[
https://issues.apache.org/jira/browse/FLINK-1728?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16780468#comment-16780468
]
Till Rohrmann commented on FLINK-1728:
--------------------------------------
Hi [~sclee01], I would discourage you from working on this issue since the
community is currently discussing to start a new machine learning library. This
would entail that {{flink-ml}} would be ditched.
> Add random forest ensemble method to machine learning library
> -------------------------------------------------------------
>
> Key: FLINK-1728
> URL: https://issues.apache.org/jira/browse/FLINK-1728
> Project: Flink
> Issue Type: New Feature
> Components: Library / Machine Learning
> Reporter: Till Rohrmann
> Priority: Major
> Labels: ML
>
> Random forests [2,3] are a well-established mean to mitigate the decision
> trees' weakness of overfitting. Therefore this would be a valuable
> contribution to Flink's machine learning library.
> Google [1] describes some of the techniques they used to do ensemble learning
> of MapReduce. This could be helpful while implementing a distributed random
> forest.
> Resources:
> [1]
> [http://static.googleusercontent.com/media/research.google.com/en/us/pubs/archive/36296.pdf]
> [2] [http://www.stat.berkeley.edu/~breiman/randomforest2001.pdf]
> [3] [http://www.stat.berkeley.edu/~breiman/Using_random_forests_V3.1.pdf]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
