Peter Steiner

I would like to use ```sklearn.model_selection.RandomizedSearchCV``` to optimize several HDBSCAN parameters. I noticed that HDBSCAN only allows the dtype ```float``` but not ```np.float64 (from scipy.stats.loguniform)```. The following code (mwe.py) can...

Is it possible to run Ridge() without memmap?

2

Hello, many thanks for providing your great library. I tried to develop an adapter class for ReservoirPy so that it fits the API of scikit-learn. It seems as if this...

This would allow to export a trained model http://onnx.ai/sklearn-onnx/

Since probabilities cannot be larger than 1.0, this would make the functions predict_proba(X) more meaningful. Currently, I do not see additional benefits.