Skip to content

Latest commit

 

History

History
139 lines (84 loc) · 5.57 KB

python-installation.md

File metadata and controls

139 lines (84 loc) · 5.57 KB
Raw

A clean Python installation using conda

(as of May 2023)

It is finally the moment to move from the safe harbours of Google Colab into the perilous waters of local Python installations!

For now follow the instructions below - but don't be scared, it will all make sense in the end!

Note: a local Python installation means a Python interpreter running Python instructions on the CPU of your computer (this is different from running Python instructions on a remote server, which is what we have been doing with Google Colab)

Before you start

Do you actually need a new Python installation? Pick the option that best describes you:

  • I have never used Python on my computer before: you need a new Python installation. Continue to Python installation

  • I remember doing something with this Python thing a while ago on my computer, but I don't remember what. We'll need more information:

    • I installed Anaconda a while back: you probably want to uninstall it and start from scratch. To uninstall it, follow the instructions here.
    • I used Python to run some important code which I might need again: Contact me, we'll figure out what to do to keep your stuff running.
    • I installed Python but not through Anaconda: contact me, we'll figure out what to do to remove it

  • I have a Python installation on my computer, and I know the basics of how to run it locally: you can skip this installation. Just make sure you have the following packages installed for now:

    • numpy
    • scipy
    • matplotlib
    • pandas
    • scikit-learn

    And that you can run Jupyter notebooks.

Python installation

We will be managing our local Python installation using conda. conda is a package manager, which means it is a program that will help you install and update Python and its libraries. It is also a virtual environment manager, which means it will help you create and manage different Python environments - more on this in the next lectures!

0. Download and install conda

Go here and dowload the latest installer for your operating system.

Screenshot 2023-05-22 at 21.35.05.png

After the download, run the installer and follow the instructions. The only option I recommend checking and eventually changing is the installation path. I recommend installing conda in your home directory, so that it is easy to find and manage. Make sure it is directly under /Users/your-username/ (Mac) or C:\Users\your-username\ (Windows).

Test your conda installation

Open a terminal (Mac; you can search it with the OS search) or an Anaconda prompt (Windows; you can search it with the OS search) and type conda. If you see a list of commands, you are good to go! If you get something like conda: command not found, ask for help!

1. Install jupyter

We will be using Jupyter notebooks to write and run our code. To install it, type the following command in the terminal:

conda install jupyter

This will install Jupyter and all the packages it needs to run. If you are asked to confirm the installation, type y and press enter. It can take a bit!

Test your Jupyter installation

Type the following command in the terminal:

jupyter notebook

You should see a browser opening with a Jupyter notebook. If you do, you are good to go! If you get something like jupyter: command not found, you might want to close the terminal and open it again. If the problem persists, ask for help!

Note: After you open a notebook, you won't be able to work on the terminal you have launched it from! For the following steps, first close the notebook! (by first closing the webpage, then pressing Ctrl+C on the terminal).

2. Create a new Python environment

We will now create a new Python environment. This is a good practice to keep your Python installations clean and tidy.

From the terminal you opened for testing, write the following command:

conda create -n course-env python=3.9

This will create a new Python environment called course-env and install Python 3.9 in it.

Test your new environment

To activate your new environment, and check that it is working, type the following commands:

conda activate course-env
python --version

This should print Python 3.9.something. If it doesn't, ask for help!

3. Install the packages we need

We will now install the packages we need for the course. In the terminal you opened for testing, copy the following commands:

conda install numpy scipy matplotlib pandas scikit-learn scikit-image ipykernel
ipython kernel install --name=course-env

Debug tip: if the above line is failing with an authorization/privileges error, try to add the --user argument:

ipython kernel install --name=course-env --user

Test your packages

Close and reopen the terminal (or command prompt) - or type conda deactivate (we need to run jupyter notebook from the base environment), then type the following command:

jupyter notebook

This should open a browser with a Jupyter notebook, something that looks like this:

Screen notebook

Click on the New menu and check that you see course-env in the list of available kernels: Screen notebook

Select course-env and create a new notebook. In the first cell, type the following code:

import numpy as np
import matplotlib.pyplot as plt

plt.plot(np.random.randn(100))