DEV Community

Cover image for universal data science template for python and jupyter (for now)
Lorenzo Tinfena
Lorenzo Tinfena

Posted on

universal data science template for python and jupyter (for now)

Hi everyone! I'm new in this community.

I have little experience with python and data science in general but i would like to start with a strong foundation so i decided to try doing this little project.

Now if someone that want to do some algorithm in datascience, like machine learning with Tensorflow (in this case I use python language), he has to set up an environment, "classic" with Python stock, and install pip, and with pip install dependencies. But for virtual environments exist conda, that is beatiful! But has some limitation, like cross platform, and collaboration, (suppose that your friend try your code with a single library version different), and you have to choose between "conda packages" or "pip/pyenv packages"? And exists also other environment managing tools...

The project is based on VSCode, and VSCode remote container extension, that uses docker (that for windows10 use WSL2)

I would like to make this project with these main points:
Scalable, Generic (not python environment, or conda environment, or R environment), simple (simple -> scalable, just better xd), complete, cross platform, updated, and collaborative (you can share your project to everyone and everyone has your exactly enviroment)

Link to project:

Now is very very basic, and the main part for now is the idea, but if someone want to take part, is welcome :)

Top comments (2)

stereobooster profile image
stereobooster • Edited

Oh that's nice. I used Docker previously, but without remote containers. I prefer jupyter/scipy-notebook though

lorenzotinfena profile image
Lorenzo Tinfena

Oh well, I was exactly in your situation :) but program in jupyter (web server )or jupyterlab i think is not the best now, the debugger is not in official python kernel, intellisense is not good as that in vscode with intellicode, and the i don't like the interface, and then I discover remote container, that i think has a lot of advantages, also for scalability. Now jupyter for vscode and remote-extension are relatively new softwares, so theg have to be imrpoved, but i think they have a future. if you want to tell your opinion fell free to do :)