You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Kuang Chen (JIRA)" <ji...@apache.org> on 2018/01/16 19:08:00 UTC
[jira] [Created] (SPARK-23091) Incorrect unit test for
approxQuantile
Kuang Chen created SPARK-23091:
----------------------------------
Summary: Incorrect unit test for approxQuantile
Key: SPARK-23091
URL: https://issues.apache.org/jira/browse/SPARK-23091
Project: Spark
Issue Type: Improvement
Components: ML
Affects Versions: 2.2.1
Reporter: Kuang Chen
Currently, test for `approxQuantile` (quantile estimation algorithm) checks whether estimated quantile is within +- 2*`relativeError` from the true quantile. See the code below:
[https://github.com/apache/spark/blob/master/sql/core/src/test/scala/org/apache/spark/sql/DataFrameStatSuite.scala#L157]
However, based on the original paper by Greenwald and Khanna, the estimated quantile is guaranteed to be within +- `relativeError` from the true quantile. Using the double "tolerance" is misleading and incorrect, and we should fix it.
--
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