Skip to content

[SPARK-29144][ML] Binarizer handle sparse vectors incorrectly with negative threshold #25829

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
wants to merge 3 commits into from
Closed

[SPARK-29144][ML] Binarizer handle sparse vectors incorrectly with negative threshold #25829

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e3a5ff4
create pr
zhengruifeng Sep 18, 2019
cb52a09
nit
zhengruifeng Sep 18, 2019
190c3b8
swith to ML fashion
zhengruifeng Sep 19, 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
48 changes: 29 additions & 19 deletions mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,30 +75,40 @@ final class Binarizer @Since("1.4.0") (@Since("1.4.0") override val uid: String)
val schema = dataset.schema
val inputType = schema($(inputCol)).dataType
val td = $(threshold)
val metadata = outputSchema($(outputCol)).metadata

val binarizerDouble = udf { in: Double => if (in > td) 1.0 else 0.0 }
val binarizerVector = udf { (data: Vector) =>
val indices = ArrayBuilder.make[Int]
val values = ArrayBuilder.make[Double]

data.foreachActive { (index, value) =>
if (value > td) {
indices += index
values += 1.0
val binarizerUDF = inputType match {
case DoubleType =>
udf { in: Double => if (in > td) 1.0 else 0.0 }

case _: VectorUDT if td >= 0 =>
udf { vector: Vector =>
val indices = ArrayBuilder.make[Int]
val values = ArrayBuilder.make[Double]
vector.foreachActive { (index, value) =>
if (value > td) {
indices += index
values += 1.0
}
}
Vectors.sparse(vector.size, indices.result(), values.result()).compressed
}
}

Vectors.sparse(data.size, indices.result(), values.result()).compressed
case _: VectorUDT if td < 0 =>
this.logWarning(s"Binarization operations on sparse dataset with negative threshold " +
Copy link
Member

Choose a reason for hiding this comment

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

I think this is OK. It will almost always be dense but not always. The warning is spurious if the input is dense already, but, a negative threshold is rare... I think. I'm trying to recall whether this is ever applied to outputs of classifiers like SVMs that output [-1, 1].

s"$td will build a dense output, so take care when applying to sparse input.")
udf { vector: Vector =>
val values = Array.fill(vector.size)(1.0)
vector.foreachActive { (index, value) =>
if (value <= td) {
values(index) = 0.0
}
}
Vectors.dense(values).compressed
}
}

val metadata = outputSchema($(outputCol)).metadata

inputType match {
case DoubleType =>
dataset.select(col("*"), binarizerDouble(col($(inputCol))).as($(outputCol), metadata))
case _: VectorUDT =>
dataset.select(col("*"), binarizerVector(col($(inputCol))).as($(outputCol), metadata))
}
dataset.withColumn($(outputCol), binarizerUDF(col($(inputCol))), metadata)
}

@Since("1.4.0")
Expand Down
14 changes: 14 additions & 0 deletions mllib/src/test/scala/org/apache/spark/ml/feature/BinarizerSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,6 +101,20 @@ class BinarizerSuite extends MLTest with DefaultReadWriteTest {
}
}

test("Binarizer should support sparse vector with negative threshold") {
val data = Seq(
(Vectors.sparse(3, Array(1), Array(0.5)), Vectors.dense(Array(1.0, 1.0, 1.0))),
(Vectors.dense(Array(0.0, 0.5, 0.0)), Vectors.dense(Array(1.0, 1.0, 1.0))))
val df = data.toDF("feature", "expected")
val binarizer = new Binarizer()
.setInputCol("feature")
.setOutputCol("binarized_feature")
.setThreshold(-0.5)
binarizer.transform(df).select("binarized_feature", "expected").collect().foreach {
case Row(x: Vector, y: Vector) =>
assert(x == y, "The feature value is not correct after binarization.")
}
}

test("read/write") {
val t = new Binarizer()
Expand Down