[SPARK-7407] [MLLIB] use uid + name to identify parameters A param instance is strongly attached to an parent in the current implementation. So if we make a copy of an estimator or a transformer in pipelines and other meta-algorithms, it becomes error-prone to copy the params to the copied instances. In this PR, a param is identified by its parent's UID and the param name. So it becomes loosely attached to its parent and all its derivatives. The UID is preserved during copying or fitting. All components now have a default constructor and a constructor that takes a UID as input. I keep the constructors for Param in this PR to reduce the amount of diff and moved `parent` as a mutable field.
This PR still needs some clean-ups, and there are several spark.ml PRs pending. I'll try to get them merged first and then update this PR. jkbradley Author: Xiangrui Meng <[email protected]> Closes #6019 from mengxr/SPARK-7407 and squashes the following commits: c4c8120 [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-7407 520f0a2 [Xiangrui Meng] address comments 2569168 [Xiangrui Meng] fix tests 873caca [Xiangrui Meng] fix tests in OneVsRest; fix a racing condition in shouldOwn 409ea08 [Xiangrui Meng] minor updates 83a163c [Xiangrui Meng] update JavaDeveloperApiExample 5db5325 [Xiangrui Meng] update OneVsRest 7bde7ae [Xiangrui Meng] merge master 697fdf9 [Xiangrui Meng] update Bucketizer 7b4f6c2 [Xiangrui Meng] Merge remote-tracking branch 'apache/master' into SPARK-7407 629d402 [Xiangrui Meng] fix LRSuite 154516f [Xiangrui Meng] merge master aa4a611 [Xiangrui Meng] fix examples/compile a4794dd [Xiangrui Meng] change Param to use to reduce the size of diff fdbc415 [Xiangrui Meng] all tests passed c255f17 [Xiangrui Meng] fix tests in ParamsSuite 818e1db [Xiangrui Meng] merge master e1160cf [Xiangrui Meng] fix tests fbc39f0 [Xiangrui Meng] pass test:compile 108937e [Xiangrui Meng] pass compile 8726d39 [Xiangrui Meng] use parent uid in Param eaeed35 [Xiangrui Meng] update Identifiable (cherry picked from commit 1b8625f4258d6d1a049d0ba60e39e9757f5a568b) Signed-off-by: Xiangrui Meng <[email protected]> Project: http://git-wip-us.apache.org/repos/asf/spark/repo Commit: http://git-wip-us.apache.org/repos/asf/spark/commit/e45cd9f7 Tree: http://git-wip-us.apache.org/repos/asf/spark/tree/e45cd9f7 Diff: http://git-wip-us.apache.org/repos/asf/spark/diff/e45cd9f7 Branch: refs/heads/branch-1.4 Commit: e45cd9f73a624d86e09f3a0f5649fa5dc7090b38 Parents: c80e0cf Author: Xiangrui Meng <[email protected]> Authored: Thu May 14 01:22:15 2015 -0700 Committer: Xiangrui Meng <[email protected]> Committed: Thu May 14 01:22:32 2015 -0700 ---------------------------------------------------------------------- .../examples/ml/JavaDeveloperApiExample.java | 43 +++++-- .../spark/examples/ml/DeveloperApiExample.scala | 11 +- .../main/scala/org/apache/spark/ml/Model.scala | 10 +- .../scala/org/apache/spark/ml/Pipeline.scala | 11 +- .../scala/org/apache/spark/ml/Predictor.scala | 2 +- .../classification/DecisionTreeClassifier.scala | 13 ++- .../spark/ml/classification/GBTClassifier.scala | 13 ++- .../ml/classification/LogisticRegression.scala | 11 +- .../spark/ml/classification/OneVsRest.scala | 27 ++--- .../classification/RandomForestClassifier.scala | 13 ++- .../BinaryClassificationEvaluator.scala | 7 +- .../org/apache/spark/ml/feature/Binarizer.scala | 7 +- .../apache/spark/ml/feature/Bucketizer.scala | 8 +- .../spark/ml/feature/ElementwiseProduct.scala | 6 +- .../org/apache/spark/ml/feature/HashingTF.scala | 5 +- .../scala/org/apache/spark/ml/feature/IDF.scala | 10 +- .../apache/spark/ml/feature/Normalizer.scala | 5 +- .../apache/spark/ml/feature/OneHotEncoder.scala | 8 +- .../spark/ml/feature/PolynomialExpansion.scala | 6 +- .../spark/ml/feature/StandardScaler.scala | 10 +- .../apache/spark/ml/feature/StringIndexer.scala | 10 +- .../org/apache/spark/ml/feature/Tokenizer.scala | 10 +- .../spark/ml/feature/VectorAssembler.scala | 6 +- .../apache/spark/ml/feature/VectorIndexer.scala | 13 ++- .../org/apache/spark/ml/feature/Word2Vec.scala | 10 +- .../org/apache/spark/ml/param/params.scala | 115 +++++++++++++------ .../apache/spark/ml/recommendation/ALS.scala | 10 +- .../ml/regression/DecisionTreeRegressor.scala | 13 ++- .../spark/ml/regression/GBTRegressor.scala | 13 ++- .../spark/ml/regression/LinearRegression.scala | 17 +-- .../ml/regression/RandomForestRegressor.scala | 12 +- .../apache/spark/ml/tuning/CrossValidator.scala | 10 +- .../org/apache/spark/ml/util/Identifiable.scala | 21 +++- .../JavaLogisticRegressionSuite.java | 4 +- .../apache/spark/ml/param/JavaTestParams.java | 52 ++++++--- .../regression/JavaLinearRegressionSuite.java | 4 +- .../spark/ml/util/IdentifiableSuite.scala | 40 +++++++ .../DecisionTreeClassifierSuite.scala | 2 +- .../ml/classification/GBTClassifierSuite.scala | 2 +- .../LogisticRegressionSuite.scala | 39 +++---- .../ml/classification/OneVsRestSuite.scala | 6 +- .../RandomForestClassifierSuite.scala | 2 +- .../org/apache/spark/ml/param/ParamsSuite.scala | 16 ++- .../org/apache/spark/ml/param/TestParams.scala | 5 +- .../regression/DecisionTreeRegressorSuite.scala | 2 +- .../spark/ml/regression/GBTRegressorSuite.scala | 3 +- .../regression/RandomForestRegressorSuite.scala | 2 +- 47 files changed, 452 insertions(+), 213 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/examples/src/main/java/org/apache/spark/examples/ml/JavaDeveloperApiExample.java ---------------------------------------------------------------------- diff --git a/examples/src/main/java/org/apache/spark/examples/ml/JavaDeveloperApiExample.java b/examples/src/main/java/org/apache/spark/examples/ml/JavaDeveloperApiExample.java index eac4f89..ec533d1 100644 --- a/examples/src/main/java/org/apache/spark/examples/ml/JavaDeveloperApiExample.java +++ b/examples/src/main/java/org/apache/spark/examples/ml/JavaDeveloperApiExample.java @@ -28,6 +28,7 @@ import org.apache.spark.ml.classification.Classifier; import org.apache.spark.ml.classification.ClassificationModel; import org.apache.spark.ml.param.IntParam; import org.apache.spark.ml.param.ParamMap; +import org.apache.spark.ml.util.Identifiable$; import org.apache.spark.mllib.linalg.BLAS; import org.apache.spark.mllib.linalg.Vector; import org.apache.spark.mllib.linalg.Vectors; @@ -103,7 +104,23 @@ public class JavaDeveloperApiExample { * However, this should still compile and run successfully. */ class MyJavaLogisticRegression - extends Classifier<Vector, MyJavaLogisticRegression, MyJavaLogisticRegressionModel> { + extends Classifier<Vector, MyJavaLogisticRegression, MyJavaLogisticRegressionModel> { + + public MyJavaLogisticRegression() { + init(); + } + + public MyJavaLogisticRegression(String uid) { + this.uid_ = uid; + init(); + } + + private String uid_ = Identifiable$.MODULE$.randomUID("myJavaLogReg"); + + @Override + public String uid() { + return uid_; + } /** * Param for max number of iterations @@ -117,7 +134,7 @@ class MyJavaLogisticRegression int getMaxIter() { return (Integer) getOrDefault(maxIter); } - public MyJavaLogisticRegression() { + private void init() { setMaxIter(100); } @@ -137,7 +154,7 @@ class MyJavaLogisticRegression Vector weights = Vectors.zeros(numFeatures); // Learning would happen here. // Create a model, and return it. - return new MyJavaLogisticRegressionModel(this, weights); + return new MyJavaLogisticRegressionModel(uid(), weights).setParent(this); } } @@ -149,17 +166,21 @@ class MyJavaLogisticRegression * However, this should still compile and run successfully. */ class MyJavaLogisticRegressionModel - extends ClassificationModel<Vector, MyJavaLogisticRegressionModel> { - - private MyJavaLogisticRegression parent_; - public MyJavaLogisticRegression parent() { return parent_; } + extends ClassificationModel<Vector, MyJavaLogisticRegressionModel> { private Vector weights_; public Vector weights() { return weights_; } - public MyJavaLogisticRegressionModel(MyJavaLogisticRegression parent_, Vector weights_) { - this.parent_ = parent_; - this.weights_ = weights_; + public MyJavaLogisticRegressionModel(String uid, Vector weights) { + this.uid_ = uid; + this.weights_ = weights; + } + + private String uid_ = Identifiable$.MODULE$.randomUID("myJavaLogReg"); + + @Override + public String uid() { + return uid_; } // This uses the default implementation of transform(), which reads column "features" and outputs @@ -204,6 +225,6 @@ class MyJavaLogisticRegressionModel */ @Override public MyJavaLogisticRegressionModel copy(ParamMap extra) { - return copyValues(new MyJavaLogisticRegressionModel(parent_, weights_), extra); + return copyValues(new MyJavaLogisticRegressionModel(uid(), weights_), extra); } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/examples/src/main/scala/org/apache/spark/examples/ml/DeveloperApiExample.scala ---------------------------------------------------------------------- diff --git a/examples/src/main/scala/org/apache/spark/examples/ml/DeveloperApiExample.scala b/examples/src/main/scala/org/apache/spark/examples/ml/DeveloperApiExample.scala index 2a2d067..3ee456e 100644 --- a/examples/src/main/scala/org/apache/spark/examples/ml/DeveloperApiExample.scala +++ b/examples/src/main/scala/org/apache/spark/examples/ml/DeveloperApiExample.scala @@ -20,6 +20,7 @@ package org.apache.spark.examples.ml import org.apache.spark.{SparkConf, SparkContext} import org.apache.spark.ml.classification.{ClassificationModel, Classifier, ClassifierParams} import org.apache.spark.ml.param.{IntParam, ParamMap} +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.linalg.{BLAS, Vector, Vectors} import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.sql.{DataFrame, Row, SQLContext} @@ -106,10 +107,12 @@ private trait MyLogisticRegressionParams extends ClassifierParams { * * NOTE: This is private since it is an example. In practice, you may not want it to be private. */ -private class MyLogisticRegression +private class MyLogisticRegression(override val uid: String) extends Classifier[Vector, MyLogisticRegression, MyLogisticRegressionModel] with MyLogisticRegressionParams { + def this() = this(Identifiable.randomUID("myLogReg")) + setMaxIter(100) // Initialize // The parameter setter is in this class since it should return type MyLogisticRegression. @@ -125,7 +128,7 @@ private class MyLogisticRegression val weights = Vectors.zeros(numFeatures) // Learning would happen here. // Create a model, and return it. - new MyLogisticRegressionModel(this, weights) + new MyLogisticRegressionModel(uid, weights).setParent(this) } } @@ -135,7 +138,7 @@ private class MyLogisticRegression * NOTE: This is private since it is an example. In practice, you may not want it to be private. */ private class MyLogisticRegressionModel( - override val parent: MyLogisticRegression, + override val uid: String, val weights: Vector) extends ClassificationModel[Vector, MyLogisticRegressionModel] with MyLogisticRegressionParams { @@ -173,6 +176,6 @@ private class MyLogisticRegressionModel( * This is used for the default implementation of [[transform()]]. */ override def copy(extra: ParamMap): MyLogisticRegressionModel = { - copyValues(new MyLogisticRegressionModel(parent, weights), extra) + copyValues(new MyLogisticRegressionModel(uid, weights), extra) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/Model.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/Model.scala b/mllib/src/main/scala/org/apache/spark/ml/Model.scala index 9974efe..7fd5153 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/Model.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/Model.scala @@ -32,7 +32,15 @@ abstract class Model[M <: Model[M]] extends Transformer { * The parent estimator that produced this model. * Note: For ensembles' component Models, this value can be null. */ - val parent: Estimator[M] + var parent: Estimator[M] = _ + + /** + * Sets the parent of this model (Java API). + */ + def setParent(parent: Estimator[M]): M = { + this.parent = parent + this.asInstanceOf[M] + } override def copy(extra: ParamMap): M = { // The default implementation of Params.copy doesn't work for models. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/Pipeline.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/Pipeline.scala b/mllib/src/main/scala/org/apache/spark/ml/Pipeline.scala index 33d430f..fac5418 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/Pipeline.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/Pipeline.scala @@ -22,6 +22,7 @@ import scala.collection.mutable.ListBuffer import org.apache.spark.Logging import org.apache.spark.annotation.{AlphaComponent, DeveloperApi} import org.apache.spark.ml.param.{Param, ParamMap, Params} +import org.apache.spark.ml.util.Identifiable import org.apache.spark.sql.DataFrame import org.apache.spark.sql.types.StructType @@ -80,7 +81,9 @@ abstract class PipelineStage extends Params with Logging { * an identity transformer. */ @AlphaComponent -class Pipeline extends Estimator[PipelineModel] { +class Pipeline(override val uid: String) extends Estimator[PipelineModel] { + + def this() = this(Identifiable.randomUID("pipeline")) /** * param for pipeline stages @@ -148,7 +151,7 @@ class Pipeline extends Estimator[PipelineModel] { } } - new PipelineModel(this, transformers.toArray) + new PipelineModel(uid, transformers.toArray).setParent(this) } override def copy(extra: ParamMap): Pipeline = { @@ -171,7 +174,7 @@ class Pipeline extends Estimator[PipelineModel] { */ @AlphaComponent class PipelineModel private[ml] ( - override val parent: Pipeline, + override val uid: String, val stages: Array[Transformer]) extends Model[PipelineModel] with Logging { @@ -190,6 +193,6 @@ class PipelineModel private[ml] ( } override def copy(extra: ParamMap): PipelineModel = { - new PipelineModel(parent, stages) + new PipelineModel(uid, stages) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/Predictor.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/Predictor.scala b/mllib/src/main/scala/org/apache/spark/ml/Predictor.scala index f6a5f27..ec0f76a 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/Predictor.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/Predictor.scala @@ -88,7 +88,7 @@ abstract class Predictor[ // This handles a few items such as schema validation. // Developers only need to implement train(). transformSchema(dataset.schema, logging = true) - copyValues(train(dataset)) + copyValues(train(dataset).setParent(this)) } override def copy(extra: ParamMap): Learner = { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/classification/DecisionTreeClassifier.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/classification/DecisionTreeClassifier.scala b/mllib/src/main/scala/org/apache/spark/ml/classification/DecisionTreeClassifier.scala index dcebea1..7c96133 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/classification/DecisionTreeClassifier.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/classification/DecisionTreeClassifier.scala @@ -21,7 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.{PredictionModel, Predictor} import org.apache.spark.ml.param.ParamMap import org.apache.spark.ml.tree.{TreeClassifierParams, DecisionTreeParams, DecisionTreeModel, Node} -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.tree.