Page 80 - Special Topic Session (STS) - Volume 3

STS515 Jeremiah D. D. et al.
Secondly, we employ Kiviat diagrams to visually compare the DS profiles.
Mapping of Requisites
Besides measuring the overall DS programme structure, we will also look
at the instrinsic requirements in terms of teaching DS, using the Machine
Learning course as an example. Machine Learning is chosen because it is a
subject that can be either taught by a Statistics or a Computer Science
department, and is considered a core subject within a DS programme. By
looking at some specific, core concepts and algorithms taught within the
course, we plan to work out some desirable requisites that may be covered by
some prerequisites or early core courses. This will show how a particular DS
design may accommodate the needs for a successful delivery of Machine
Learning.

3. DS Profiling: Findings
4
Using DS programme links listed at various online sources , we score a
number of undergraduate DS programmes on four dimensions, using the
afore-mentioned method. The total weight as the sum of the four scores, gives
an indication about the overall technical intensity of these programmes. The
results are shown in Table 2.
We can visualize these DS profiles using Kiviat diagrams (Kolence, 1973) as
shown in Figure 1. It can be seen from Figure 1 that the DS profiles vary from
each other in terms of sizes and orientations. Maryville and ChineseU both
have envelops of larger sizes, while Auckland and Kansas State etc. give
smaller envelops. Their focuses seem to differ: Maryville leans to statistics
rather than computing, while Canterbury is on the contrary.
Therefore some interesting questions arise. Why is there some significant
diversity as observed in these DS programmes? We do not have further
information in this regard, but we suspect a number of factors may contribute
to the diversity: the subjective understanding of what DS stands for, the
availability and preference of the teaching staff with expertise, the industrial
demands from multiple sectors, etc. Unlike traditional disciplines such as
computer science and electronic engineering, there are no professional
agencies such as IEEE and ACM yet to provide curriculum recommendations
or accreditations. On the other hand, the multi-disciplinary nature of Data
Science will be persistent, and institutions may see the diversity an opportunity
as to promote the uniqueness of their own DS programmes as features
desirable for student recruitment or appealing to particular niches of job
markets. One may see it important to equip DS graduates with skills in
developping new statistical algorithms for new data types; some may find it

4 E.g., https://www.discoverdatascience.org/programs/bachelors-in-data-science/
demands may make DS curriculum design seem quite subjective and arbitrary.
69 | I S I W S C 2 0 1 9