Joseph K. Bradley created SPARK-6025:
----------------------------------------
Summary: Helper method for GradientBoostedTrees to compute
validation error
Key: SPARK-6025
URL: https://issues.apache.org/jira/browse/SPARK-6025
Project: Spark
Issue Type: New Feature
Components: MLlib
Affects Versions: 1.3.0
Reporter: Joseph K. Bradley
Priority: Minor
Create a helper method for computing the error at each iteration of boosting.
This should be used post-hoc to compute the error efficiently on a new dataset.
E.g.:
{code}
def evaluateEachIteration(data: RDD[LabeledPoint], evaluator): Array[Double]
{code}
Notes:
* It should run in the same big-O time as predict() by keeping a running total
(residual).
* A different method name could be good.
* It could take an evaluator and/or could evaluate using the training metric by
default.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
