Sequence Rules for Web Clickstream Analysis - Advances in Data Mining - page 3

at 0 the midnight of the preceeding day). More importantly for our analysis, such
dataset contains binary variables that describe whether each page is visited at least
once (modality 1) or not (modality (0). Finally, there is a bianry variable, named
purchase, that indicates whether the session has led to (at least) one e-commerce
transaction.
Table 2 shows part of a row from the visitors dataset, that corresponds to the
session in Table 1.

Table 2. The derived dataset.

c_value c_time length clicks time home

70ee683a6df… 14OCT97:11:09:01 24 5 11:09:01 1

c_value catalog addcart program product

70ee683a6df… 1 0 1 1

While the rows in Table 1 correspond to clicks, in table 2 the rows correspond to
sessions (or, equivalently, visitors, as they are unique). There are as many rows as the
total number of visits to the web site. In particular, looking at the last five columns, it
is obtained a binary data matrix that expresses which pages, among the 36 considered,
have been visited at least once in each session.
To give an idea of the type of considered pages, we now list some among the most
frequent ones.
HOME: the home page of the web site;
LOGIN: where a user has to enter its name and other personal information, during
the first registration, in order to access to certain services and products, reserved to
the customers;
LOGPOST: prompts a message that informs whether the login has been successful
or if it has failed;
LOGOUT: on this page the user can leave the personal characterisation given in
the login page;
REGISTER: in order to be later recognized, the visitor has to prompt a userid and
password;
REGPOST: shows the partial results of the registration, asking for missing
information;
RESULTS: once the registration is accomplished, this page summarizes the
information given;
REGFORM1: here the visitor has to insert data that enable him/her to buy a
product, such as a personal identification number;