{DecisionTree => OldDecisionTree} @@ -39,10 +39,12 @@ import org.apache.spark.sql.DataFrame * features. */ @AlphaComponent -final class DecisionTreeClassifier +final class DecisionTreeClassifier(override val uid: String) extends Predictor[Vector, DecisionTreeClassifier, DecisionTreeClassificationModel] with DecisionTreeParams with TreeClassifierParams { + def this() = this(Identifiable.randomUID("dtc")) + // Override parameter setters from parent trait for Java API compatibility. override def setMaxDepth(value: Int): this.type = super.setMaxDepth(value) @@ -101,7 +103,7 @@ object DecisionTreeClassifier { */ @AlphaComponent final class DecisionTreeClassificationModel private[ml] ( - override val parent: DecisionTreeClassifier, + override val uid: String, override val rootNode: Node) extends PredictionModel[Vector, DecisionTreeClassificationModel] with DecisionTreeModel with Serializable { @@ -114,7 +116,7 @@ final class DecisionTreeClassificationModel private[ml] ( } override def copy(extra: ParamMap): DecisionTreeClassificationModel = { - copyValues(new DecisionTreeClassificationModel(parent, rootNode), extra) + copyValues(new DecisionTreeClassificationModel(uid, rootNode), extra) } override def toString: String = { @@ -138,6 +140,7 @@ private[ml] object DecisionTreeClassificationModel { s"Cannot convert non-classification DecisionTreeModel (old API) to" + s" DecisionTreeClassificationModel (new API). Algo is: ${oldModel.algo}") val rootNode = Node.fromOld(oldModel.topNode, categoricalFeatures) - new DecisionTreeClassificationModel(parent, rootNode) + val uid = if (parent != null) parent.uid else Identifiable.randomUID("dtc") + new DecisionTreeClassificationModel(uid, rootNode) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/classification/GBTClassifier.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/classification/GBTClassifier.scala b/mllib/src/main/scala/org/apache/spark/ml/classification/GBTClassifier.scala index ae51b05..d504d84 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/classification/GBTClassifier.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/classification/GBTClassifier.scala @@ -25,7 +25,7 @@ import org.apache.spark.ml.{PredictionModel, Predictor} import org.apache.spark.ml.param.{Param, ParamMap} import org.apache.spark.ml.regression.DecisionTreeRegressionModel import org.apache.spark.ml.tree.{GBTParams, TreeClassifierParams, DecisionTreeModel, TreeEnsembleModel} -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.tree.{GradientBoostedTrees => OldGBT} @@ -44,10 +44,12 @@ import org.apache.spark.sql.DataFrame * Note: Multiclass labels are not currently supported. */ @AlphaComponent -final class GBTClassifier +final class GBTClassifier(override val uid: String) extends Predictor[Vector, GBTClassifier, GBTClassificationModel] with GBTParams with TreeClassifierParams with Logging { + def this() = this(Identifiable.randomUID("gbtc")) + // Override parameter setters from parent trait for Java API compatibility. // Parameters from TreeClassifierParams: @@ -160,7 +162,7 @@ object GBTClassifier { */ @AlphaComponent final class GBTClassificationModel( - override val parent: GBTClassifier, + override val uid: String, private val _trees: Array[DecisionTreeRegressionModel], private val _treeWeights: Array[Double]) extends PredictionModel[Vector, GBTClassificationModel] @@ -184,7 +186,7 @@ final class GBTClassificationModel( } override def copy(extra: ParamMap): GBTClassificationModel = { - copyValues(new GBTClassificationModel(parent, _trees, _treeWeights), extra) + copyValues(new GBTClassificationModel(uid, _trees, _treeWeights), extra) } override def toString: String = { @@ -210,6 +212,7 @@ private[ml] object GBTClassificationModel { // parent, fittingParamMap for each tree is null since there are no good ways to set these. DecisionTreeRegressionModel.fromOld(tree, null, categoricalFeatures) } - new GBTClassificationModel(parent, newTrees, oldModel.treeWeights) + val uid = if (parent != null) parent.uid else Identifiable.randomUID("gbtc") + new GBTClassificationModel(parent.uid, newTrees, oldModel.treeWeights) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala b/mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala index 2b10362..8694c96 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala @@ -26,6 +26,7 @@ import breeze.optimize.{CachedDiffFunction, DiffFunction} import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.linalg._ import org.apache.spark.mllib.linalg.BLAS._ import org.apache.spark.mllib.regression.LabeledPoint @@ -50,10 +51,12 @@ private[classification] trait LogisticRegressionParams extends ProbabilisticClas * Currently, this class only supports binary classification. */ @AlphaComponent -class LogisticRegression +class LogisticRegression(override val uid: String) extends ProbabilisticClassifier[Vector, LogisticRegression, LogisticRegressionModel] with LogisticRegressionParams with Logging { + def this() = this(Identifiable.randomUID("logreg")) + /** * Set the regularization parameter. * Default is 0.0. @@ -213,7 +216,7 @@ class LogisticRegression (weightsWithIntercept, 0.0) } - new LogisticRegressionModel(this, weights.compressed, intercept) + new LogisticRegressionModel(uid, weights.compressed, intercept) } } @@ -224,7 +227,7 @@ class LogisticRegression */ @AlphaComponent class LogisticRegressionModel private[ml] ( - override val parent: LogisticRegression, + override val uid: String, val weights: Vector, val intercept: Double) extends ProbabilisticClassificationModel[Vector, LogisticRegressionModel] @@ -276,7 +279,7 @@ class LogisticRegressionModel private[ml] ( } override def copy(extra: ParamMap): LogisticRegressionModel = { - copyValues(new LogisticRegressionModel(parent, weights, intercept), extra) + copyValues(new LogisticRegressionModel(uid, weights, intercept), extra) } override protected def raw2prediction(rawPrediction: Vector): Double = { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/classification/OneVsRest.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/classification/OneVsRest.scala b/mllib/src/main/scala/org/apache/spark/ml/classification/OneVsRest.scala index afb8d75..1543f05 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/classification/OneVsRest.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/classification/OneVsRest.scala @@ -25,7 +25,7 @@ import org.apache.spark.annotation.{AlphaComponent, Experimental} import org.apache.spark.ml._ import org.apache.spark.ml.attribute._ import org.apache.spark.ml.param.Param -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.sql.{DataFrame, Row} import org.apache.spark.sql.functions._ @@ -40,19 +40,17 @@ private[ml] trait OneVsRestParams extends PredictorParams { type ClassifierType = Classifier[F, E, M] forSome { type F type M <: ClassificationModel[F, M] - type E <: Classifier[F, E, M] + type E <: Classifier[F, E, M] } /** * param for the base binary classifier that we reduce multiclass classification into. * @group param */ - val classifier: Param[ClassifierType] = - new Param(this, "classifier", "base binary classifier ") + val classifier: Param[ClassifierType] = new Param(this, "classifier", "base binary classifier") /** @group getParam */ def getClassifier: ClassifierType = $(classifier) - } /** @@ -70,10 +68,10 @@ private[ml] trait OneVsRestParams extends PredictorParams { * (taking label 0). */ @AlphaComponent -class OneVsRestModel private[ml] ( - override val parent: OneVsRest, - labelMetadata: Metadata, - val models: Array[_ <: ClassificationModel[_,_]]) +final class OneVsRestModel private[ml] ( + override val uid: String, + labelMetadata: Metadata, + val models: Array[_ <: ClassificationModel[_,_]]) extends Model[OneVsRestModel] with OneVsRestParams { override def transformSchema(schema: StructType): StructType = { @@ -145,11 +143,13 @@ class OneVsRestModel private[ml] ( * is picked to label the example. */ @Experimental -final class OneVsRest extends Estimator[OneVsRestModel] with OneVsRestParams { +final class OneVsRest(override val uid: String) + extends Estimator[OneVsRestModel] with OneVsRestParams { + + def this() = this(Identifiable.randomUID("oneVsRest")) /** @group setParam */ - def setClassifier(value: Classifier[_,_,_]): this.type = { - // TODO: Find a better way to do this. Existential Types don't work with Java API so cast needed + def setClassifier(value: Classifier[_, _, _]): this.type = { set(classifier, value.asInstanceOf[ClassifierType]) } @@ -204,6 +204,7 @@ final class OneVsRest extends Estimator[OneVsRestModel] with OneVsRestParams { NominalAttribute.defaultAttr.withName("label").withNumValues(numClasses) case attr: Attribute => attr } - copyValues(new OneVsRestModel(this, labelAttribute.toMetadata(), models)) + val model = new OneVsRestModel(uid, labelAttribute.toMetadata(), models).setParent(this) + copyValues(model) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/classification/RandomForestClassifier.