skopt is a python module for automatically performing bayesian optimization on scikit-learn machine learning algorithms.
skopt has been tested to work with the following packages:
- GPyOpt==1.2.5
- scikit-learn==0.21.2
The following algorithms are currently supported by skopt:
- LinearSVC
- RandomForestClassifier
- LinearSVR
- RandomForestRegressor
Below is a simple example of using skopt to find the optimal parameters for a random forest classifiers with features X and labels y:
from skopt import model_searcher
searcher = model_searcher.ModelSearcher('RandomForestClassifier')
searcher.fit(X, y, 'accuracy')