search

MilesCranmer / PySR

High-Performance Symbolic Regression in Python and Julia
https://astroautomata.com/PySR
Apache License 2.0
2.09k stars 197 forks source link

[Feature]: Self-contained install/user interface #461

Open iwelch opened 8 months ago

iwelch commented 8 months ago

Feature Request

I am looking around to try out a few symbolic regression packages. it would be nice to be able to give it a few data points and see what packages can do.

I would also love to ask my students to play with it.

alas, this only works if the set-up is easy --- a webapp (probably not feasible, because you need a lot of CPU), or a binary app. many students run macos these days, too.

just a feature suggestion...

MilesCranmer commented 8 months ago

Thanks @iwelch, indeed I think a web app or self-contained desktop app would be awesome to have.

The closest thing to it is the interactive API here: https://astroautomata.com/PySR/interactive-docs/. But at the end of the day it still requires one to run some Python code (see examples here: https://astroautomata.com/PySR/examples/).

I wonder if there is some scikit-learn graphical user interface that already exists? Surely it would be popular. Since PySR is scikit-learn compatible it can probably just plug in.