. These users are paid customers. Finding behavior patterns are not easy. Someone with experience can tell easily. Awstats is fine, but not useful for me. Awstats is about the website traffic rather than details of individual activity.	[reply]
Re^3: Login Chart by davidrw (Prior) on Jun 13, 2005 at 18:21 UTC
So you want time parts? what kind? What's an example of a pattern an experienced person would find? Either way, maybe start by parsing the time with on of the Date modules (i personally like Date::Calc once you have the date pieces). Depending on what you want, you could hash it up on hour, or hour range (e.g. early morning, morning, early afternoon, etc), or on day of week and see what the buckets look like. Depending on your data, it might be very cool to convert the time to a epoch value and run some Fourier Transforms (FFT) against it and see what frequencies pop out.	[reply]
Re: Login Chart by poqui (Deacon) on Jun 13, 2005 at 21:46 UTC
If you have to ask "Does a dog have buddha nature?" you will never know.	[reply]
Re: Login Chart by TedPride (Priest) on Jun 15, 2005 at 04:35 UTC
So you want to interpret the "nature" of the user just from their login info? About the only thing you can find out from that is whether they use the same IP all the time or not. Can you please clarify why you're trying to do this?	[reply]
Re: Login Chart by TedPride (Priest) on Jun 13, 2005 at 18:54 UTC
I don't know what you mean by "nature", but you can probably determine interests by how long they spend on pages of each topic. You need to make a list of all the URLs on your site and categorize each according to whatever topic area they would correspond to, then estimate time spent on each page, where possible, and add together the times for each topic area and sort by time spent. Also of interest might be average time spent per page and a flowchart of the user's path through the site, and for overall information, it's very useful to know which links people take to visit other pages. For each page you should have a frequency list of links followed to that page, and links followed out of that page. I'm not going to write it for you, though :)	[reply]
Re^2: Login Chart by artist (Parson) on Jun 13, 2005 at 19:04 UTC
As I said, I just care about the login part and not anything else. So, it's not about page viewing or topics of the sites etc.. My log file has only 2 columns. time in epoch seconds and the user id.	[reply]