0.3.2

• Fix Cubist per its updated interface

0.3.1

• Add XGBoost
• Fixes for leave-one-out cross-validation
• Fix deprecation warnings for RandomForest

0.3.0

This version introduces the following breaking changes:

1. Calls on model.train and model.train_with_sequence now accept as input a Dataset object instead of the explicit X and Y arrays.
2. PLS, Cubist and RF now work with multi-output data; as a consequence the return value of the model.train function is now a list of dictionaries. In the case of one output value this list has a length of 1.

The following updates took place:

1. Azure Pipelines are now used for CI
2. pytest now used to run the tests
3. An initial explainability component is added to help create visualizations of e.g. feature importance as identified by the model
4. code formatted with black