You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@madlib.apache.org by "Frank McQuillan (JIRA)" <ji...@apache.org> on 2016/09/23 20:59:20 UTC

[jira] [Comment Edited] (MADLIB-951) Elastic net - support column names instead of an array for features

    [ https://issues.apache.org/jira/browse/MADLIB-951?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15090329#comment-15090329 ] 

Frank McQuillan edited comment on MADLIB-951 at 9/23/16 8:58 PM:
-----------------------------------------------------------------

We can't give the column names since MADlib internally never sees the names. We use the array expression as is without parsing it, so the elastic net function only sees an array of numbers. 
This was done to allow the user to send any expression leading to an array - for instance if you have a function that returns an array of feature values then you can give that instead of the long array expression. 

Looks like this flexibility isn't normally used (?) and we would be better-off just asking for a comma-separated list of column names that we can parse through. We built that like decision trees. 


was (Author: fmcquillan):
We can't give the column names since MADlib internally never sees the names. We use the array expression as is without parsing it, so the elastic net function only sees an array of numbers. 
This was done to allow the user to send any expression leading to an array - for instance if you have a function that returns an array of feature values then you can give that instead of the long array expression. 

Looks like this flexibility isn't normally used and we would be better-off just asking for a comma-separated list of column names that we can parse through. We built that like decision trees. 

> Elastic net - support column names instead of an array for features
> -------------------------------------------------------------------
>
>                 Key: MADLIB-951
>                 URL: https://issues.apache.org/jira/browse/MADLIB-951
>             Project: Apache MADlib
>          Issue Type: Improvement
>          Components: Module: Regularized Regression
>            Reporter: Frank McQuillan
>            Priority: Minor
>             Fix For: v2.0
>
>         Attachments: screenshot-1.png
>
>
> From a user:
> One usability issue: when you are displaying the results how come you give the index of the features ? why not the feature name? 



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