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/classification/RandomForestClassifier.scala b/mllib/src/main/scala/org/apache/spark/ml/classification/RandomForestClassifier.scala index 9954893..a1de791 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/classification/RandomForestClassifier.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/classification/RandomForestClassifier.scala @@ -23,7 +23,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.{PredictionModel, Predictor} import org.apache.spark.ml.param.ParamMap import org.apache.spark.ml.tree.{RandomForestParams, TreeClassifierParams, DecisionTreeModel, TreeEnsembleModel} -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.tree.{RandomForest => OldRandomForest} @@ -41,10 +41,12 @@ import org.apache.spark.sql.DataFrame * features. */ @AlphaComponent -final class RandomForestClassifier +final class RandomForestClassifier(override val uid: String) extends Predictor[Vector, RandomForestClassifier, RandomForestClassificationModel] with RandomForestParams with TreeClassifierParams { + def this() = this(Identifiable.randomUID("rfc")) + // Override parameter setters from parent trait for Java API compatibility. // Parameters from TreeClassifierParams: @@ -118,7 +120,7 @@ object RandomForestClassifier { */ @AlphaComponent final class RandomForestClassificationModel private[ml] ( - override val parent: RandomForestClassifier, + override val uid: String, private val _trees: Array[DecisionTreeClassificationModel]) extends PredictionModel[Vector, RandomForestClassificationModel] with TreeEnsembleModel with Serializable { @@ -146,7 +148,7 @@ final class RandomForestClassificationModel private[ml] ( } override def copy(extra: ParamMap): RandomForestClassificationModel = { - copyValues(new RandomForestClassificationModel(parent, _trees), extra) + copyValues(new RandomForestClassificationModel(uid, _trees), extra) } override def toString: String = { @@ -172,6 +174,7 @@ private[ml] object RandomForestClassificationModel { // parent, fittingParamMap for each tree is null since there are no good ways to set these. DecisionTreeClassificationModel.fromOld(tree, null, categoricalFeatures) } - new RandomForestClassificationModel(parent, newTrees) + val uid = if (parent != null) parent.uid else Identifiable.randomUID("rfc") + new RandomForestClassificationModel(uid, newTrees) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/evaluation/BinaryClassificationEvaluator.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/evaluation/BinaryClassificationEvaluator.scala b/mllib/src/main/scala/org/apache/spark/ml/evaluation/BinaryClassificationEvaluator.scala index e5a73c6..c1af09c 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/evaluation/BinaryClassificationEvaluator.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/evaluation/BinaryClassificationEvaluator.scala @@ -21,7 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.Evaluator import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ -import org.apache.spark.ml.util.SchemaUtils +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.mllib.evaluation.BinaryClassificationMetrics import org.apache.spark.mllib.linalg.{Vector, VectorUDT} import org.apache.spark.sql.{DataFrame, Row} @@ -33,7 +33,10 @@ import org.apache.spark.sql.types.DoubleType * Evaluator for binary classification, which expects two input columns: score and label. */ @AlphaComponent -class BinaryClassificationEvaluator extends Evaluator with HasRawPredictionCol with HasLabelCol { +class BinaryClassificationEvaluator(override val uid: String) + extends Evaluator with HasRawPredictionCol with HasLabelCol { + + def this() = this(Identifiable.randomUID("binEval")) /** * param for metric name in evaluation http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala index 6eb1db6..62f4a63 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala @@ -22,7 +22,7 @@ import org.apache.spark.ml.Transformer import org.apache.spark.ml.attribute.BinaryAttribute import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared.{HasInputCol, HasOutputCol} -import org.apache.spark.ml.util.SchemaUtils +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.sql._ import org.apache.spark.sql.functions._ import org.apache.spark.sql.types.{DoubleType, StructType} @@ -32,7 +32,10 @@ import org.apache.spark.sql.types.{DoubleType, StructType} * Binarize a column of continuous features given a threshold. */ @AlphaComponent -final class Binarizer extends Transformer with HasInputCol with HasOutputCol { +final class Binarizer(override val uid: String) + extends Transformer with HasInputCol with HasOutputCol { + + def this() = this(Identifiable.randomUID("binarizer")) /** * Param for threshold used to binarize continuous features. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/Bucketizer.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/Bucketizer.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/Bucketizer.scala index d8f1961..ac8dfb5 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/Bucketizer.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/Bucketizer.scala @@ -21,11 +21,11 @@ import java.{util => ju} import org.apache.spark.SparkException import org.apache.spark.annotation.AlphaComponent +import org.apache.spark.ml.Model import org.apache.spark.ml.attribute.NominalAttribute import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared.{HasInputCol, HasOutputCol} -import org.apache.spark.ml.util.SchemaUtils -import org.apache.spark.ml.{Estimator, Model} +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.sql._ import org.apache.spark.sql.functions._ import org.apache.spark.sql.types.{DoubleType, StructField, StructType} @@ -35,10 +35,10 @@ import org.apache.spark.sql.types.{DoubleType, StructField, StructType} * `Bucketizer` maps a column of continuous features to a column of feature buckets. */ @AlphaComponent -final class Bucketizer private[ml] (override val parent: Estimator[Bucketizer]) +final class Bucketizer(override val uid: String) extends Model[Bucketizer] with HasInputCol with HasOutputCol { - def this() = this(null) + def this() = this(Identifiable.randomUID("bucketizer")) /** * Parameter for mapping continuous features into buckets. With n+1 splits, there are n buckets. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/ElementwiseProduct.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/ElementwiseProduct.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/ElementwiseProduct.scala index f8b5629..8b32eee 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/ElementwiseProduct.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/ElementwiseProduct.scala @@ -20,6 +20,7 @@ package org.apache.spark.ml.feature import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.UnaryTransformer import org.apache.spark.ml.param.Param +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.feature import org.apache.spark.mllib.linalg.{Vector, VectorUDT} import org.apache.spark.sql.types.DataType @@ -31,7 +32,10 @@ import org.apache.spark.sql.types.DataType * multiplier. */ @AlphaComponent -class ElementwiseProduct extends UnaryTransformer[Vector, Vector, ElementwiseProduct] { +class ElementwiseProduct(override val uid: String) + extends UnaryTransformer[Vector, Vector, ElementwiseProduct] { + + def this() = this(Identifiable.randomUID("elemProd")) /** * the vector to multiply with input vectors http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/HashingTF.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/HashingTF.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/HashingTF.scala index c305a81..30033ce 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/HashingTF.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/HashingTF.scala @@ -20,6 +20,7 @@ package org.apache.spark.ml.feature import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.UnaryTransformer import org.apache.spark.ml.param.