v0.4.8

What's Changed

• Fix from xgboost by @jinlow in #107

Full Changelog: v0.4.7...v0.4.8

v0.4.7

What's Changed

• Feature/leaf preds by @jinlow in #104

Full Changelog: v0.4.6...v0.4.7

v0.4.6

What's Changed

• Feature/validate y weight by @jinlow in #101
• Update pyproject.toml by @jinlow in #102

Full Changelog: v0.4.5...v0.4.6

v0.4.5

Release fixes issue where model trained in rust couldn’t be loaded in python.

v0.4.4

This release adds several new hyperparameters.

• colsample_bytree - randomly sample columns from the data at each iteration.
• l1 - apply l1 regularization to the predicted weights.
• max_delta_step - specify maximum change in prediction at each leaf.

v0.4.3

Add the “Shapley” option as a contribution method, which if specified will use the tree shap algorithm for calculating contributions.

v0.4.2

Fixed bug where missing might not be considered missing if greater than the minimum.

v0.4.1

Add number_of_trees attribute, and also clean up tree loop a little bit.

v0.4.0

Change evaluation dataset order, so that the last value is used for early stopping to align with XGBoost.

v0.3.4

New release introduces trees_to_dataframe method.