[ 
https://issues.apache.org/jira/browse/SPARK-30532?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17048839#comment-17048839
 ] 

Oleksii Kachaiev commented on SPARK-30532:
------------------------------------------

The same applies to other stats functions: {{cov}}, {{corr}} and {{freqItems}}.

I'm working on PR to fix all of them (the problem is similar in all cases).

> DataFrameStatFunctions.approxQuantile doesn't work with TABLE.COLUMN syntax
> ---------------------------------------------------------------------------
>
>                 Key: SPARK-30532
>                 URL: https://issues.apache.org/jira/browse/SPARK-30532
>             Project: Spark
>          Issue Type: Bug
>          Components: Spark Core
>    Affects Versions: 2.4.4
>            Reporter: Chris Suchanek
>            Priority: Minor
>
> The DataFrameStatFunctions.approxQuantile doesn't work with fully qualified 
> column name (i.e TABLE_NAME.COLUMN_NAME) which is often the way you refer to 
> the column when working with joined dataframes having ambiguous column names.
> See code below for example.
> {code:java}
> import scala.util.Random
> val l = (0 to 1000).map(_ => Random.nextGaussian() * 1000)
> val df1 = sc.parallelize(l).toDF("num").as("tt1")
> val df2 = sc.parallelize(l).toDF("num").as("tt2")
> val dfx = df2.crossJoin(df1)
> dfx.stat.approxQuantile("tt1.num", Array(0.1), 0.0)
> // throws: java.lang.IllegalArgumentException: Field "tt1.num" does not exist.
> Available fields: num
> dfx.stat.approxQuantile("num", Array(0.1), 0.0)
> // throws: org.apache.spark.sql.AnalysisException: Reference 'num' is 
> ambiguous, could be: tt2.num, tt1.num.;{code}
>  
>  



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org

Reply via email to