Github user XXXShao commented on a diff in the pull request:
https://github.com/apache/spark/pull/16722#discussion_r137897201
--- Diff:
mllib/src/main/scala/org/apache/spark/ml/tree/impl/RandomForest.scala ---
@@ -1002,9 +1018,9 @@ private[spark] object RandomForest extends Logging {
val numSplits = metadata.numSplits(featureIndex)
// get count for each distinct value
- val (valueCountMap, numSamples) =
featureSamples.foldLeft((Map.empty[Double, Int], 0)) {
- case ((m, cnt), x) =>
- (m + ((x, m.getOrElse(x, 0) + 1)), cnt + 1)
+ val (valueCountMap, numSamples) =
featureSamples.foldLeft((Map.empty[Double, Double], 0.0)) {
--- End diff --
Hi, thanks for your contribution~ I have a question about considering
weight info in findSplitsForContinuousFeature here. It looks the continuous
features will be influenced much more by instance weight because the weight
part is considered twice: (1)make split (2) calculate impurity. Normally weight
is only mentioned in impurity calculation part according to limited papers I
have read. Could you provide some reference you refer here? And correct me if
I misunderstand your code. :) Thanks!
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
