Github user srowen commented on a diff in the pull request:
https://github.com/apache/spark/pull/19438#discussion_r143684083
--- Diff:
sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/util/QuantileSummariesSuite.scala
---
@@ -58,7 +58,7 @@ class QuantileSummariesSuite extends SparkFunSuite {
if (data.nonEmpty) {
val approx = summary.query(quant).get
// The rank of the approximation.
- val rank = data.count(_ < approx) // has to be <, not <= to be exact
+ val rank = data.count(_ <= approx)
--- End diff --
@wzhfy that formula is asymmetric which feels wrong; it may happen to fix
this but maybe would fail another future case. It would be a little more
principled to round the average.
Yeah I know that [1,2,2,2,2,2,2,2,3] can't happen in this test, just
illustrating a general point.
Hm, what's the case where the quantile is between 39 and 40? the input is
0-99 in that case? I don't see a test for the 40% quantile so wondering if we
really do have a problem or are misunderstanding the failure.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
