Mailing-List: contact commits-help@spark.apache.org; run by ezmlm
Precedence: bulk
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
From: srowen@apache.org
To: commits@spark.apache.org
Message-Id: <bf0986eed5564b6aa7243dec6204d9b1@git.apache.org>
Subject: spark git commit: Log warnings for numIterations * miniBatchFraction
 < 1.0
Date: Wed, 25 May 2016 21:52:34 +0000 (UTC)
archived-at: Wed, 25 May 2016 22:29:10 -0000

Repository: spark
Updated Branches:
  refs/heads/master 9c297df3d -> 589cce93c


Log warnings for numIterations * miniBatchFraction < 1.0

## What changes were proposed in this pull request?

Add a warning log for the case that `numIterations * miniBatchFraction <1.0` during gradient descent. If the product of those two numbers is less than `1.0`, then not all training examples will be used during optimization. To put this concretely, suppose that `numExamples = 100`, `miniBatchFraction = 0.2` and `numIterations = 3`. Then, 3 iterations will occur each sampling approximately 6 examples each. In the best case, each of the 6 examples are unique; hence 18/100 examples are used.

This may be counter-intuitive to most users and led to the issue during the development of another Spark  ML model: https://github.com/zhengruifeng/spark-libFM/issues/11. If a user actually does not require the training data set, it would be easier and more intuitive to use `RDD.sample`.

## How was this patch tested?

`build/mvn -DskipTests clean package` build succeeds

Author: Gio Borje <gborje@linkedin.com>

Closes #13265 from Hydrotoast/master.


Project: http://git-wip-us.apache.org/repos/asf/spark/repo
Commit: http://git-wip-us.apache.org/repos/asf/spark/commit/589cce93
Tree: http://git-wip-us.apache.org/repos/asf/spark/tree/589cce93
Diff: http://git-wip-us.apache.org/repos/asf/spark/diff/589cce93

Branch: refs/heads/master
Commit: 589cce93c821ac28e9090a478f6e7465398b7c30
Parents: 9c297df
Author: Gio Borje <gborje@linkedin.com>
Authored: Wed May 25 16:52:31 2016 -0500
Committer: Sean Owen <sowen@cloudera.com>
Committed: Wed May 25 16:52:31 2016 -0500

----------------------------------------------------------------------
 .../org/apache/spark/mllib/optimization/GradientDescent.scala   | 5 +++++
 1 file changed, 5 insertions(+)
----------------------------------------------------------------------


http://git-wip-us.apache.org/repos/asf/spark/blob/589cce93/mllib/src/main/scala/org/apache/spark/mllib/optimization/GradientDescent.scala
----------------------------------------------------------------------
diff --git a/mllib/src/main/scala/org/apache/spark/mllib/optimization/GradientDescent.scala b/mllib/src/main/scala/org/apache/spark/mllib/optimization/GradientDescent.scala
index a67ea83..735e780 100644
--- a/mllib/src/main/scala/org/apache/spark/mllib/optimization/GradientDescent.scala
+++ b/mllib/src/main/scala/org/apache/spark/mllib/optimization/GradientDescent.scala
@@ -197,6 +197,11 @@ object GradientDescent extends Logging {
         "< 1.0 can be unstable because of the stochasticity in sampling.")
     }
 
+    if (numIterations * miniBatchFraction < 1.0) {
+      logWarning("Not all examples will be used if numIterations * miniBatchFraction < 1.0: " +
+        s"numIterations=$numIterations and miniBatchFraction=$miniBatchFraction")
+    }
+
     val stochasticLossHistory = new ArrayBuffer[Double](numIterations)
     // Record previous weight and current one to calculate solution vector difference
 

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@spark.apache.org
For additional commands, e-mail: commits-help@spark.apache.org