Piml

1. Support data loading by Spark
2. Support H2O model registration
3. Support data-dependent model interpretation and explanation
4. Add Friedman’s H-statistic to post hoc explanation
5. Add data integrity check and more outlier detection methods
6. Add log loss and Brier score to binary classification metrics
7. Add segmented performance diagnostics
8. Add hyperparameter tuning by grid and randomized search
9. Pipeline optimization, code cleanup, and bug fixes

1. Add model-free diagnostics test APIs.
2. Add support for a wider range of computing environments, including Mac ARM.
3. Enhance overfit diagnostic module to support AUC metric.
4. Add sliced 1D plot as alternatives of 2D heatmap.
5. Code refactorization and bug fixes.

1. Release of PiML user guide.
2. Add data quality module.
3. Add interpretable XGB1 model.
4. Add train-test split methods in data_prepare.
5. Improve binning search in model_fairness.
6. Some bug fixes and miscellaneous changes.