[
https://issues.apache.org/jira/browse/SPARK-7921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14571378#comment-14571378
]
Joseph K. Bradley commented on SPARK-7921:
------------------------------------------
Note: I just realized that R glm's default behavior is to drop the first
element. (It sorts named categories alphabetically and then drops the first.)
But I don't think we want to emulate that since sorting by frequency makes more
sense.
> Change includeFirst to dropLast in OneHotEncoder
> ------------------------------------------------
>
> Key: SPARK-7921
> URL: https://issues.apache.org/jira/browse/SPARK-7921
> Project: Spark
> Issue Type: Improvement
> Components: ML
> Affects Versions: 1.4.0
> Reporter: Xiangrui Meng
> Assignee: Xiangrui Meng
> Fix For: 1.4.0
>
>
> Change includeFirst to dropLast and leave the default to true. There are
> couple benefits:
> a. consistent with other tutorials of one-hot encoding (or dummy coding)
> (e.g., http://www.ats.ucla.edu/stat/mult_pkg/faq/general/dummy.htm)
> b. keep the indices unmodified in the output vector. If we drop the first,
> all indices will be shifted by 1.
> c. If users use StringIndex, the last element is the least frequent one.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
