zhengruifeng commented on issue #26972: [SPARK-30321][ML] Log weightSum in Algo
that has weights support
URL: https://github.com/apache/spark/pull/26972#issuecomment-569875692
Merged to master, thanks all!
Happy new year!
zhengruifeng commented on issue #26972: [SPARK-30321][ML] Log weightSum in Algo
that has weights support
URL: https://github.com/apache/spark/pull/26972#issuecomment-569864240
retest this please
This is an automated message
zhengruifeng commented on issue #26972: [SPARK-30321][Ml] Log weightSum in Algo
that has weights support
URL: https://github.com/apache/spark/pull/26972#issuecomment-568991530
@huaxingao Can we pass the instr of GBTClassifier/GBTRegressor into the
internal impl, and use it to log the
zhengruifeng commented on issue #26972: [SPARK-30321][Ml] Log weightSum in Algo
that has weights support
URL: https://github.com/apache/spark/pull/26972#issuecomment-568828877
Yes, current GBT impls will log these metrics for each iteration. But I
guess we can just log weightSum there, if
zhengruifeng commented on issue #26972: [SPARK-30321][Ml] Log weightSum in Algo
that has weights support
URL: https://github.com/apache/spark/pull/26972#issuecomment-568610797
I notice that other statistics for tree models are logged in
`RandomForest.run`:
```scala
instr match {