[
https://issues.apache.org/jira/browse/SPARK-13969?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15203148#comment-15203148
]
Nick Pentreath commented on SPARK-13969:
----------------------------------------
We could do this - though to be honest I think the name "HashingTF" is a bit
specific to term frequencies. I'd prefer to create a new {{FeatureHasher}} and
then either have {{HashingTF}} use it under the hood, or since we're moving to
Spark 2.0.0, deprecate {{HashingTF}} (this would also solve any issues in
persistence / versioning between different versions' behaviour for
{{HashingTF}} as mentioned in SPARK-10574). What do you think?
> Extend input format that feature hashing can handle
> ---------------------------------------------------
>
> Key: SPARK-13969
> URL: https://issues.apache.org/jira/browse/SPARK-13969
> Project: Spark
> Issue Type: Sub-task
> Components: ML, MLlib
> Reporter: Nick Pentreath
> Priority: Minor
>
> Currently {{HashingTF}} works like {{CountVectorizer}} (the equivalent in
> scikit-learn is {{HashingVectorizer}}). That is, it works on a sequence of
> strings and computes term frequencies.
> The use cases for feature hashing extend to arbitrary feature values (binary,
> count or real-valued). For example, scikit-learn's {{FeatureHasher}} can
> accept a sequence of (feature_name, value) pairs (e.g. a map, list). In this
> way, feature hashing can operate as both "one-hot encoder" and "vector
> assembler" at the same time.
> Investigate adding a more generic feature hasher (that in turn can be used by
> {{HashingTF}}).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
