diff --git a/core/src/main/scala/org/apache/spark/rdd/RDD.scala b/core/src/main/scala/org/apache/spark/rdd/RDD.scala index 8baf199f215fb438f9c7bbfdde1c08ae05761602..7dad30ecbdd2f9a65144eba230539c44f9cbe1ef 100644 --- a/core/src/main/scala/org/apache/spark/rdd/RDD.scala +++ b/core/src/main/scala/org/apache/spark/rdd/RDD.scala @@ -1161,8 +1161,8 @@ abstract class RDD[T: ClassTag]( */ @Experimental def countApproxDistinct(p: Int, sp: Int): Long = withScope { - require(p >= 4, s"p ($p) must be at least 4") - require(sp <= 32, s"sp ($sp) cannot be greater than 32") + require(p >= 4, s"p ($p) must be >= 4") + require(sp <= 32, s"sp ($sp) must be <= 32") require(sp == 0 || p <= sp, s"p ($p) cannot be greater than sp ($sp)") val zeroCounter = new HyperLogLogPlus(p, sp) aggregate(zeroCounter)( @@ -1187,8 +1187,9 @@ abstract class RDD[T: ClassTag]( * It must be greater than 0.000017. */ def countApproxDistinct(relativeSD: Double = 0.05): Long = withScope { + require(relativeSD > 0.000017, s"accuracy ($relativeSD) must be greater than 0.000017") val p = math.ceil(2.0 * math.log(1.054 / relativeSD) / math.log(2)).toInt - countApproxDistinct(p, 0) + countApproxDistinct(if (p < 4) 4 else p, 0) } /** diff --git a/core/src/test/scala/org/apache/spark/rdd/RDDSuite.scala b/core/src/test/scala/org/apache/spark/rdd/RDDSuite.scala index ef8c36a28655b174882869936f608ff333d8ab37..afc11bdc4d6aba7155baf81cb81ae4605bf35276 100644 --- a/core/src/test/scala/org/apache/spark/rdd/RDDSuite.scala +++ b/core/src/test/scala/org/apache/spark/rdd/RDDSuite.scala @@ -89,6 +89,8 @@ class RDDSuite extends FunSuite with SharedSparkContext { val simpleRdd = sc.makeRDD(uniformDistro, 10) assert(error(simpleRdd.countApproxDistinct(8, 0), size) < 0.2) assert(error(simpleRdd.countApproxDistinct(12, 0), size) < 0.1) + assert(error(simpleRdd.countApproxDistinct(0.02), size) < 0.1) + assert(error(simpleRdd.countApproxDistinct(0.5), size) < 0.22) } test("SparkContext.union") { diff --git a/python/pyspark/rdd.py b/python/pyspark/rdd.py index d254deb527d1052843ab16e515fb44f4604dfa88..545c5ad20cb96cbed06103522c40dea177e52b44 100644 --- a/python/pyspark/rdd.py +++ b/python/pyspark/rdd.py @@ -2239,8 +2239,6 @@ class RDD(object): """ if relativeSD < 0.000017: raise ValueError("relativeSD should be greater than 0.000017") - if relativeSD > 0.37: - raise ValueError("relativeSD should be smaller than 0.37") # the hash space in Java is 2^32 hashRDD = self.map(lambda x: portable_hash(x) & 0xFFFFFFFF) return hashRDD._to_java_object_rdd().countApproxDistinct(relativeSD) diff --git a/python/pyspark/tests.py b/python/pyspark/tests.py index ea63a396da5b8826f28a70cfca680d6350755e01..09de4d159fdcf4a65981b1a697ef3b9fdd1d9821 100644 --- a/python/pyspark/tests.py +++ b/python/pyspark/tests.py @@ -644,7 +644,6 @@ class RDDTests(ReusedPySparkTestCase): self.assertTrue(18 < rdd.map(lambda x: (x, -x)).countApproxDistinct() < 22) self.assertRaises(ValueError, lambda: rdd.countApproxDistinct(0.00000001)) - self.assertRaises(ValueError, lambda: rdd.countApproxDistinct(0.5)) def test_histogram(self): # empty