Skip to content

add randomized search of hyperparameters to sklearn classifiers

Patrick Godwin requested to merge sklearn_randomizedcv into master

This essentially takes care of reed.essick/iDQ#40.

Changes:

  • Adds randomized cross-validation to sklearn classifiers, can sample from either a uniform or a log-uniform distribution.
  • Update docs to show how this is used in classifiers.py
  • Kept grid-based CV and changed name for grid from cv_grid to param_grid to match better sklearn API.
  • Added log_uniform distribution in utils for sampling. This is used within RandomizedCV for hyperparameter sampling.

Merge request reports