[jira] [Comment Edited] (SYSTEMML-700) Inflexible category labels for Multinomial Logistic Regression

["Matthias Boehm (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SYSTEMML-700?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15518574#comment-15518574 ] 

Matthias Boehm edited comment on SYSTEMML-700 at 9/24/16 7:07 AM:
------------------------------------------------------------------

that's a good point - could you clarify what exactly you have in mind (e.g., scale values, non-contiguous integers, or even frames with strings). 

Just to bring everybody on the same page, here is what we currently do for adjusting y (after preparation we require integers starting from 1)
{code}
if (min (Y_vec) <= 0) { 
    # Category labels "0", "-1" etc. are converted into the largest label
    max_y = max (Y_vec);
    Y_vec  = Y_vec  + (- Y_vec  + max_y + 1) * (Y_vec <= 0);
}
{code}

One could imaging something like {{transformencode(as.frame(y))}} to recode anything into the required range. 


was (Author: mboehm7):
that's a good point - could you clarify what exactly you have in mind (e.g., scale values, non-contiguous integers, or even frames with strings). 

Just to bring everybody on the same page, here is what we currently do for adjusting y (after preparation we require integers starting from 1)
{code}
if (min (Y_vec) <= 0) { 
    # Category labels "0", "-1" etc. are converted into the largest label
    max_y = max (Y_vec);
    Y_vec  = Y_vec  + (- Y_vec  + max_y + 1) * (Y_vec <= 0);
}
{code}

One could imaging something like transformencode(as.frame(y)) to recode anything into the required range. 

> Inflexible category labels for Multinomial Logistic Regression
> --------------------------------------------------------------
>
>                 Key: SYSTEMML-700
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-700
>             Project: SystemML
>          Issue Type: Bug
>          Components: Algorithms
>            Reporter: Jeremy
>            Priority: Minor
>   Original Estimate: 4h
>  Remaining Estimate: 4h
>
> The Logistic Regression algorithm requires that category labels be labeled as 0 up to the number of classes-1. It should be able to handle any set of category labels provided by the user. B_out should have the appropriate size regardless of the values of the labels given, and the algorithm should also preserve the original labeling for the user.



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