[GitHub] [spark] zero323 commented on issue #27241: [SPARK-30533][ML][PYSPARK] Add classes to represent Java Regressors and RegressionModels
zero323 commented on issue #27241: [SPARK-30533][ML][PYSPARK] Add classes to represent Java Regressors and RegressionModels URL: https://github.com/apache/spark/pull/27241#issuecomment-575855169 Thanks @huaxingao @srowen @zhengruifeng This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] zero323 commented on issue #27241: [SPARK-30533][ML][PYSPARK] Add classes to represent Java Regressors and RegressionModels
zero323 commented on issue #27241: [SPARK-30533][ML][PYSPARK] Add classes to represent Java Regressors and RegressionModels URL: https://github.com/apache/spark/pull/27241#issuecomment-575582976 > The Regressor class is basically empty and I am not sure if we should add another layer of abstraction, so I chose not to add Regressor/Regressor class on python side when I did #27168. But I know you can argue that we need change python too to keep the parity between scala and python. I am OK either way. I am actually not that interested in parity (as it is right now it provides little or no value to the end user, inflates `pyspark.ml` codebase, and actually increases effort required to maintain the whole thing) as much as practical value. As I argued in discussion around https://github.com/apache/spark/pull/25776#issuecomment-533488999 ability to distinguish between types of predictors is fundamental for building complex ML workflows, and current API is not sufficient to that (`Classifiers` and non classifier `Predictors` usually have the same API, and produce identical output schema). > the Regressor class is basically empty I am afraid that's, for good or bad, argument you can make against significant chunk of the API. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] zero323 commented on issue #27241: [SPARK-30533][ML][PYSPARK] Add classes to represent Java Regressors and RegressionModels
zero323 commented on issue #27241: [SPARK-30533][ML][PYSPARK] Add classes to represent Java Regressors and RegressionModels URL: https://github.com/apache/spark/pull/27241#issuecomment-575426906 > Is this more for symmetry than anything else? I don't have a strong opinion. I'd defer to @huaxingao and @zhengruifeng who have been watching this much more closely. To some extent. In general it useful to be able to distinguish between `Regressors`, `Classiffiers` and other types of `Params` (more about [here](https://issues.apache.org/jira/browse/SPARK-29212)). Additionally we cannot really get Scala parity without these. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
