Introduction to Machine Learning

Here are the answers to the quiz.

Supervised, unsupervised or reinforcement learning?

1. You want to identify the different distinct groups of people who are turning up at hospital A&Es with COVID. Understanding this case-mix will help the hospital to better respond to the needs of these different types of patients. [Unsupervised learning]

2. A book distributor has a collection of books, which it has classified into different categories, e.g. ‘Young adults’, ‘Biography’, and ‘Horror’. It wants to use this information to build a system to classify its new products automatically. [Supervised learning]

3. You work for Spotify and want to group together similar songs based on their characteristics (e.g. their tempo and their lyrics). You want to do this so you can create playlists with similar songs and suggest them to users. [Unsupervised learning]

4. You are trying to train a robot to open your fridge and bring you an ice-cold 0% Heineken beer. [Reinforcement learning]

Classification or regression?

1. You want to predict the number of people who will arrive at a hospital with Covid-19 the next day. You can use information about how many people have tested positive and historic information about how many positive tests have resulted in people arriving at hospital. [Regression]

2. You are trying to predict whether an individual has COVID based on their current symptoms. You can use historical data from people have previously tested positive or negative and what their symptoms were at the time. [Classification]