search

8080labs / ppscore

Predictive Power Score (PPS) in Python
MIT License
1.12k stars 168 forks source link

Model pipelining support #50

Open HectorBarrio opened 3 years ago

HectorBarrio commented 3 years ago

Hello, pulling into "dev" branch, I hope this is the intended branch.

FlorianWetschoreck commented 3 years ago

Thank you! I will review this in the following days :)

FlorianWetschoreck commented 3 years ago

Thank you for providing the PR.

Can you please create a sample notebook which shows and explains the usage of the new API? The focus is on showing in which exact scenario you would use the new API and how it is useful compared to an alternative solution

HectorBarrio commented 3 years ago

Hello Florian, let me know whether you can access this: https://view.datalore.jetbrains.com/notebook/2VvLnxXJU9c70cR774ufgz?force_sso=true

If not: https://github.com/HectorBarrio/ppscore/blob/Model-pipelining-support/examples/PPS%20Model%20Pipelining.ipynb

FlorianWetschoreck commented 3 years ago

Thank you for providing this - the github.com link worked - datalore did not. I will think about this in the following days :)

FlorianWetschoreck commented 3 years ago

Hey Hector, thank you for providing the demo notebook. That was very important for the actual decision because we needed to see the full thing in a live context.

Currently, we want to keep the package short and sweet without integrating pipeline support at this time. This might change if there are other good examples and more demand. Also, people can use the code from your github repo in the meantime if they really want to

Also, feel free to create a separate repo where you add more features on top of the base ppscore package e.g. including the multiprocessing support that was proposed in another branch

HectorBarrio commented 3 years ago

Ok Florian.