Lab exercises for COMP0088 Introduction to Machine Learning

TL;DR

Install Git (if you don't already have it)
Install Python 3 (if you don't already have it)

Clone this repository to your local machine:

git clone https://github.com/comp0088/labs.git

Create and activate a virtual environment:

cd labs
python3 -m venv .venv
# On Unix/MacOS:
source .venv/bin/activate
# On Windows:
.venv\Scripts\activate.bat

Install Python package requirements:
```
pip install -r requirements.txt
```
Read the week's lab exercises document lab_N.pdf
Add your code to the script week_N.py
Run the script from the command line to see your code in action:
```
python week_N.py
```

About

This repository contains lab exercises for the COMP0088 Introduction to Machine Learning module for taught MSc students at UCL, delivered in Autumn 2021. Exercises are designed to be attempted in the on-campus lab sessions on Thursday mornings, though you are free to do additional work in your own time if you wish.

Lab attendance will be monitored, but the exercises are not graded. They are intended as learning experiences, to help you understand and apply different machine learning algorithms. You are welcome to discuss and help each other with these tasks and to ask for assistance and clarification from the TAs, but there is nothing to be gained by simply copying each others' work.

Exercises for week N are specified in document lab_N.pdf. Skeleton code for the exercises is provided in the script week_N.py. You should add you solution code to this file. The script can be run at the command line like this:

python week_N.py

Each week's script will print some progress messages and generate an output file called week_N.pdf containing a number of plots. An example showing (approximately) what the finished output should look like is included in the exercises document.

In addition to the spec and script for each week, there are a few other files in the repo:

README.md: this file.
plotting_examples.py: a script containing a few examples of plotting with Matplotlib for those unfamiliar with this.
utils.py: a library of utility functions that are used by the script code, or are available to be used by your own solution code.
requirements.txt: a list of additional Python packages to install.
LICENSE: text of the MIT License that applies to all code and documentation in this repository. (Summary: in the unlikely event that you have any reason to do so, you are free to reuse this material for any purpose you like.)

Cloning & Updating

In order to copy the repo you need to have a working installation of the Git version control system. If you don't already have this, you can download it here.

Choose a convenient location for your working directory and download an initial copy of the repository with git:

git clone https://github.com/comp0088/labs.git

New lab exercises will be added to the repo each week. You can download updates with the following commands:

git fetch
git merge origin/main

Note however that git may report issues when you try to merge our upstream changes with your own if you have uncommitted changes in your directory, or if you have made changes that conflict with changes we have made in the main repo.

We will do our best to avoid making any changes that are likely to cause conflicts. You can generally assume that the week_N.py files will not be updated and can be freely edited and committed. (You should not assume this about utils.py or requirements.txt -- try to avoid editing these files if you can.) If conflicting changes do become necessary, for example to fix significant errors in one of the supplied scripts, we will announce it on Moodle and explain what you'll need to do about it.

We recommend using git to track your own changes as you work on the exercises. Commit your work at appropriate intervals and only fetch/merge new changes when your own changes are up to date.

However, if you have made changes that you don't want to commit for some reason, you can also temporarily get them out of the way using git stash:

git stash push
git fetch
git merge origin/main
git stash apply

Python Setup

The exercises require a local installation of Python 3, along with a number of additional packages for numerical programming, plotting and machine learning. We suggest using the latest stable release of Python (currently 3.9.7) from python.org. If you know and prefer Anaconda/Miniconda you are welcome to use that instead.

(Although we recommend the latest Python, the code has also been tested on Python 3.6.8, which is the version currently installed on some of the CS lab machines. It is possible, albeit suboptimal, to set up and run the exercises on one of those machines via SSH.)

Virtual Environments

The package requirements for the lab exercises are pretty vanilla, but we strongly recommend working in a dedicated virtual environment in order to avoid any conflicts or compatibility issues with any other Python work you may be doing.

There are several options for how and where to set up such a virtual environment. If you already have experience doing so then feel free to use any configuration you are comfortable with. If you haven't done this before and/or would rather not think about it, follow the default setup instructions below.

Default virtual environment setup

A straightforward way to configure your virtual environment is to store it in a hidden subdirectory of your working directory (ie, the directory containing this repository). Once you have cloned the repo as described above, change into the directory:

cd labs

Initialise a new virtual environment:

python3 -m venv .venv

(The name .venv is a reasonably common convention that should be recognised by Python-aware editors such as VSCode and PyCharm, but you can use a different and more informative name if you wish. In that case, also replace .venv with your chosen name in the commands below.)

Make the virtual environment active. The command for this varies depending on your operating system and shell.

On Unix-esque systems (Linux and MacOS), run:

source .venv/bin/activate

(If you are running a variant of the C-shell csh rather than bash or zsh, you should instead source .venv/bin/activate.csh.)

On Windows, run:

.venv\Scripts\activate.bat

(If you are running PowerShell, you should instead run .venv\Scripts\Activate.ps1)

When the virtual environment is active, your commmand prompt will be modified with the prefix [.venv].

Installing Required Packages

With your virtual environment active, you should be able to install all required packages using pip, like this:

pip install -r requirements.txt

Working Environment

If you are an experienced Python coder with a preferred development environment, you should use that. Note, however, that the exercises are not really designed to be run within Jupyter Notebook. You will likely be able to if you really insist, but in general there will be no benefit to doing so and it probably won't be very convenient.

If you don't have any existing preference, we recommend using a Python-aware editor such as VS Code (be sure to install the Python extension) or the PyCharm IDE (the Pro version is free to students), together with an IPython interactive session running in a terminal window for testing and debugging.

Feedback

Please post questions, comments, issues or bug reports to the COMP0088 Moodle forum or raise them with the TAs during your lab sessions.

MichaelKim39/labs