Scikit-psl

Fixed

- MaxEntropyBinarizer also uses stricter test to detect binary features

Fixed

- PSL uses stricter test to detect binary features

Added

- PSL classifier
- probability predictions can return confidence intervals
- probability calibration using BetaCalibration
- stages can now be sliced and iterated (\__getitem\__(), \__iter\__())
- Metrics
- Weighted loss metric
- Rank loss metric

Fixed

- PSL more robust against non-standard class labels like "True"/"False" instead of boolean values

Added

- PSL supports Dataframes as inputs

Added

- PSL supports instance weights

Added

- Extended precision at recall function