[
https://issues.apache.org/jira/browse/SPARK-57865?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mao Li updated SPARK-57865:
---------------------------
Description:
Apply the HasIntermediateStorageLevel shared param (introduced in SPARK-57860)
to the
remaining estimators that hardcode StorageLevel.MEMORY_AND_DISK for
intermediate datasets:
Classification: LogisticRegression, LinearSVC, FMClassifier, OneVsRest
Clustering: GaussianMixture, LDA
Regression: LinearRegression, AFTSurvivalRegression, FMRegressor,
IsotonicRegression
FPM: FPGrowth
Includes the corresponding PySpark changes and tests. Part of SPARK-47103.
was:
Apply HasIntermediateStorageLevel (introduced in the first sub-task) to tree
ensemble estimators. These require passing the storage level through to
internal implementation classes (GradientBoostedTrees.scala,
RandomForest.scala).
Affected estimators:
- GBTClassifier
- GBTRegressor
- RandomForestClassifier
- RandomForestRegressor
Also update the corresponding PySpark classes.
Part of SPARK-47103.
> Apply HasIntermediateStorageLevel to remaining
> classification/clustering/regression estimators
> ----------------------------------------------------------------------------------------------
>
> Key: SPARK-57865
> URL: https://issues.apache.org/jira/browse/SPARK-57865
> Project: Spark
> Issue Type: Sub-task
> Components: MLlib
> Affects Versions: 4.0.0
> Reporter: Mao Li
> Priority: Major
>
> Apply the HasIntermediateStorageLevel shared param (introduced in
> SPARK-57860) to the
> remaining estimators that hardcode StorageLevel.MEMORY_AND_DISK for
> intermediate datasets:
> Classification: LogisticRegression, LinearSVC, FMClassifier, OneVsRest
> Clustering: GaussianMixture, LDA
> Regression: LinearRegression, AFTSurvivalRegression, FMRegressor,
> IsotonicRegression
> FPM: FPGrowth
> Includes the corresponding PySpark changes and tests. Part of SPARK-47103.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
