Github user MLnick commented on a diff in the pull request:
https://github.com/apache/spark/pull/11553#discussion_r59171165
--- Diff:
mllib/src/test/scala/org/apache/spark/ml/feature/QuantileDiscretizerSuite.scala
---
@@ -17,78 +17,59 @@
package org.apache.spark.ml.feature
-import org.apache.spark.{SparkContext, SparkFunSuite}
-import org.apache.spark.ml.attribute.{Attribute, NominalAttribute}
+import org.apache.spark.SparkFunSuite
import org.apache.spark.ml.util.DefaultReadWriteTest
import org.apache.spark.mllib.util.MLlibTestSparkContext
-import org.apache.spark.sql.{Row, SQLContext}
+import org.apache.spark.sql.SQLContext
class QuantileDiscretizerSuite
extends SparkFunSuite with MLlibTestSparkContext with
DefaultReadWriteTest {
- import org.apache.spark.ml.feature.QuantileDiscretizerSuite._
-
- test("Test quantile discretizer") {
- checkDiscretizedData(sc,
- Array[Double](1, 2, 3, 3, 3, 3, 3, 3, 3),
- 10,
- Array[Double](1, 2, 3, 3, 3, 3, 3, 3, 3),
- Array("-Infinity, 1.0", "1.0, 2.0", "2.0, 3.0", "3.0, Infinity"))
-
- checkDiscretizedData(sc,
- Array[Double](1, 2, 3, 3, 3, 3, 3, 3, 3),
- 4,
- Array[Double](1, 2, 3, 3, 3, 3, 3, 3, 3),
- Array("-Infinity, 1.0", "1.0, 2.0", "2.0, 3.0", "3.0, Infinity"))
-
- checkDiscretizedData(sc,
- Array[Double](1, 2, 3, 3, 3, 3, 3, 3, 3),
- 3,
- Array[Double](0, 1, 2, 2, 2, 2, 2, 2, 2),
- Array("-Infinity, 2.0", "2.0, 3.0", "3.0, Infinity"))
+ test("Test observed number of buckets and their sizes match expected
values") {
+ val sqlCtx = SQLContext.getOrCreate(sc)
+ import sqlCtx.implicits._
- checkDiscretizedData(sc,
- Array[Double](1, 2, 3, 3, 3, 3, 3, 3, 3),
- 2,
- Array[Double](0, 1, 1, 1, 1, 1, 1, 1, 1),
- Array("-Infinity, 2.0", "2.0, Infinity"))
+ val datasetSize = 100000
+ val numBuckets = 5
+ val df = sc.parallelize(1.0 to datasetSize by
1.0).map(Tuple1.apply).toDF("input")
+ val discretizer = new QuantileDiscretizer()
+ .setInputCol("input")
+ .setOutputCol("result")
+ .setNumBuckets(numBuckets)
+ val result = discretizer.fit(df).transform(df)
- }
+ val observedNumBuckets = result.select("result").distinct.count
+ assert(observedNumBuckets === numBuckets,
+ "Observed number of buckets does not equal expected number of
buckets.")
- test("Test getting splits") {
- val splitTestPoints = Array(
- Array[Double]() -> Array(Double.NegativeInfinity, 0,
Double.PositiveInfinity),
- Array(Double.NegativeInfinity) -> Array(Double.NegativeInfinity, 0,
Double.PositiveInfinity),
- Array(Double.PositiveInfinity) -> Array(Double.NegativeInfinity, 0,
Double.PositiveInfinity),
- Array(Double.NegativeInfinity, Double.PositiveInfinity)
- -> Array(Double.NegativeInfinity, 0, Double.PositiveInfinity),
- Array(0.0) -> Array(Double.NegativeInfinity, 0,
Double.PositiveInfinity),
- Array(1.0) -> Array(Double.NegativeInfinity, 1,
Double.PositiveInfinity),
- Array(0.0, 1.0) -> Array(Double.NegativeInfinity, 0, 1,
Double.PositiveInfinity)
- )
- for ((ori, res) <- splitTestPoints) {
- assert(QuantileDiscretizer.getSplits(ori) === res, "Returned splits
are invalid.")
+ val relativeError = discretizer.getRelativeError
+ val isGoodBucket = org.apache.spark.sql.functions.udf {
+ (size: Int) => math.abs( size - (datasetSize / numBuckets)) <=
(relativeError * datasetSize)
}
+ val numGoodBuckets =
result.groupBy("result").count.filter(isGoodBucket($"count")).count
+ assert(numGoodBuckets === numBuckets,
+ "Bucket sizes are not as expected.")
}
- test("Test splits on dataset larger than minSamplesRequired") {
+ test("Test transform method on unseen data") {
val sqlCtx = SQLContext.getOrCreate(sc)
import sqlCtx.implicits._
- val datasetSize = QuantileDiscretizer.minSamplesRequired + 1
- val numBuckets = 5
- val df = sc.parallelize((1.0 to datasetSize by
1.0).map(Tuple1.apply)).toDF("input")
+ val trainDF = sc.parallelize(1.0 to 100.0 by
1.0).map(Tuple1.apply).toDF("input")
+ val testDF = sc.parallelize(-10.0 to 110.0 by
1.0).map(Tuple1.apply).toDF("input")
val discretizer = new QuantileDiscretizer()
.setInputCol("input")
.setOutputCol("result")
- .setNumBuckets(numBuckets)
- .setSeed(1)
+ .setNumBuckets(5)
- val result = discretizer.fit(df).transform(df)
- val observedNumBuckets = result.select("result").distinct.count
+ val result = discretizer.fit(trainDF).transform(testDF)
+ val firstBucketSize = result.filter(result("result") === 0.0).count
+ val lastBucketSize = result.filter(result("result") === 4.0).count
- assert(observedNumBuckets === numBuckets,
- "Observed number of buckets does not equal expected number of
buckets.")
+ assert(firstBucketSize === 30L,
+ s"Size of first bucket ${firstBucketSize} did not equal expected
value 30.")
--- End diff --
expected value of 30
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
