Searching over warm_start in SGDClassifier, not tuning learning rate

rhiever / sklearn-benchmarks

A centralized repository to report scikit-learn model performance across a variety of parameter settings and data sets.

MIT License

210 stars 54 forks source link

Searching over warm_start in SGDClassifier, not tuning learning rate #19

Closed amueller closed 8 years ago

amueller commented 8 years ago

Searching over warm_start in SGDClassifier is not meaningful. Also, using a constant learning rate and not adjusting the rate will probably not work. the "optimal" rate will only work with scaled data.