Skip to content

[SPARK-27539][SQL] Fix inaccurate aggregate outputRows estimation with column containing null values #24436

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
pengbo wants to merge 2 commits into from
Closed

[SPARK-27539][SQL] Fix inaccurate aggregate outputRows estimation with column containing null values #24436

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
db8701f
SPARK-27539: Inaccurate aggregate outputRows estimation with null val…
pengbo Apr 22, 2019
50ae5cf
style fixing
pengbo Apr 22, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions ...ala/org/apache/spark/sql/catalyst/plans/logical/statsEstimation/AggregateEstimation.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,8 +42,8 @@ object AggregateEstimation {
(res, expr) => {
val columnStat = childStats.attributeStats(expr.asInstanceOf[Attribute])
val distinctCount = columnStat.distinctCount.get
val distinctValue: BigInt = if (distinctCount == 0 && columnStat.nullCount.get > 0) {
1
val distinctValue: BigInt = if (columnStat.nullCount.get > 0) {
distinctCount + 1
Copy link
Member

@HyukjinKwon HyukjinKwon May 23, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hm, actually, do we need to count null as distinct value? It's not counted as a distinct value in SQL (F.countDistinct or count(DISTINCT col)) and Pandas (unique() by default) at least.

Copy link
Member

@HyukjinKwon HyukjinKwon May 23, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looking into the current impl, seems we ignore null as distinct values:

spark/sql/core/src/main/scala/org/apache/spark/sql/execution/command/CommandUtils.scala

Line 270 in 239082d

val numNonNulls = if (col.nullable) Count(col) else Count(one)

spark/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/statsEstimation/EstimationUtils.scala

Lines 50 to 51 in b1857a4

ColumnStat(distinctCount = Some(0), min = None, max = None, nullCount = Some(rowCount),
avgLen = Some(dataType.defaultSize), maxLen = Some(dataType.defaultSize))

} else {
distinctCount
}
Expand Down
13 changes: 12 additions & 1 deletion ...c/test/scala/org/apache/spark/sql/catalyst/statsEstimation/AggregateEstimationSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,9 @@ class AggregateEstimationSuite extends StatsEstimationTestBase with PlanTest {
attr("key31") -> ColumnStat(distinctCount = Some(0), min = None, max = None,
nullCount = Some(0), avgLen = Some(4), maxLen = Some(4)),
attr("key32") -> ColumnStat(distinctCount = Some(0), min = None, max = None,
nullCount = Some(4), avgLen = Some(4), maxLen = Some(4))
nullCount = Some(4), avgLen = Some(4), maxLen = Some(4)),
attr("key33") -> ColumnStat(distinctCount = Some(2), min = None, max = None,
nullCount = Some(2), avgLen = Some(4), maxLen = Some(4))
))

private val nameToAttr: Map[String, Attribute] = columnInfo.map(kv => kv._1.name -> kv._1)
Expand Down Expand Up @@ -126,6 +128,15 @@ class AggregateEstimationSuite extends StatsEstimationTestBase with PlanTest {
expectedOutputRowCount = nameToColInfo("key22")._2.distinctCount.get)
}

test("group-by column with null value") {
checkAggStats(
tableColumns = Seq("key21", "key33"),
tableRowCount = 6,
groupByColumns = Seq("key21", "key33"),
expectedOutputRowCount = nameToColInfo("key21")._2.distinctCount.get *
(nameToColInfo("key33")._2.distinctCount.get + 1))
}

test("non-cbo estimation") {
val attributes = Seq("key12").map(nameToAttr)
val child = StatsTestPlan(
Expand Down