{IntParam, ParamValidators} +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.feature import org.apache.spark.mllib.linalg.{Vector, VectorUDT} import org.apache.spark.sql.types.DataType @@ -29,7 +30,9 @@ import org.apache.spark.sql.types.DataType * Maps a sequence of terms to their term frequencies using the hashing trick. */ @AlphaComponent -class HashingTF extends UnaryTransformer[Iterable[_], Vector, HashingTF] { +class HashingTF(override val uid: String) extends UnaryTransformer[Iterable[_], Vector, HashingTF] { + + def this() = this(Identifiable.randomUID("hashingTF")) /** * Number of features. Should be > 0. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/IDF.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/IDF.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/IDF.scala index d901a20..788c392 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/IDF.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/IDF.scala @@ -21,7 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml._ import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ -import org.apache.spark.ml.util.SchemaUtils +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.mllib.feature import org.apache.spark.mllib.linalg.{Vector, VectorUDT} import org.apache.spark.sql._ @@ -62,7 +62,9 @@ private[feature] trait IDFBase extends Params with HasInputCol with HasOutputCol * Compute the Inverse Document Frequency (IDF) given a collection of documents. */ @AlphaComponent -final class IDF extends Estimator[IDFModel] with IDFBase { +final class IDF(override val uid: String) extends Estimator[IDFModel] with IDFBase { + + def this() = this(Identifiable.randomUID("idf")) /** @group setParam */ def setInputCol(value: String): this.type = set(inputCol, value) @@ -74,7 +76,7 @@ final class IDF extends Estimator[IDFModel] with IDFBase { transformSchema(dataset.schema, logging = true) val input = dataset.select($(inputCol)).map { case Row(v: Vector) => v } val idf = new feature.IDF($(minDocFreq)).fit(input) - copyValues(new IDFModel(this, idf)) + copyValues(new IDFModel(uid, idf).setParent(this)) } override def transformSchema(schema: StructType): StructType = { @@ -88,7 +90,7 @@ final class IDF extends Estimator[IDFModel] with IDFBase { */ @AlphaComponent class IDFModel private[ml] ( - override val parent: IDF, + override val uid: String, idfModel: feature.IDFModel) extends Model[IDFModel] with IDFBase { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/Normalizer.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/Normalizer.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/Normalizer.scala index 755b46a..3f689d1 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/Normalizer.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/Normalizer.scala @@ -20,6 +20,7 @@ package org.apache.spark.ml.feature import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.UnaryTransformer import org.apache.spark.ml.param.{DoubleParam, ParamValidators} +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.feature import org.apache.spark.mllib.linalg.{Vector, VectorUDT} import org.apache.spark.sql.types.DataType @@ -29,7 +30,9 @@ import org.apache.spark.sql.types.DataType * Normalize a vector to have unit norm using the given p-norm. */ @AlphaComponent -class Normalizer extends UnaryTransformer[Vector, Vector, Normalizer] { +class Normalizer(override val uid: String) extends UnaryTransformer[Vector, Vector, Normalizer] { + + def this() = this(Identifiable.randomUID("normalizer")) /** * Normalization in L^p^ space. Must be >= 1. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/OneHotEncoder.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/OneHotEncoder.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/OneHotEncoder.scala index 46514ae..1fb9b9a 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/OneHotEncoder.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/OneHotEncoder.scala @@ -24,7 +24,7 @@ import org.apache.spark.ml.attribute.{Attribute, BinaryAttribute, NominalAttribu import org.apache.spark.mllib.linalg.{Vector, Vectors, VectorUDT} import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared.{HasInputCol, HasOutputCol} -import org.apache.spark.ml.util.SchemaUtils +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.sql.types.{DataType, DoubleType, StructType} /** @@ -37,8 +37,10 @@ import org.apache.spark.sql.types.{DataType, DoubleType, StructType} * linearly dependent because they sum up to one. */ @AlphaComponent -class OneHotEncoder extends UnaryTransformer[Double, Vector, OneHotEncoder] - with HasInputCol with HasOutputCol { +class OneHotEncoder(override val uid: String) + extends UnaryTransformer[Double, Vector, OneHotEncoder] with HasInputCol with HasOutputCol { + + def this() = this(Identifiable.randomUID("oneHot")) /** * Whether to include a component in the encoded vectors for the first category, defaults to true. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/PolynomialExpansion.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/PolynomialExpansion.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/PolynomialExpansion.scala index 9e6177c..4156441 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/PolynomialExpansion.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/PolynomialExpansion.scala @@ -22,6 +22,7 @@ import scala.collection.mutable import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.UnaryTransformer import org.apache.spark.ml.param.{IntParam, ParamValidators} +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.linalg._ import org.apache.spark.sql.types.DataType @@ -34,7 +35,10 @@ import org.apache.spark.sql.types.DataType * `(x, y)`, if we want to expand it with degree 2, then we get `(x, x * x, y, x * y, y * y)`. */ @AlphaComponent -class PolynomialExpansion extends UnaryTransformer[Vector, Vector, PolynomialExpansion] { +class PolynomialExpansion(override val uid: String) + extends UnaryTransformer[Vector, Vector, PolynomialExpansion] { + + def this() = this(Identifiable.randomUID("poly")) /** * The polynomial degree to expand, which should be >= 1. A value of 1 means no expansion. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/StandardScaler.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/StandardScaler.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/StandardScaler.scala index 7cad59f..5ccda15 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/StandardScaler.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/StandardScaler.scala @@ -21,6 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml._ import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.feature import org.apache.spark.mllib.linalg.{Vector, VectorUDT} import org.apache.spark.sql._ @@ -55,7 +56,10 @@ private[feature] trait StandardScalerParams extends Params with HasInputCol with * statistics on the samples in the training set. */ @AlphaComponent -class StandardScaler extends Estimator[StandardScalerModel] with StandardScalerParams { +class StandardScaler(override val uid: String) extends Estimator[StandardScalerModel] + with StandardScalerParams { + + def this() = this(Identifiable.randomUID("stdScal")) setDefault(withMean -> false, withStd -> true) @@ -76,7 +80,7 @@ class StandardScaler extends Estimator[StandardScalerModel] with StandardScalerP val input = dataset.select($(inputCol)).map { case Row(v: Vector) => v } val scaler = new feature.StandardScaler(withMean = $(withMean), withStd = $(withStd)) val scalerModel = scaler.fit(input) - copyValues(new StandardScalerModel(this, scalerModel)) + copyValues(new StandardScalerModel(uid, scalerModel).setParent(this)) } override def transformSchema(schema: StructType): StructType = { @@ -96,7 +100,7 @@ class StandardScaler extends Estimator[StandardScalerModel] with StandardScalerP */ @AlphaComponent class StandardScalerModel private[ml] ( - override val parent: StandardScaler, + override val uid: String, scaler: feature.StandardScalerModel) extends Model[StandardScalerModel] with StandardScalerParams { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/StringIndexer.