Under the hood

Philosophy

Log(psycho)Analyst's philosophy is to make the user life as easy as possible. Two major reasons:

not all users are webmasters or highly skilled in computers, but have a web site;
easy configuration and use allow to reduce the user's questions (so we can use the saved time to enhance the software)

Thus, just after downloading the setup program, you will be asked for some information (e.g. your domain name), then all the possible information is collected automatically. Scheduled tasks are automatically created to process the log files. And it's all! Log files are then retrieved automatically, as well as databases and software updates.

Differentiation between human visitors and robots

We recognize more than 750 user-agents, and classify them into human visitors and robots users. This differentiation is really important to do, because humans and robots behavior is clearly not the same:

Robots usually get only about one page per visit, while humans read more;
A lot of information is unknown in robots (e.g. plugins, screen size, language), thus merging robot and human stats increases the percentage of unknown information.
Robots do not produce direct benefits for you (they don't buy your products)
...

This allows us to generate statistics more precise than other log analyzers tools. We recognize even robots that are spoofing well known HTTP user-agent strings such as Microsoft Internet Explorer or Mozilla Firefox.

Proxy identification

A lot of visitors are now surfing the web via proxy servers. Two major proxy types exist: single proxy and multiple proxy. When using single proxies, several different human visitors are browsing the web with the same IP (this is typically the case for transparent entreprise proxies). Contrarily, multiple proxies allow the visitors to spread the visits requests (images, HTML pages, PDF documents, etc.) on several IP addresses.

If these proxies are not identified, you will get less visits than in reality for single proxies, and more visits than in reality for multiple proxies.

Log(psycho)Analyst recognizes both types of proxies. This ensures that your statistics will be the closest to the reality. This explain why you can get different results with our tool than with other log file analyzers.

Processing speed

Log(psycho)Analyst processing speed can vary depending on the amount and type of information to process. Usually, we can process about 2000-3000 log lines per second, which is about the average speed of other tools. However, you should remember that a lot of graphics and reports are generated, so this speed can be considered as very good.

We have also tested our fast reverse DNS feature on a modem line and got an incredible speed of 1000 log lines per second!

Libraries used

We use the following LGPL open source libraries:

JFreeChart
JRobin
Jung
JDom