Course software requirements

Python

Make sure you run Python 3. Check by starting the interpreter, for instance

$ python3
Python 3.9.7 (default, Sep 24 2021, 09:43:00)
[GCC 10.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>>

On some systems, the default python is python3.

Minimal local setup

The minimal set of Python packages you need to install are listed in requirements.txt. Install them on your local machine, e.g. using your system's package manager (say apt on Debian or Ubuntu GNU/Linux).

If you use the pip Python package manager, then you may want to create a virtual environment in case you need to separate packages from the rest of your system.

$ python3 -m venv course_env && . ./course_env/bin/activate

Install packages:

(course_env) $ pip install -r requirements.txt

Note that we install tensorflow-cpu by default. If you have a GPU, then you may use tensorflow instead, which will utilize that. But all examples we use are very light weight, so the CPU version will suffice.

You can now run

(course_env) $ ./check_install.py

which should print

numpy                ... ok
matplotlib           ... ok
sklearn              ... ok
pandas               ... ok
seaborn              ... ok
tensorflow.keras     ... ok

Anaconda distribution + Conda package manager (local)

Instead of using your system tools or pip, you can install Python itself and packages using the Anaconda distribution. Once installed, you can install more packages using the Conda package manager which comes with Anaconda. This might be the easiest option on Windows systems. More details can be found here.

Google Colab (cloud)

You can use a free cloud-backed Jupyter notebook running on Google Colab. Using that, most of the packages we require are already installed. You can install missing ones by following the Google Colab documentation.

Interactive environment

Some of the course code material is presented in form of Jupyter notebooks and many people like to use the JupyterLab or Jupyter Notebook interactive environments to run them (e.g. Google Colab Jupyter notebook).

But this is not required. If you don't have/want a Jupyter environment, you can also execute all code from the lesson notebooks

• in your normal Python interactive shell (terminal based, start python or python3)
• using the more powerful ipython interactive shell (start ipython or ipython3)