[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14966008#comment-14966008 ] holdenk commented on SPARK-11171: - Is there an different API in mind for this, or would matching the MLLib toPMML be a good starting point? > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14966077#comment-14966077 ] Joseph K. Bradley commented on SPARK-11171: --- I think matching should be fine, but it'd be worth thinking about issues with the current one. I haven't heard complaints though. > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14967558#comment-14967558 ] holdenk commented on SPARK-11171: - Skimming the mailing list it seems that people mostly just want more models to support PMML export and I don't see many issues with the API its self. I'll get started on doing this (break it up into different tickets this afternoon and do some code over the next few days). > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14967608#comment-14967608 ] Joseph K. Bradley commented on SPARK-11171: --- Sounds great, thanks! > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14967956#comment-14967956 ] holdenk commented on SPARK-11171: - So from the implementation point of view, we could keep extending the factory - but I was thinking since we are looking at adding more save methods directly in the pipeline models anyways (from the parent) maybe not going through the factory might be more reasonable. I'll try and do this for KMeans and make a PR for it :) > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14968214#comment-14968214 ] Apache Spark commented on SPARK-11171: -- User 'holdenk' has created a pull request for this issue: https://github.com/apache/spark/pull/9207 > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.3.4#6332) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org
[jira] [Commented] (SPARK-11171) PMML for Pipelines API
[ https://issues.apache.org/jira/browse/SPARK-11171?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16276640#comment-16276640 ] Apache Spark commented on SPARK-11171: -- User 'holdenk' has created a pull request for this issue: https://github.com/apache/spark/pull/19876 > PMML for Pipelines API > -- > > Key: SPARK-11171 > URL: https://issues.apache.org/jira/browse/SPARK-11171 > Project: Spark > Issue Type: New Feature > Components: ML >Reporter: Joseph K. Bradley >Assignee: holdenk > > We need to add PMML export to the spark.ml Pipelines API. > We should make 1 subtask JIRA per model. Hopefully we can reuse the > underlying implementation, adding simple wrappers. -- This message was sent by Atlassian JIRA (v6.4.14#64029) - To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org For additional commands, e-mail: issues-h...@spark.apache.org