Other Applications: Sequential Dependence Modelling and Data Mining - On Statistical Pattern Recognition in Independent Component Analysis Mixture Modelling - page 164

Table 7.5 Association between learning styles and web activities
Learning style Web event activity

1 Sensory-intuitive Perception Chats, forum participation, course access

2 Visual-auditory Input Chats, forum participation, news reading,

email exchange

3 Inductive-deductive Organization Workgroup document, news reading,

course achievement,
content consulting

4 Active-reflective Processing Email exchange, content consulting,

workgroup document, exercise practice

5 Sequential-global Understanding Course access, agenda using,

content consulting, course achievement

Note that some web activities are associated with more than one dimension; this
makes sense because a web activity could require several capabilities of the stu-
dents to be used in their learning process. By allowing this kind of relationship, we
can obtain more real and versatile descriptions of student learning styles as well as
include all the dimensions of the learning framework. In [ 21 ], just three dimen-
sions of Felder's model were considered and the Bayesian network proposed
constrained the relationship of the web activities with just one dimension of the
learning model.
Figure 7.8 shows the sources 3, 4, and 5 (organization, processing, under-
standing) obtained for the graded graduate course dataset. Four labelled charac-
terised zones in the learning style space are displayed: (1) This zone represents the
most important learning style in the population. The learning for the students in
this zone emphasizes global understanding, active processing, and deductive logic
(natural human teaching style), and high grades. (2) This zone focuses on inductive
logic (natural human learning style), sequential understanding, and relative active
processing. Students within this learning style could have natural skills for virtual
education. (3) This zone is characterised by global understanding, deductive logic,
and reflective processing. Students within this learning style would have higher
abstraction skills that require teaching. (4) This zone basically represents outliers
with individual learning styles.
We can conclude that the understanding dimension facilitates the clear projection
of the learning styles of the students. Its principal components are course achieve-
ment, content consulting, and use of agenda. This finding confirms the assumption
that the quickest way to change the learning style of the student is to change the
assessment style, i.e., expected evaluation biases how the student learns [ 27 ].
We made a cluster validation procedure to determine the best quality of cluster
configuration for the data in Fig. 7.8 . It consisted of estimating the partition and
partition entropy coefficients for different number of clusters [ 28 ], which were
explained in the Introduction chapter. The evolution of those coefficients is shown
in Fig. 7.9 . By applying the criterion to Fig. 7.9 to detect the best partitioning (the
point where the slope of the two curves changes), we found that the optimum
number of clusters is 4.