Our great sponsors
-
cookiecutter-data-science
A logical, reasonably standardized, but flexible project structure for doing and sharing data science work.
-
WorkOS
The modern identity platform for B2B SaaS. The APIs are flexible and easy-to-use, supporting authentication, user identity, and complex enterprise features like SSO and SCIM provisioning.
I have to set up a GitHub repo for an upcoming project and was researching some data science templates to follow. I came across cookie cutter and this template by drivendata: https://github.com/drivendata/cookiecutter-data-science
I would say yes because you won’t know if the newer model has better or worse performance until you test it so you may elect to keep the older one. As your data changes and the model drifts one of the previously trained models might begin to perform better. You also will likely want to switch approaches and may have several different implementations. I suggest checking out MLFlow and setting up a model pipeline and registry. https://mlflow.org/
I also thought that the drivendata template is a great starting point, but is too comprehensive for my needs. Ended up putting together a quick and dirty version that I gradually update as my needs evolve (https://github.com/tnwei/cookiecutter-datascience-lite/). Otherwise I concur w/ the sentiment to just take what works for you and evolve upon that
Related posts
-
aim VS cascade - a user suggested alternative
2 projects | 5 Dec 2023
-
cascade alternatives - clearml and MLflow
3 projects | 1 Nov 2023
- intelligent-trading-bot: NEW Other Models - star count:567.0
- intelligent-trading-bot: NEW Other Models - star count:567.0
- intelligent-trading-bot: NEW Other Models - star count:567.0