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/StringIndexer.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/StringIndexer.scala index 3d78537..3f79b67 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/StringIndexer.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/StringIndexer.scala @@ -23,6 +23,7 @@ import org.apache.spark.ml.{Estimator, Model} import org.apache.spark.ml.attribute.NominalAttribute import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.sql.DataFrame import org.apache.spark.sql.functions._ import org.apache.spark.sql.types.{NumericType, StringType, StructType} @@ -58,7 +59,10 @@ private[feature] trait StringIndexerBase extends Params with HasInputCol with Ha * So the most frequent label gets index 0. */ @AlphaComponent -class StringIndexer extends Estimator[StringIndexerModel] with StringIndexerBase { +class StringIndexer(override val uid: String) extends Estimator[StringIndexerModel] + with StringIndexerBase { + + def this() = this(Identifiable.randomUID("strIdx")) /** @group setParam */ def setInputCol(value: String): this.type = set(inputCol, value) @@ -73,7 +77,7 @@ class StringIndexer extends Estimator[StringIndexerModel] with StringIndexerBase .map(_.getString(0)) .countByValue() val labels = counts.toSeq.sortBy(-_._2).map(_._1).toArray - copyValues(new StringIndexerModel(this, labels)) + copyValues(new StringIndexerModel(uid, labels).setParent(this)) } override def transformSchema(schema: StructType): StructType = { @@ -87,7 +91,7 @@ class StringIndexer extends Estimator[StringIndexerModel] with StringIndexerBase */ @AlphaComponent class StringIndexerModel private[ml] ( - override val parent: StringIndexer, + override val uid: String, labels: Array[String]) extends Model[StringIndexerModel] with StringIndexerBase { private val labelToIndex: OpenHashMap[String, Double] = { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/Tokenizer.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/Tokenizer.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/Tokenizer.scala index 649c217..36d9e17 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/Tokenizer.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/Tokenizer.scala @@ -20,6 +20,7 @@ package org.apache.spark.ml.feature import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.UnaryTransformer import org.apache.spark.ml.param._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.sql.types.{ArrayType, DataType, StringType} /** @@ -27,7 +28,9 @@ import org.apache.spark.sql.types.{ArrayType, DataType, StringType} * A tokenizer that converts the input string to lowercase and then splits it by white spaces. */ @AlphaComponent -class Tokenizer extends UnaryTransformer[String, Seq[String], Tokenizer] { +class Tokenizer(override val uid: String) extends UnaryTransformer[String, Seq[String], Tokenizer] { + + def this() = this(Identifiable.randomUID("tok")) override protected def createTransformFunc: String => Seq[String] = { _.toLowerCase.split("\\s") @@ -48,7 +51,10 @@ class Tokenizer extends UnaryTransformer[String, Seq[String], Tokenizer] { * It returns an array of strings that can be empty. */ @AlphaComponent -class RegexTokenizer extends UnaryTransformer[String, Seq[String], RegexTokenizer] { +class RegexTokenizer(override val uid: String) + extends UnaryTransformer[String, Seq[String], RegexTokenizer] { + + def this() = this(Identifiable.randomUID("regexTok")) /** * Minimum token length, >= 0. http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/VectorAssembler.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/VectorAssembler.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/VectorAssembler.scala index 796758a..1c00094 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/VectorAssembler.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/VectorAssembler.scala @@ -23,6 +23,7 @@ import org.apache.spark.SparkException import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.Transformer import org.apache.spark.ml.param.shared._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.linalg.{Vector, VectorUDT, Vectors} import org.apache.spark.sql.{DataFrame, Row} import org.apache.spark.sql.functions._ @@ -33,7 +34,10 @@ import org.apache.spark.sql.types._ * A feature transformer that merges multiple columns into a vector column. */ @AlphaComponent -class VectorAssembler extends Transformer with HasInputCols with HasOutputCol { +class VectorAssembler(override val uid: String) + extends Transformer with HasInputCols with HasOutputCol { + + def this() = this(Identifiable.randomUID("va")) /** @group setParam */ def setInputCols(value: Array[String]): this.type = set(inputCols, value) http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/VectorIndexer.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/VectorIndexer.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/VectorIndexer.scala index 0f83a29..6d1d052 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/VectorIndexer.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/VectorIndexer.scala @@ -22,7 +22,7 @@ import org.apache.spark.ml.{Estimator, Model} import org.apache.spark.ml.attribute._ import org.apache.spark.ml.param.{IntParam, ParamValidators, Params} import org.apache.spark.ml.param.shared._ -import org.apache.spark.ml.util.SchemaUtils +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.mllib.linalg.{DenseVector, SparseVector, Vector, VectorUDT} import org.apache.spark.sql.{DataFrame, Row} import org.apache.spark.sql.functions.callUDF @@ -87,7 +87,10 @@ private[ml] trait VectorIndexerParams extends Params with HasInputCol with HasOu * - Add option for allowing unknown categories. */ @AlphaComponent -class VectorIndexer extends Estimator[VectorIndexerModel] with VectorIndexerParams { +class VectorIndexer(override val uid: String) extends Estimator[VectorIndexerModel] + with VectorIndexerParams { + + def this() = this(Identifiable.randomUID("vecIdx")) /** @group setParam */ def setMaxCategories(value: Int): this.type = set(maxCategories, value) @@ -110,7 +113,9 @@ class VectorIndexer extends Estimator[VectorIndexerModel] with VectorIndexerPara iter.foreach(localCatStats.addVector) Iterator(localCatStats) }.reduce((stats1, stats2) => stats1.merge(stats2)) - copyValues(new VectorIndexerModel(this, numFeatures, categoryStats.getCategoryMaps)) + val model = new VectorIndexerModel(uid, numFeatures, categoryStats.getCategoryMaps) + .setParent(this) + copyValues(model) } override def transformSchema(schema: StructType): StructType = { @@ -238,7 +243,7 @@ private object VectorIndexer { */ @AlphaComponent class VectorIndexerModel private[ml] ( - override val parent: VectorIndexer, + override val uid: String, val numFeatures: Int, val categoryMaps: Map[Int, Map[Double, Int]]) extends Model[VectorIndexerModel] with VectorIndexerParams { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/feature/Word2Vec.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/feature/Word2Vec.scala b/mllib/src/main/scala/org/apache/spark/ml/feature/Word2Vec.scala index 34ff929..8ace8c5 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/feature/Word2Vec.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/feature/Word2Vec.scala @@ -21,7 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.{Estimator, Model} import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ -import org.apache.spark.ml.util.SchemaUtils +import org.apache.spark.ml.util.{Identifiable, SchemaUtils} import org.apache.spark.mllib.feature import org.apache.spark.mllib.linalg.