search

rhiever / sklearn-benchmarks

A centralized repository to report scikit-learn model performance across a variety of parameter settings and data sets.
MIT License
209 stars 53 forks source link

Docs for new classifier #43

Open elyase opened 4 years ago

elyase commented 4 years ago

What would be the easiest way / best starting point to test a new classifier using the same testing procedure/datasets as in the paper?

rhiever commented 4 years ago

Yes, unless you want to rerun everything. Might not be a bad idea to to rerun everything just to be thorough.

elyase commented 4 years ago

@rhiever I might rerun everything, can you give me a rough idea on how much computing it would need? But my original question is, if I want to test a new classifier, where should I start?

rhiever commented 4 years ago

All of the parameter tuning code is in these directories: https://github.com/rhiever/sklearn-benchmarks/tree/master/model_code

and the code to analyze the results of those searches is in this notebook: https://github.com/rhiever/sklearn-benchmarks/blob/master/notebooks/analyze-sklearn-benchmark6.ipynb