[
https://issues.apache.org/jira/browse/SPARK-13600?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15176580#comment-15176580
]
Oliver Pierson commented on SPARK-13600:
----------------------------------------
Yeah, you can assign this to me. However, it may be a few days or even a week
before I can get a PR together. I was hoping to get people's opinion on
reimplementing findSplitCandidates using another
[method|https://en.wikipedia.org/wiki/Quantile#Estimating_quantiles_from_a_sample]
(see the second paragraph of that section). I believe it's done this way in
Numpy/Scipy. [~srowen][~mengxr][~yinxusen]
> Incorrect number of buckets in QuantileDiscretizer
> --------------------------------------------------
>
> Key: SPARK-13600
> URL: https://issues.apache.org/jira/browse/SPARK-13600
> Project: Spark
> Issue Type: Bug
> Components: MLlib
> Affects Versions: 1.6.0, 2.0.0
> Reporter: Oliver Pierson
>
> Under certain circumstances, QuantileDiscretizer fails to calculate the
> correct splits resulting in an incorrect number of buckets/bins.
> E.g.
> val df = sc.parallelize(1.0 to 10.0 by 1.0).map(Tuple1.apply).toDF("x")
> val discretizer = new
> QuantileDiscretizer().setInputCol("x").setOutputCol("y").setNumBuckets(5)
> discretizer.fit(df).getSplits
> gives:
> Array(-Infinity, 2.0, 4.0, 6.0, 8.0, 10.0, Infinity)
> which corresponds to 6 buckets (not 5).
> The problem appears to be in the QuantileDiscretizer.findSplitsCandidates
> method.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
