Github user srowen commented on a diff in the pull request: https://github.com/apache/spark/pull/13624#discussion_r68203172 --- Diff: mllib/src/main/scala/org/apache/spark/ml/tree/impl/GradientBoostedTrees.scala --- @@ -205,31 +205,31 @@ private[spark] object GradientBoostedTrees extends Logging { case _ => data } - val numIterations = trees.length - val evaluationArray = Array.fill(numIterations)(0.0) - val localTreeWeights = treeWeights - - var predictionAndError = computeInitialPredictionAndError( - remappedData, localTreeWeights(0), trees(0), loss) - - evaluationArray(0) = predictionAndError.values.mean() - val broadcastTrees = sc.broadcast(trees) - (1 until numIterations).foreach { nTree => - predictionAndError = remappedData.zip(predictionAndError).mapPartitions { iter => - val currentTree = broadcastTrees.value(nTree) - val currentTreeWeight = localTreeWeights(nTree) - iter.map { case (point, (pred, error)) => - val newPred = updatePrediction(point.features, pred, currentTree, currentTreeWeight) - val newError = loss.computeError(newPred, point.label) - (newPred, newError) - } + val localTreeWeights = treeWeights + val treesIndices = trees.indices + + val dataCount = remappedData.count() + val evaluation = remappedData + .map { (point: LabeledPoint) => + treesIndices + .map(idx => { + val prediction = broadcastTrees.value(idx) + .rootNode + .predictImpl(point.features) + .prediction + prediction * localTreeWeights(idx) + } + ) + .scanLeft(0.0)(_ + _).drop(1) + .map(prediction => loss.computeError(prediction, point.label)) } - evaluationArray(nTree) = predictionAndError.values.mean() - } - - broadcastTrees.unpersist() - evaluationArray + .aggregate(treesIndices.map(_ => 0.0))( + (aggregated, row) => treesIndices.map(idx => aggregated(idx) + row(idx)), + (a, b) => treesIndices.map(idx => a(idx) + b(idx))) + .map(d => d / dataCount) --- End diff -- `.map(_ / count)`? trivial but hey
--- 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