ARTICLE #6 | 107
DATA MINING FOR LEARNING ANALYTICS
to view separate pages in tabs or know how to download and save
files in their own file stores. This repeated opening of files therefore
is not increasing engagement, it is just the manifestation of their
pre-existing digital practices or perhaps the increased reliance of
having files available on demand online and not stored locally.
Conclusions
It is clear therefore that although interactions with digital resources
can represent engagement for students, there are other factors such
as a student’s prior experience and characteristics of their degree
programme that need to be taken into account. For computing in
particular this will become an increasingly important issue with the
increased focus on programming within the National Curriculum
and students coming into degrees with expected higher levels of
experience and knowledge.
It is also important to note that this study has been performed on
one Level 4 module with a particular structure both in face-to-face
delivery and in the resources that are provided. Models that LA
systems use to measure engagement and progress must be able to
take into account the differences in delivery styles across modules,
degree subjects, teaching teams and universities. Currently, we are
using the same method to analyse student behaviour on a Level 5
programming module at Keele and will be producing classification
models in the form of decision trees that will indicate the likely
trajectory of a learner, given their activity on the VLE. In the first
instance this will allow us to determine how generalisable our
method is across different modules and institutions. The longer
term hope however is that such models will help us to identify early
on those students that can be supported further and offer that
support to them.
References
(1) In effect, the intranet is a bespoke VLE
(2) i.e. the more a student attends, the higher the mark they
achieve
(3) As part of regular teaching activity