[jira] [Closed] (SPARK-13291) Numerical models should preserve label attributes

["Joseph K. Bradley (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/SPARK-13291?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Joseph K. Bradley closed SPARK-13291.
-------------------------------------
    Resolution: Duplicate

> Numerical models should preserve label attributes
> -------------------------------------------------
>
>                 Key: SPARK-13291
>                 URL: https://issues.apache.org/jira/browse/SPARK-13291
>             Project: Spark
>          Issue Type: Improvement
>          Components: ML
>    Affects Versions: 1.6.0
>            Reporter: Piotr Smolinski
>            Priority: Minor
>
> I tried building a simple pipeline for Random Forest classification. The predictors are some doubles, some ints and some strings. The response is string. The RFormula seems to be a perfect candidate. RFormulaModel produces nicely *labelCol* column with StringIndexer derived metadata and RandomForestClassificationModel converts the *featuresCol* to *predictionCol*. The problem is that there is no way to convert the *predictionCol* (which is factor index) back to the label. The metadata created by StringIndexer is lost. 
> The numerical models should create the *predictionCol* columns with metadata seen on the *labelCol* column during the model fitting.
> Preserving metadata allows for example to pipeline RFormula, RandomForestClassifier and IndexToString.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

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