[jira] [Commented] (MADLIB-1299) Baseline referencing for independent variables

["Rahul Iyer (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MADLIB-1299?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16762162#comment-16762162 ] 

Rahul Iyer commented on MADLIB-1299:
------------------------------------

@ [~pratikpawar]

By baseline reference, do you mean the reference for dummy encoding of a categorical independent variable?

If yes, then MADlib expects that the data has been encoded before passing it to the regression methods. See [Encoding Categorical Variables|http://madlib.apache.org/docs/latest/group__grp__encode__categorical.html] for the procedure to dummy encode. The {{value_to_drop}} parameter provides the functionality for user-defined reference level.

> Baseline referencing for independent variables
> ----------------------------------------------
>
>                 Key: MADLIB-1299
>                 URL: https://issues.apache.org/jira/browse/MADLIB-1299
>             Project: Apache MADlib
>          Issue Type: Improvement
>          Components: Module: Linear Regression, Module: Logistic Regression, Module: Multinomial Logistic Regression
>            Reporter: Pratik
>            Assignee: Rahul Iyer
>            Priority: Minor
>
> When we perform regression between dep vs independent variables currently function does not support control over baseline reference for the independent variable. 
> I am not sure if Madlib does it internally with 1.15.1 release. If that is the case it would be awesome to have documentation and get the return value of the baseline variable.
> [https://stackoverflow.com/questions/54523454/linear-regression-baseline-referencing-madlib]
>  
> R way to do it - 
> https://stackoverflow.com/questions/3872070/how-to-force-r-to-use-a-specified-factor-level-as-reference-in-a-regression?noredirect=1&lq=1 
>  
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)