FastAPI-template VS cookiecutter-data-science

I'm pretty partial to s3rius's template boiler plate. Used it while contracting a while back and it just tackled everything I'd want for the first few months leaving me time to focus on CRUD / business logic instead of Devops stuff. Do wish it used ruff but that was an ultra easy swap.

I found myself rebuilding a lot of basic stuff such as sign up, login, async database functionality and unit tests every time I wanted to quickly POC something. Now I like to keep things simple and customisable when starting and thus usually don't opt for using one of the already existing templates (such as this great generator: https://github.com/s3rius/FastAPI-template) since I feel I spend more time cutting stuff out I don't need.

I opened an Anaconda cmd window and ran `cookiecutter https://github.com/drivendata/cookiecutter-data-science ` . I answered all prompted questions. After searching for a while I found where the project folder was created. However, how do I get this on GitHub? The only thing I can figure out is to create a brand new repo on GitHub with the exact same name, open it in GitHub desktop, click "show in explorer", and then drag and drop all files from the Cookiecutter folder into the GitHub Desktop folder. However to me this does not sound like the intended way to create a new project and put it on GitHub.

I'm not sure what "data scraping" means exactly, but for data science generally I think this is a pretty good template: https://github.com/drivendata/cookiecutter-data-science

Perhaps this cookiecutter template will help: https://github.com/drivendata/cookiecutter-data-science

I've found https://github.com/drivendata/cookiecutter-data-science as a guide, but haven't found any repos that solve a problem end to end actually use it. Are there any good repos or resources that exemplify how to solve a DS/ML case end-to-end? Including any UI (a report, stream, dash etc) needed for delivery, handling data, preprocessing, training and local development.

Project don't need to be 100% original, do the project in a such a way that other people has not done yet. There are plenty of datasets and notebooks available on Kaggle. Those are just bunch of notebooks. Take the inspiration from the notebook and build the project in modular structure and organize your project in proper folders and modules. I am using this cookiecutter for building my portfolio projects. https://github.com/drivendata/cookiecutter-data-science

Hi all,I am working on ML project and have created project using the cookiecutter-data-science scaffolding. The structure of the project somewhat looks like this,

While data science is not SE, it's fundamental to have some structure in your projects since you want the work to be somewhat reproducible. I recommend you start here https://github.com/drivendata/cookiecutter-data-science Since it's a cookie cutter it will be easier to implement at first since they can create the structure by running a short command, after some time you will tailor it to your specific company needs :) For notebooks it's kind of hard, they can't be peer reviewd that easily since cells are editable even after code has been run, keeping the old result... I recommend tools like deepnote, but I'm not sure how well they work for collaboration in notebooks because I never used them yet, I just know they are working on solving these problems. I hope these things help!

Personally I like using cookiecutter’s data science project template. It is easy to set up and has a clear structure. Here is their github: https://github.com/drivendata/cookiecutter-data-science