Github user yanboliang commented on a diff in the pull request:
https://github.com/apache/spark/pull/7694#discussion_r35959887
--- Diff:
mllib/src/main/scala/org/apache/spark/ml/tree/impl/RandomForest.scala ---
@@ -734,12 +703,10 @@ private[ml] object RandomForest extends Logging {
val rightChildStats =
binAggregates.getImpurityCalculator(nodeFeatureOffset,
numSplits)
rightChildStats.subtract(leftChildStats)
- predictionAndImpurity = Some(predictionAndImpurity.getOrElse(
- calculatePredictImpurity(leftChildStats, rightChildStats)))
- val gainStats = calculateGainForSplit(leftChildStats,
- rightChildStats, binAggregates.metadata,
predictionAndImpurity.get._2)
- (splitIdx, gainStats)
- }.maxBy(_._2.gain)
+ gainAndImpurityStats = Some(calculateImpurityStats(
--- End diff --
Here we don't need to use ```getOrElse``` because the
```gainAndImpurityStats``` need to be recompute for each split candidate, the
only changeless members are ```impurity``` and ```impurityCalculator```. I have
make optimization in the function calculateImpurityStats to directly get these
two variables rather than recompute them.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
