You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This interface is unsuitable to being a public class as part of the IEstimator/ITransformer idioms, being more of implementation interface for the infrastructure of the calibrators. So we don't want that exposed, but we do nonetheless want the ability to given some data, produce a calibrator out of it.
Happily we almost have it, in the form of the subclasses of the calibrator estimators.
The only real trouble is that this object assumes, at a fairly deep level, that internally there is an IPredictor in it, but as far as I know this is mostly due to some internal convenience code that I hope could be refactored to just take an IDataView directly.
One of the legacy interfaces we have is the calibrator trainer.
machinelearning/src/Microsoft.ML.Data/Prediction/Calibrator.cs
Line 95 in f269adc
This interface is unsuitable to being a public class as part of the
IEstimator
/ITransformer
idioms, being more of implementation interface for the infrastructure of the calibrators. So we don't want that exposed, but we do nonetheless want the ability to given some data, produce a calibrator out of it.Happily we almost have it, in the form of the subclasses of the calibrator estimators.
machinelearning/src/Microsoft.ML.Data/Prediction/CalibratorCatalog.cs
Line 52 in f269adc
The only real trouble is that this object assumes, at a fairly deep level, that internally there is an
IPredictor
in it, but as far as I know this is mostly due to some internal convenience code that I hope could be refactored to just take anIDataView
directly.This might be viewed as a sub-part of #1871.
/ccing @sfilipi for visibility.
The text was updated successfully, but these errors were encountered: