[jira] [Created] (SPARK-26721) Bug in feature importance calculation in GBM (and possibly other decision tree classifiers)

["Daniel Jumper (JIRA)" <ji...@apache.org>]

Daniel Jumper created SPARK-26721:
-------------------------------------

             Summary: Bug in feature importance calculation in GBM (and possibly other decision tree classifiers)
                 Key: SPARK-26721
                 URL: https://issues.apache.org/jira/browse/SPARK-26721
             Project: Spark
          Issue Type: Bug
          Components: ML
    Affects Versions: 2.4.0
            Reporter: Daniel Jumper


The feature importance calculation in org.apache.spark.ml.classification.GBTClassificationModel.featureImportances follows a flawed implementation from scikit-learn. An error was recently discovered and updated in scikit-learn version 0.20.0. This error is inherited in the spark implementation and needs to be fixed here as well.

As described in the [scikit-learn release notes|[https://scikit-learn.org/stable/whats_new.html#version-0-20-0]|https://scikit-learn.org/stable/whats_new.html#version-0-20-0]:]  :
{quote}
Fix Fixed a bug in ensemble.GradientBoostingRegressor and ensemble.GradientBoostingClassifier to have feature importances summed and then normalized, rather than normalizing on a per-tree basis. The previous behavior over-weighted the Gini importance of features that appear in later stages. This issue only affected feature importances. #11176 by Gil Forsyth.
{quote}
Full discussion of this error and debate ultimately validating the correctness of the change can be found in the comment thread of the scikit-learn pull request: [https://github.com/scikit-learn/scikit-learn/pull/11176] 

 

I believe the main change required would be to the featureImportances function in mllib/src/main/scala/org/apache/spark/ml/tree/treeModels.scala , however, I do not have the experience to make this change myself.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org