c) An online music store wants to send out recommendations about
available music downloads to customers who are likely to be very
interested in them. For each of their current customers, they know their
age, gender, occupation and address. Each available music download
has a title, a genre, artist and date of release, as well as a unique
catalogue number. In the past, some customers have provided ratings
for the tracks they have downloaded and the store has roughly 1000
past ratings available.
i. Suppose you are hired to build a prototype system for the store.
What data might you actually use as a basis for making
recommendations? You should explain your answer carefully.
[7 marks]
ii. Describe an approach to constructing a suitable system based
on the data that you identify in part i). You should include a brief
critical discussion of the approach you describe.
[10 marks]
d) What is it about the Naïve Bayes Classifier that is naïve? Your answer
should provide an example of a machine learning scenario where the
assumptions underlying the Naïve Bayes Classifier do not hold.
[10 marks]
e) What is regression? You answer should give a concrete example of
the task and suggest an algorithm that might be used to learn an
appropriate model.
[6 marks]
