[SPARK-3130][MLLIB] detect negative values in naive Bayes

because NB treats feature values as term frequencies. jkbradley

Author: Xiangrui Meng <meng@databricks.com>

Closes #2038 from mengxr/nb-neg and squashes the following commits:

52c37c3 [Xiangrui Meng] address comments
65f892d [Xiangrui Meng] detect negative values in nb

1 files changed, 28 insertions, 0 deletions

diff --git a/mllib/src/test/scala/org/apache/spark/mllib/classification/NaiveBayesSuite.scala b/mllib/src/test/scala/org/apache/spark/mllib/classification/NaiveBayesSuite.scala
index 06cdd04f5f..80989bc074 100644
--- a/mllib/src/test/scala/org/apache/spark/mllib/classification/NaiveBayesSuite.scala
+++ b/mllib/src/test/scala/org/apache/spark/mllib/classification/NaiveBayesSuite.scala
@@ -21,6 +21,7 @@ import scala.util.Random
 
 import org.scalatest.FunSuite
 
+import org.apache.spark.SparkException
 import org.apache.spark.mllib.linalg.Vectors
 import org.apache.spark.mllib.regression.LabeledPoint
 import org.apache.spark.mllib.util.{LocalClusterSparkContext, LocalSparkContext}
@@ -95,6 +96,33 @@ class NaiveBayesSuite extends FunSuite with LocalSparkContext {
     // Test prediction on Array.
     validatePrediction(validationData.map(row => model.predict(row.features)), validationData)
   }
+
+  test("detect negative values") {
+    val dense = Seq(
+      LabeledPoint(1.0, Vectors.dense(1.0)),
+      LabeledPoint(0.0, Vectors.dense(-1.0)),
+      LabeledPoint(1.0, Vectors.dense(1.0)),
+      LabeledPoint(1.0, Vectors.dense(0.0)))
+    intercept[SparkException] {
+      NaiveBayes.train(sc.makeRDD(dense, 2))
+    }
+    val sparse = Seq(
+      LabeledPoint(1.0, Vectors.sparse(1, Array(0), Array(1.0))),
+      LabeledPoint(0.0, Vectors.sparse(1, Array(0), Array(-1.0))),
+      LabeledPoint(1.0, Vectors.sparse(1, Array(0), Array(1.0))),
+      LabeledPoint(1.0, Vectors.sparse(1, Array.empty, Array.empty)))
+    intercept[SparkException] {
+      NaiveBayes.train(sc.makeRDD(sparse, 2))
+    }
+    val nan = Seq(
+      LabeledPoint(1.0, Vectors.sparse(1, Array(0), Array(1.0))),
+      LabeledPoint(0.0, Vectors.sparse(1, Array(0), Array(Double.NaN))),
+      LabeledPoint(1.0, Vectors.sparse(1, Array(0), Array(1.0))),
+      LabeledPoint(1.0, Vectors.sparse(1, Array.empty, Array.empty)))
+    intercept[SparkException] {
+      NaiveBayes.train(sc.makeRDD(nan, 2))
+    }
+  }
 }
 
 class NaiveBayesClusterSuite extends FunSuite with LocalClusterSparkContext {