Version 1.3.4

• Added support for pandas >= 1

  • Previously didn’t support latest pandas.

• Add sklearn OneHotEncoder

  • Previously used category_encoders, use scikit-learn’s instead for better and more reliable performance.

  • This object can be accessed as a transformer under ‘one hot encoder’.

• Added initial support for in-place FIs

  • Moving from plotting via ML to plotting from the Feature Importance object itself

  • Only fully supports global right now.

• Allow transformers to be skipped if out of scope

  • Previously would cause error.