[jira] [Commented] (SPARK-10291) Add statsByKey method to compute StatCounters for each key in an RDD
[ https://issues.apache.org/jira/browse/SPARK-10291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15064266#comment-15064266 ] Sean Owen commented on SPARK-10291: --- My POV is that this isn't likely worth adding a method for. I appreciate the value of utility methods but have to weight it against adding another item to a core API and how often it'd be used. This is also straightforward to express in Spark SQL on a dataframe, no? > Add statsByKey method to compute StatCounters for each key in an RDD > > > Key: SPARK-10291 > URL: https://issues.apache.org/jira/browse/SPARK-10291 > Project: Spark > Issue Type: New Feature > Components: PySpark >Reporter: Erik Shilts >Priority: Minor > > A common task is to summarize numerical data for different groups. Having a > `statsByKey` method would simplify this so the user would not have to write > the aggregators for all the statistics or manage collecting by key and > computing individual StatCounters. > This should be a straightforward addition to PySpark. I can look into adding > to Scala and R if we want to maintain feature parity. -- 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-10291) Add statsByKey method to compute StatCounters for each key in an RDD
[ https://issues.apache.org/jira/browse/SPARK-10291?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14723944#comment-14723944 ] Apache Spark commented on SPARK-10291: -- User 'eshilts' has created a pull request for this issue: https://github.com/apache/spark/pull/8539 > Add statsByKey method to compute StatCounters for each key in an RDD > > > Key: SPARK-10291 > URL: https://issues.apache.org/jira/browse/SPARK-10291 > Project: Spark > Issue Type: New Feature > Components: PySpark >Reporter: Erik Shilts >Priority: Minor > > A common task is to summarize numerical data for different groups. Having a > `statsByKey` method would simplify this so the user would not have to write > the aggregators for all the statistics or manage collecting by key and > computing individual StatCounters. > This should be a straightforward addition to PySpark. I can look into adding > to Scala and R if we want to maintain feature parity. -- 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