Member-only story
Don't Install Python for Data Science. Use Docker Instead!
A beginner's step-by-step guide to start using Docker Containers for Data Science development and avoid complex Python environment managers.
Due to changes in Medium.com policy concerning non-members reads, implemented in Sep 2023, this post is now freely available on geocorner.net: https://www.geocorner.net/post/don-t-install-python-for-data-science-use-docker-instead
Introduction
Docker containers provide a lightweight and efficient way to package and deploy applications, making it easier to move them between different environments, such as development, testing, and production. However, while Docker is widely used for deployment, it has been underutilized by developers for their day-to-day work. Many developers still rely on traditional local development environments that can be difficult to set up, maintain, and share with others. This can lead to issues with version conflicts, dependencies, and different operating systems, which can slow down the development process and make it harder to collaborate with others.
A few months ago, I wrote an article entitled ‘Why You Should Use Devcontainers for Your Geospatial Development’ on Towards Data Science publication, where I discussed the advantages of using VSCode’s Devcontainers for geospatial development over traditional local Python environments. Most of this was motivated by the complex dependencies that exist in specialized geospatial libraries. Since then, I have continued to develop using Devcontainers for most of my work, but I recently had to reformat my notebook, which gave me the opportunity to start fresh. After installing Windows 11 and Ubuntu 22.04, I decided not to install anything related to Python, such as Conda, VirtualEnv, or Miniconda. Instead, I opted for Git and Docker, and as I will show here, we don’t even need VSCode, although I recommend it as a full-fledged IDE if you are doing more than just data analysis.
Since adopting Docker for my development work, I have found it to be a game-changer. By using Docker containers, I am able to quickly and easily set up and manage my development environment without the hassle of dealing with conflicting…
