Total suggested work for this week

6 hours lectures (including Python lecture), 1 hour video recommendations, 2 hours pencil and paper self-study, 1 hour computer lab self-study, 2 hours paper and pencil tutorial. Total 12 hours.

1.1 What is Machine Learning?

Start with this! No nonsense introduction to what we are actually doing when we do supervised machine learning using Netflix.

The presenter of this talk, Sabine Hauert, (the real one, not the cartoon) visited Uppsala to give a talk in 2016 about social robots.

I also chose this video to introduce the area because it clearly distinguishes AI and Machine Learning. With a simple example it explains what learning means in the context of Supervised machine learning.

Please beware of the ridiculously overhyped videos out there. All we are doing in machine learning is adjusting weights in a statistical model as outlined in these two videos.

1.2 Introduction and 'need to know' maths

Lecture on Monday 26th 15:15-17:00.

Lecture notes are available here.

1.3 Linear regression: the basics

This slightly patronising, but to the point video (9 minutes) goes through some key points from linear regression as it is used in statistics for a single variable.

This video (1 minute) is a straightforward illustration of the key ideas of multiple variable linear regression.

1.4 Linear regression: machine learning approach

Lecture will be on Tuesday 27th 13:15-15:00.

Lecture notes available here.

1.5 Paper and Pencil Exercises

Start this work on Wednesday 28th October. Spend about two hours attempting the questions before the tutorial on Friday.

The exercise sheet is available here.

Recommended problems: 1.1, 1.2, 1.3
Additional problems: 1.4, 1.5

This will be covered at the tutorial on Friday 30th October 08:15-10:00. Please attempt the questions before the tutorial. Also feel free to ask questions in the discussion section.

Padlet: https://padlet.com/davidwidmann1/90a3sqm6tnf0lfim

1.6 Python Intro Lecture

Due to the fire alarm the Python lecture is moved to Monday 2nd of November 15:15-17:00.

The code is available here:

Lecture notes available here.

1.7 Computer Lab

Start this work on Monday 2nd November. Spend about two hours attempting the questions before the lab session on Wednesday 4th of November 10:15-12:00. In particular, make sure you have a Python environment set up on your computer.

All the computer labs will be in the form of Python notebooks, You can use these either locally on you laptop (I recommend downloading Anaconda) or using Google Colab notebook.

The first lab sessions are available here:

Python notebook
Google Colab

Recommended problems: 2.1, 2.2, 2.3, 2.4
Additional problems: 2.5

Suggested solutions are available here.

Please attempt the questions before the tutorial. Also feel free to ask questions in the discussion section.

1.8 I love this and I want to do more

Completely optional. This is material I found when preparing this lecture that can be interesting. Nothing here is required for the exam.

Here are the most comprehensive online lecture notes (by Brandon Stewart) I have found about the connection between linear algebra and linear regression. I found the intuition in the picture below in particular very helpful.

The World Happiness Report primarily uses linear regression to determine what makes us happy throughout the world. Below are the results of a regression over many different countries.

The biggest limitation of this kind of study is that it does not deal with causation. The machine just learns correlations in the data. It is very important as an aspiring 'machine learner' that you understand how your statistical work ties in to scientific experimental and observational work. A very good example of how this is done can be found in the work of Lara Aknin and co-workers, also in the World Happiness Report.

If you want to investigate the data for yourself, it is available here.