Github user srowen commented on a diff in the pull request:
https://github.com/apache/spark/pull/2928#discussion_r19364103
--- Diff: core/src/main/scala/org/apache/spark/partial/StudentTCacher.scala
---
@@ -35,7 +37,8 @@ private[spark] class StudentTCacher(confidence: Double) {
} else {
val size = sampleSize.toInt
if (cache(size) < 0) {
- cache(size) = Probability.studentTInverse(1 - confidence, size - 1)
+ val tDist = new TDistribution(size - 1)
+ cache(size) = tDist.inverseCumulativeProbability(1 - (1 -
confidence) / 2)
--- End diff --
It should still be one-sided. The argument to the Colt method
(http://acs.lbl.gov/ACSSoftware/colt/api/cern/jet/stat/Probability.html#studentTInverse(double,
int)) was `alpha` such that the cumulative probability is `1-alpha/2`. For the
Commons Math3 method, the argument is just the cumulative probability
(http://commons.apache.org/proper/commons-math/apidocs/org/apache/commons/math3/distribution/AbstractRealDistribution.html#inverseCumulativeProbability(double)).
So I translated as `1 - (1 - confidence) / 2`.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
