[
https://issues.apache.org/jira/browse/SPARK-61612?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hyukjin Kwon resolved SPARK-61612.
----------------------------------
Resolution: Fixed
Issue resolved by pull request 605
[https://github.com/apache/spark-connect-rust/pull/605]
> Support native FMClassifier, MultilayerPerceptronClassifier and OneVsRest
> -------------------------------------------------------------------------
>
> Key: SPARK-61612
> URL: https://issues.apache.org/jira/browse/SPARK-61612
> Project: Spark
> Issue Type: Improvement
> Components: Connect
> Affects Versions: connect-rust-4.3.0
> Reporter: Ruifeng Zheng
> Assignee: Ruifeng Zheng
> Priority: Major
> Labels: pull-request-available
> Fix For: connect-rust-4.3.0
>
>
> Add the three remaining classification estimators to the native Rust and
> Rust-backed Python APIs. FMClassifier and MultilayerPerceptronClassifier use
> Spark Connect cached models, with full constructor parameters, named setters,
> layer arrays and dense/sparse initial weights. OneVsRest performs parallel
> binary fitting and score aggregation in Rust with all eight native base
> classifiers, weights, cache cleanup and lazy model ownership. Validate wire
> encoding against official PySpark and numerical behavior with both clients.
> This covers estimator configuration, fitting and DataFrame prediction; model
> inspection and persistence remain separate API work.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]