imbs-hl / ranger

A Fast Implementation of Random Forests
http://imbs-hl.github.io/ranger/
770 stars 193 forks source link

Put ranger into Python? #523

Open PhilippPro opened 4 years ago

PhilippPro commented 4 years ago

Does it make sense to put ranger into Python? I guess more benchmarks would be necessary to prove superiority above the current RandomForest implementation.

mnwright commented 4 years ago

That would be nice. The problem is maintenance capacity... I'm already struggling to maintain the R and C++ version and definitely cannot take care of another one. I'm happy to support a python wrapper in another repository, maintained by someone else.

See also #61 (it's from 2016 and severely outdated so better focus on the discussion).

PhilippPro commented 4 years ago

Entwicklung der Downloadzahlen in RStudio. Trend ist glaub ich eindeutig. ;) (mit cranlogs package erstellt) ranger_entwicklung

crflynn commented 4 years ago

I maintain a scikit-learn compatible python package which binds ranger here: https://github.com/crflynn/skranger

PhilippPro commented 4 years ago

@crflynn That is awesome. Maybe make it more official @mnwright ?

mnwright commented 4 years ago

Looks great! We could link to it in ranger's README.md and the FAQ. Any other ideas?

PhilippPro commented 4 years ago

Yea, I think this is the easiest option. If you are very convinced of the implementation, you could include it also in the repository.