If you are talking about distributed training of a single XGBoost model, there is no built-in capability in SparkML. SparkML supports gradient boosted trees, but not XGBoost specifically. However, there are 3rd party packages, such as XGBoost4J that ...
