From c5912ecc7b392a13089ae735c07c2d7256de36c6 Mon Sep 17 00:00:00 2001
From: Xiangrui Meng <meng@databricks.com>
Date: Mon, 3 Nov 2014 18:50:37 -0800
Subject: [PATCH] [FIX][MLLIB] fix seed in BaggedPointSuite
Saw Jenkins test failures due to random seeds.
jkbradley manishamde
Author: Xiangrui Meng <meng@databricks.com>
Closes #3084 from mengxr/fix-baggedpoint-suite and squashes the following commits:
f735a43 [Xiangrui Meng] fix seed in BaggedPointSuite
---
.../spark/mllib/tree/impl/BaggedPointSuite.scala | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/mllib/src/test/scala/org/apache/spark/mllib/tree/impl/BaggedPointSuite.scala b/mllib/src/test/scala/org/apache/spark/mllib/tree/impl/BaggedPointSuite.scala
index c0a62e0043..5cb433232e 100644
--- a/mllib/src/test/scala/org/apache/spark/mllib/tree/impl/BaggedPointSuite.scala
+++ b/mllib/src/test/scala/org/apache/spark/mllib/tree/impl/BaggedPointSuite.scala
@@ -30,7 +30,7 @@ class BaggedPointSuite extends FunSuite with LocalSparkContext {
test("BaggedPoint RDD: without subsampling") {
val arr = EnsembleTestHelper.generateOrderedLabeledPoints(1, 1000)
val rdd = sc.parallelize(arr)
- val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, 1.0, 1, false)
+ val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, 1.0, 1, false, 42)
baggedRDD.collect().foreach { baggedPoint =>
assert(baggedPoint.subsampleWeights.size == 1 && baggedPoint.subsampleWeights(0) == 1)
}
@@ -44,7 +44,7 @@ class BaggedPointSuite extends FunSuite with LocalSparkContext {
val arr = EnsembleTestHelper.generateOrderedLabeledPoints(1, 1000)
val rdd = sc.parallelize(arr)
seeds.foreach { seed =>
- val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, 1.0, numSubsamples, true)
+ val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, 1.0, numSubsamples, true, seed)
val subsampleCounts: Array[Array[Double]] = baggedRDD.map(_.subsampleWeights).collect()
EnsembleTestHelper.testRandomArrays(subsampleCounts, numSubsamples, expectedMean,
expectedStddev, epsilon = 0.01)
@@ -60,7 +60,7 @@ class BaggedPointSuite extends FunSuite with LocalSparkContext {
val arr = EnsembleTestHelper.generateOrderedLabeledPoints(1, 1000)
val rdd = sc.parallelize(arr)
seeds.foreach { seed =>
- val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, subsample, numSubsamples, true)
+ val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, subsample, numSubsamples, true, seed)
val subsampleCounts: Array[Array[Double]] = baggedRDD.map(_.subsampleWeights).collect()
EnsembleTestHelper.testRandomArrays(subsampleCounts, numSubsamples, expectedMean,
expectedStddev, epsilon = 0.01)
@@ -75,7 +75,7 @@ class BaggedPointSuite extends FunSuite with LocalSparkContext {
val arr = EnsembleTestHelper.generateOrderedLabeledPoints(1, 1000)
val rdd = sc.parallelize(arr)
seeds.foreach { seed =>
- val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, 1.0, numSubsamples, false)
+ val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, 1.0, numSubsamples, false, seed)
val subsampleCounts: Array[Array[Double]] = baggedRDD.map(_.subsampleWeights).collect()
EnsembleTestHelper.testRandomArrays(subsampleCounts, numSubsamples, expectedMean,
expectedStddev, epsilon = 0.01)
@@ -91,7 +91,7 @@ class BaggedPointSuite extends FunSuite with LocalSparkContext {
val arr = EnsembleTestHelper.generateOrderedLabeledPoints(1, 1000)
val rdd = sc.parallelize(arr)
seeds.foreach { seed =>
- val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, subsample, numSubsamples, false)
+ val baggedRDD = BaggedPoint.convertToBaggedRDD(rdd, subsample, numSubsamples, false, seed)
val subsampleCounts: Array[Array[Double]] = baggedRDD.map(_.subsampleWeights).collect()
EnsembleTestHelper.testRandomArrays(subsampleCounts, numSubsamples, expectedMean,
expectedStddev, epsilon = 0.01)
--
GitLab