Github user jkbradley commented on a diff in the pull request: https://github.com/apache/spark/pull/3374#discussion_r20629011 --- Diff: mllib/src/main/scala/org/apache/spark/mllib/tree/GradientBoostedTrees.scala --- @@ -40,151 +39,98 @@ import org.apache.spark.storage.StorageLevel * Notes: * - This currently can be run with several loss functions. However, only SquaredError is * fully supported. Specifically, the loss function should be used to compute the gradient - * (to re-label training instances on each iteration) and to weight weak hypotheses. + * (to re-label training instances on each iteration) and to weight tree ensembles. * Currently, gradients are computed correctly for the available loss functions, - * but weak hypothesis weights are not computed correctly for LogLoss or AbsoluteError. - * Running with those losses will likely behave reasonably, but lacks the same guarantees. + * but tree predictions are not computed correctly for LogLoss or AbsoluteError since they --- End diff -- (copying comment here since it was on an outdated diff) The original explanation is correct for the original Gradient Boosting algorithm, which uses weak hypothesis weights and is oblivious to the weak learner being used. This updated explanation is really for TreeBoost, Friedman's improvement to the original algorithm which is specialized for trees (which we should add at some point but isn't what we're claiming to have now, I'd say). So I think the original explanation is more accurate since we do not claim to implement TreeBoost.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org