daenuprobst
Models
A little bit of an ambitious pull request... It would be nice to have more structure and a more compact main.py.
- moved classifiers and regressors into separate files (TODO: add some standard models to benchmark against).
- created a model class that can help select models for benchmarks
- added a simple argsparse to main.py
- started configs.py to store the settings
Sorry if I missed any recent changes; I ran the benchmark and can reproduce the results (I made gzip and smiles the defaults).
I can break it up into smaller pull requests. Feel free to let me know if you would rather try something else.
Hey could you test if you still get the same numbers with all the new changes that happened in the meantime?
I agree it might make sense to break down the script into smaller parts. @daenuprobst what do you think? tagging you because modifying the bench marking script is obviously important :)
Hey... definitely makes sense to breaking up the script (or maybe even a separate branch for now?). @EricBoittier, could you confirm whether the results are still the same?
