Database Reference
In-Depth Information
The amount of work must justify the amount of time allocated. The project specifi-
cation given in section 2 is meant for a full data mining module on a computing pro-
gramme. The scope of the project must be adjusted accordingly if the module has a
different emphasis or it is for a different programme. For instance, a module on ad-
vanced databases with a significant part on data mining may require a much smaller
scope. Everything within the project, such as the data set dimensionality, the amount
of data preparation work and the number of data mining tasks, should all be limited.
The project group size may even be increased to 4 students rather than 2 or 3 to fur-
ther spread workload.
Student Feedbacks
A thorough survey of student feedbacks on the project over years has not been con-
ducted yet mainly because of small class sizes. However, from the annual student
feedback questionnaires on the module in which a question about the coursework is
asked, comments on the use of the project are overall positive. Most students consider
the project experience useful, and it has helped enhancing the understanding of the
subject. At the same time, many students consider the project harder than practical
projects they have done for other subjects. Weka as a tool for supporting the project
has also received good feedback for its simplicity and ease of use.
Some cautious feedback comments are also received. Due to the lack of business
context and short of domain experts, some students cannot see how useful result pat-
terns can be, which in turn affect the level of their confidence towards the patterns
they have discovered. The biggest difficulty is how to deal with the uncertainty. This
could be a cultural shock for computing students who have worked largely towards
deterministic solutions. Some students also have difficulties in data understanding and
pattern evaluation due to lack of training and understanding of basic statistics.
A small minority of students have very little clues about the project. They tend to
make bad judgement and rely heavily on trial and error or hand-on guidance from the
tutor. Their comments towards the project are quite negative. Effort is certainly
needed to find out how to minimise the size of this group.
Resource Implications
Running the project may require additional resources in terms of tutor's hours. En-
hancing student learning experience has cost. Indeed, no pain, no gain. In the author's
experience, however, the extra hours spent on administrating/supervising the project
is only marginally more than that for a database design project, thanks to the aca-
demic maturity of the final-year students. The good students usually need very little
guidance. They can obtain the required knowledge from lectures and reference mate-
rials. Weak and fair students in fact take up most of the consultation time. Whether
the project idea can scale up to classes of hundreds of students is yet to be verified.
5 Concluding Remarks
This paper argues that a data mining mini-project should become a critical part of the
coursework for a data mining module. The project should benefit student learning
experience of this interesting subject in the same way as a database design project
