Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[jvm-packages] Automatically set maximize_evaluation_metrics if not explicitly given in XGBoost4J-Spark (#4446)

Browse Source 
* Automatically set maximize_evaluation_metrics if not explicitly given.

* When custom_eval is set, require maximize_evaluation_metrics.

* Update documents on early stop in XGBoost4J-Spark.

* Fix code error.
This commit is contained in:
Shaochen Shi
2019-05-09 14:49:44 -05:00
committed by Nan Zhu
parent 8da4907e89
commit 18e4fc3690
3 changed files with 20 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
doc/jvm/xgboost4j_spark_tutorial.rst
View File

@@ -239,7 +239,7 @@ Early Stopping
Early stopping is a feature to prevent the unnecessary training iterations. By specifying ``num_early_stopping_rounds`` or directly call ``setNumEarlyStoppingRounds`` over a XGBoostClassifier or XGBoostRegressor, we can define number of rounds if the evaluation metric going away from the best iteration and early stop training iterations.
In additional to ``num_early_stopping_rounds``, you also need to define ``maximize_evaluation_metrics`` or call ``setMaximizeEvaluationMetrics`` to specify whether you want to maximize or minimize the metrics in training.
When it comes to custom eval metrics, in additional to ``num_early_stopping_rounds``, you also need to define ``maximize_evaluation_metrics`` or call ``setMaximizeEvaluationMetrics`` to specify whether you want to maximize or minimize the metrics in training. For built-in eval metrics, XGBoost4J-Spark will automatically select the direction.
For example, we need to maximize the evaluation metrics (set ``maximize_evaluation_metrics`` with true), and set ``num_early_stopping_rounds`` with 5. The evaluation metric of 10th iteration is the maximum one until now. In the following iterations, if there is no evaluation metric greater than the 10th iteration's (best one), the traning would be early stopped at 15th iteration.