Journey Optimization Platform Release Notes
|Release Date||Release Type||Restrictions||AIX||Linux||Solaris||Windows|
This release includes the following new features and enhancements:
- This release improves feature ranking analysis in the Feature Analysis report, which identifies the variables having the greatest impact on outcomes for a specified metric. To achieve this improvement, the sklearn DecisionTreeRegressor/Classifier was replaced with XGBoost library (an optimized distributed gradient boosting library). In conjunction with this improvement, the Decision Tree view was removed from the Feature Analysis report.
This release contains the following resolved issues:
Model training no longer fails when the model contains high-cardinality numeric features. Previously, the method for estimating the required feature space size for numeric features overestimated the necessary size. As a result, it under-sampled the training dataset such that no records were available for training. (PRR-1509)
No special procedure is required to upgrade to release 9.0.006.13.