Github user yanboliang commented on a diff in the pull request: https://github.com/apache/spark/pull/18305#discussion_r126582816 --- Diff: mllib/src/test/scala/org/apache/spark/ml/optim/aggregator/LogisticAggregatorSuite.scala --- @@ -0,0 +1,254 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.spark.ml.optim.aggregator + +import org.apache.spark.SparkFunSuite +import org.apache.spark.ml.feature.Instance +import org.apache.spark.ml.linalg.{BLAS, Matrices, Vector, Vectors} +import org.apache.spark.ml.util.TestingUtils._ +import org.apache.spark.mllib.util.MLlibTestSparkContext + +class LogisticAggregatorSuite extends SparkFunSuite with MLlibTestSparkContext { + + import DifferentiableLossAggregatorSuite.getClassificationSummarizers + + @transient var instances: Array[Instance] = _ + @transient var instancesConstantFeature: Array[Instance] = _ + + override def beforeAll(): Unit = { + super.beforeAll() + instances = Array( + Instance(0.0, 0.1, Vectors.dense(1.0, 2.0)), + Instance(1.0, 0.5, Vectors.dense(1.5, 1.0)), + Instance(2.0, 0.3, Vectors.dense(4.0, 0.5)) + ) + instancesConstantFeature = Array( + Instance(0.0, 0.1, Vectors.dense(1.0, 2.0)), + Instance(1.0, 0.5, Vectors.dense(1.0, 1.0)), + Instance(2.0, 0.3, Vectors.dense(1.0, 0.5)) + ) + } + + + /** Get summary statistics for some data and create a new LogisticAggregator. */ + private def getNewAggregator( + instances: Array[Instance], + coefficients: Vector, + fitIntercept: Boolean, + isMultinomial: Boolean): LogisticAggregator = { + val (featuresSummarizer, ySummarizer) = + DifferentiableLossAggregatorSuite.getClassificationSummarizers(instances) + val numClasses = ySummarizer.histogram.length + val featuresStd = featuresSummarizer.variance.toArray.map(math.sqrt) + val bcFeaturesStd = spark.sparkContext.broadcast(featuresStd) + val bcCoefficients = spark.sparkContext.broadcast(coefficients) --- End diff -- I think we always try to destroy broadcast variable explicitly both in source code and test cases, like [here](https://github.com/apache/spark/pull/18152). Of course, these broadcast variables can be destroyed after spark session is torn down. The reason of why we do this in source code is users application may be long-time running, so it will accumulate lots of these variables, waste lots of resource and slower your application. The reason of why we do this in test case is we should keep same code route as in source code. Since we have encountered similar bugs which was not covered by test cases. But in this case, I think it's safe to not destroy these variables. I just suggested to follow MLlib's convention. Thanks.
--- 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 infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org