{VectorUDT, Vectors} import org.apache.spark.mllib.linalg.BLAS._ @@ -85,7 +85,9 @@ private[feature] trait Word2VecBase extends Params * natural language processing or machine learning process. */ @AlphaComponent -final class Word2Vec extends Estimator[Word2VecModel] with Word2VecBase { +final class Word2Vec(override val uid: String) extends Estimator[Word2VecModel] with Word2VecBase { + + def this() = this(Identifiable.randomUID("w2v")) /** @group setParam */ def setInputCol(value: String): this.type = set(inputCol, value) @@ -122,7 +124,7 @@ final class Word2Vec extends Estimator[Word2VecModel] with Word2VecBase { .setSeed($(seed)) .setVectorSize($(vectorSize)) .fit(input) - copyValues(new Word2VecModel(this, wordVectors)) + copyValues(new Word2VecModel(uid, wordVectors).setParent(this)) } override def transformSchema(schema: StructType): StructType = { @@ -136,7 +138,7 @@ final class Word2Vec extends Estimator[Word2VecModel] with Word2VecBase { */ @AlphaComponent class Word2VecModel private[ml] ( - override val parent: Word2Vec, + override val uid: String, wordVectors: feature.Word2VecModel) extends Model[Word2VecModel] with Word2VecBase { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/param/params.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/param/params.scala b/mllib/src/main/scala/org/apache/spark/ml/param/params.scala index 5a7ec29..247e08b 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/param/params.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/param/params.scala @@ -40,12 +40,17 @@ import org.apache.spark.ml.util.Identifiable * @tparam T param value type */ @AlphaComponent -class Param[T] (val parent: Params, val name: String, val doc: String, val isValid: T => Boolean) +class Param[T](val parent: String, val name: String, val doc: String, val isValid: T => Boolean) extends Serializable { - def this(parent: Params, name: String, doc: String) = + def this(parent: Identifiable, name: String, doc: String, isValid: T => Boolean) = + this(parent.uid, name, doc, isValid) + + def this(parent: String, name: String, doc: String) = this(parent, name, doc, ParamValidators.alwaysTrue[T]) + def this(parent: Identifiable, name: String, doc: String) = this(parent.uid, name, doc) + /** * Assert that the given value is valid for this parameter. * @@ -60,8 +65,7 @@ class Param[T] (val parent: Params, val name: String, val doc: String, val isVal */ private[param] def validate(value: T): Unit = { if (!isValid(value)) { - throw new IllegalArgumentException(s"$parent parameter $name given invalid value $value." + - s" Parameter description: $toString") + throw new IllegalArgumentException(s"$parent parameter $name given invalid value $value.") } } @@ -75,19 +79,15 @@ class Param[T] (val parent: Params, val name: String, val doc: String, val isVal */ def ->(value: T): ParamPair[T] = ParamPair(this, value) - /** - * Converts this param's name, doc, and optionally its default value and the user-supplied - * value in its parent to string. - */ - override def toString: String = { - val valueStr = if (parent.isDefined(this)) { - val defaultValueStr = parent.getDefault(this).map("default: " + _) - val currentValueStr = parent.get(this).map("current: " + _) - (defaultValueStr ++ currentValueStr).mkString("(", ", ", ")") - } else { - "(undefined)" + override final def toString: String = s"${parent}__$name" + + override final def hashCode: Int = toString.## + + override final def equals(obj: Any): Boolean = { + obj match { + case p: Param[_] => (p.parent == parent) && (p.name == name) + case _ => false } - s"$name: $doc $valueStr" } } @@ -173,49 +173,71 @@ object ParamValidators { // specialize primitive-typed params because Java doesn't recognize scala.Double, scala.Int, ... /** Specialized version of [[Param[Double]]] for Java. */ -class DoubleParam(parent: Params, name: String, doc: String, isValid: Double => Boolean) +class DoubleParam(parent: String, name: String, doc: String, isValid: Double => Boolean) extends Param[Double](parent, name, doc, isValid) { - def this(parent: Params, name: String, doc: String) = + def this(parent: String, name: String, doc: String) = this(parent, name, doc, ParamValidators.alwaysTrue) + def this(parent: Identifiable, name: String, doc: String, isValid: Double => Boolean) = + this(parent.uid, name, doc, isValid) + + def this(parent: Identifiable, name: String, doc: String) = this(parent.uid, name, doc) + override def w(value: Double): ParamPair[Double] = super.w(value) } /** Specialized version of [[Param[Int]]] for Java. */ -class IntParam(parent: Params, name: String, doc: String, isValid: Int => Boolean) +class IntParam(parent: String, name: String, doc: String, isValid: Int => Boolean) extends Param[Int](parent, name, doc, isValid) { - def this(parent: Params, name: String, doc: String) = + def this(parent: String, name: String, doc: String) = this(parent, name, doc, ParamValidators.alwaysTrue) + def this(parent: Identifiable, name: String, doc: String, isValid: Int => Boolean) = + this(parent.uid, name, doc, isValid) + + def this(parent: Identifiable, name: String, doc: String) = this(parent.uid, name, doc) + override def w(value: Int): ParamPair[Int] = super.w(value) } /** Specialized version of [[Param[Float]]] for Java. */ -class FloatParam(parent: Params, name: String, doc: String, isValid: Float => Boolean) +class FloatParam(parent: String, name: String, doc: String, isValid: Float => Boolean) extends Param[Float](parent, name, doc, isValid) { - def this(parent: Params, name: String, doc: String) = + def this(parent: String, name: String, doc: String) = this(parent, name, doc, ParamValidators.alwaysTrue) + def this(parent: Identifiable, name: String, doc: String, isValid: Float => Boolean) = + this(parent.uid, name, doc, isValid) + + def this(parent: Identifiable, name: String, doc: String) = this(parent.uid, name, doc) + override def w(value: Float): ParamPair[Float] = super.w(value) } /** Specialized version of [[Param[Long]]] for Java. */ -class LongParam(parent: Params, name: String, doc: String, isValid: Long => Boolean) +class LongParam(parent: String, name: String, doc: String, isValid: Long => Boolean) extends Param[Long](parent, name, doc, isValid) { - def this(parent: Params, name: String, doc: String) = + def this(parent: String, name: String, doc: String) = this(parent, name, doc, ParamValidators.alwaysTrue) + def this(parent: Identifiable, name: String, doc: String, isValid: Long => Boolean) = + this(parent.uid, name, doc, isValid) + + def this(parent: Identifiable, name: String, doc: String) = this(parent.uid, name, doc) + override def w(value: Long): ParamPair[Long] = super.w(value) } /** Specialized version of [[Param[Boolean]]] for Java. */ -class BooleanParam(parent: Params, name: String, doc: String) // No need for isValid +class BooleanParam(parent: String, name: String, doc: String) // No need for isValid extends Param[Boolean](parent, name, doc) { + def this(parent: Identifiable, name: String, doc: String) = this(parent.uid, name, doc) + override def w(value: Boolean): ParamPair[Boolean] = super.w(value) } @@ -265,6 +287,9 @@ trait Params extends Identifiable with Serializable { /** * Returns all params sorted by their names. The default implementation uses Java reflection to * list all public methods that have no arguments and return [[Param]]. + * + * Note: Developer should not use this method in constructor because we cannot guarantee that + * this variable gets initialized before other params. */ lazy val params: Array[Param[_]] = { val methods = this.getClass.getMethods @@ -299,15 +324,36 @@ trait Params extends Identifiable with Serializable { * those are checked during schema validation. */ def validateParams(): Unit = { - params.filter(isDefined _).foreach { param => + params.filter(isDefined).foreach { param => param.asInstanceOf[Param[Any]].validate($(param)) } } /** - * Returns the documentation of all params. + * Explains a param. + * @param param input param, must belong to this instance. + * @return a string that contains the input param name, doc, and optionally its default value and + * the user-supplied value + */ + def explainParam(param: Param[_]): String = { + shouldOwn(param) + val valueStr = if (isDefined(param)) { + val defaultValueStr = getDefault(param).map("default: " + _) + val currentValueStr = get(param).map("current: " + _) + (defaultValueStr ++ currentValueStr).mkString("(", ", ", ")") + } else { + "(undefined)" + } + s"${param.name}: ${param.doc} $valueStr" + } + + /** + * Explains all params of this instance. + * @see [[explainParam()]] */ - def explainParams(): String = params.mkString("\n") + def explainParams(): String = { + params.map(explainParam).mkString("\n") + } /** Checks whether a param is explicitly set. */ final def isSet(param: Param[_]): Boolean = { @@ -392,7 +438,6 @@ trait Params extends Identifiable with Serializable { * @param value the default value */ protected final def setDefault[T](param: Param[T], value: T): this.type = { - shouldOwn(param) defaultParamMap.put(param, value) this } @@ -430,13 +475,13 @@ trait Params extends Identifiable with Serializable { } /** - * Creates a copy of this instance with a randomly generated uid and some extra params. - * The default implementation calls the default constructor to create a new instance, then - * copies the embedded and extra parameters over and returns the new instance. + * Creates a copy of this instance with the same UID and some extra params. + * The default implementation tries to create a new instance with the same UID. + * Then it copies the embedded and extra parameters over and returns the new instance. * Subclasses should override this method if the default approach is not sufficient. */ def copy(extra: ParamMap): Params = { - val that = this.getClass.newInstance() + val that = this.getClass.getConstructor(classOf[String]).newInstance(uid) copyValues(that, extra) that } @@ -465,7 +510,7 @@ trait Params extends Identifiable with Serializable { /** Validates that the input param belongs to this instance. */ private def shouldOwn(param: Param[_]): Unit = { - require(param.parent.eq(this), s"Param $param does not belong to $this.") + require(param.parent == uid && hasParam(param.name), s"Param $param does not belong to $this.") } /** @@ -581,7 +626,7 @@ final class ParamMap private[ml] (private val map: mutable.Map[Param[Any], Any]) override def toString: String = { map.toSeq.sortBy(_._1.name).map { case (param, value) => - s"\t${param.parent.uid}-${param.name}: $value" + s"\t${param.parent}-${param.name}: $value" }.mkString("{\n", ",\n", "\n}") } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala b/mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala index d7cbffc..45c57b5 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala @@ -35,6 +35,7 @@ import org.apache.spark.annotation.DeveloperApi import org.apache.spark.ml.{Estimator, Model} import org.apache.spark.ml.param._ import org.apache.spark.ml.param.shared._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.optimization.NNLS import org.apache.spark.rdd.RDD import org.apache.spark.sql.DataFrame @@ -171,7 +172,7 @@ private[recommendation] trait ALSParams extends Params with HasMaxIter with HasR * Model fitted by ALS. */ class ALSModel private[ml] ( - override val parent: ALS, + override val uid: String, k: Int, userFactors: RDD[(Int, Array[Float])], itemFactors: RDD[(Int, Array[Float])]) @@ -235,10 +236,12 @@ class ALSModel private[ml] ( * indicated user * preferences rather than explicit ratings given to items. */ -class ALS extends Estimator[ALSModel] with ALSParams { +class ALS(override val uid: String) extends Estimator[ALSModel] with ALSParams { import org.apache.spark.ml.recommendation.ALS.Rating + def this() = this(Identifiable.randomUID("als")) + /** @group setParam */ def setRank(value: Int): this.type = set(rank, value) @@ -303,7 +306,8 @@ class ALS extends Estimator[ALSModel] with ALSParams { maxIter = $(maxIter), regParam = $(regParam), implicitPrefs = $(implicitPrefs), alpha = $(alpha), nonnegative = $(nonnegative), checkpointInterval = $(checkpointInterval), seed = $(seed)) - copyValues(new ALSModel(this, $(rank), userFactors, itemFactors)) + val model = new ALSModel(uid, $(rank), userFactors, itemFactors).setParent(this) + copyValues(model) } override def transformSchema(schema: StructType): StructType = { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/regression/DecisionTreeRegressor.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/regression/DecisionTreeRegressor.scala b/mllib/src/main/scala/org/apache/spark/ml/regression/DecisionTreeRegressor.scala index f8f0b16..e67df21 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/regression/DecisionTreeRegressor.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/regression/DecisionTreeRegressor.scala @@ -21,7 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.{PredictionModel, Predictor} import org.apache.spark.ml.param.ParamMap import org.apache.spark.ml.tree.{TreeRegressorParams, DecisionTreeParams, DecisionTreeModel, Node} -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.tree.{DecisionTree => OldDecisionTree} @@ -38,10 +38,12 @@ import org.apache.spark.sql.DataFrame * It supports both continuous and categorical features. */ @AlphaComponent -final class DecisionTreeRegressor +final class DecisionTreeRegressor(override val uid: String) extends Predictor[Vector, DecisionTreeRegressor, DecisionTreeRegressionModel] with DecisionTreeParams with TreeRegressorParams { + def this() = this(Identifiable.randomUID("dtr")) + // Override parameter setters from parent trait for Java API compatibility. override def setMaxDepth(value: Int): this.type = super.setMaxDepth(value) @@ -91,7 +93,7 @@ object DecisionTreeRegressor { */ @AlphaComponent final class DecisionTreeRegressionModel private[ml] ( - override val parent: DecisionTreeRegressor, + override val uid: String, override val rootNode: Node) extends PredictionModel[Vector, DecisionTreeRegressionModel] with DecisionTreeModel with Serializable { @@ -104,7 +106,7 @@ final class DecisionTreeRegressionModel private[ml] ( } override def copy(extra: ParamMap): DecisionTreeRegressionModel = { - copyValues(new DecisionTreeRegressionModel(parent, rootNode), extra) + copyValues(new DecisionTreeRegressionModel(uid, rootNode), extra) } override def toString: String = { @@ -128,6 +130,7 @@ private[ml] object DecisionTreeRegressionModel { s"Cannot convert non-regression DecisionTreeModel (old API) to" + s" DecisionTreeRegressionModel (new API). Algo is: ${oldModel.algo}") val rootNode = Node.fromOld(oldModel.topNode, categoricalFeatures) - new DecisionTreeRegressionModel(parent, rootNode) + val uid = if (parent != null) parent.uid else Identifiable.randomUID("dtr") + new DecisionTreeRegressionModel(uid, rootNode) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/regression/GBTRegressor.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/regression/GBTRegressor.scala b/mllib/src/main/scala/org/apache/spark/ml/regression/GBTRegressor.scala index 461905c..4249ff5 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/regression/GBTRegressor.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/regression/GBTRegressor.scala @@ -24,7 +24,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.{PredictionModel, Predictor} import org.apache.spark.ml.param.{Param, ParamMap} import org.apache.spark.ml.tree.{GBTParams, TreeRegressorParams, DecisionTreeModel, TreeEnsembleModel} -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.tree.{GradientBoostedTrees => OldGBT} @@ -42,10 +42,12 @@ import org.apache.spark.sql.DataFrame * It supports both continuous and categorical features. */ @AlphaComponent -final class GBTRegressor +final class GBTRegressor(override val uid: String) extends Predictor[Vector, GBTRegressor, GBTRegressionModel] with GBTParams with TreeRegressorParams with Logging { + def this() = this(Identifiable.randomUID("gbtr")) + // Override parameter setters from parent trait for Java API compatibility. // Parameters from TreeRegressorParams: @@ -149,7 +151,7 @@ object GBTRegressor { */ @AlphaComponent final class GBTRegressionModel( - override val parent: GBTRegressor, + override val uid: String, private val _trees: Array[DecisionTreeRegressionModel], private val _treeWeights: Array[Double]) extends PredictionModel[Vector, GBTRegressionModel] @@ -173,7 +175,7 @@ final class GBTRegressionModel( } override def copy(extra: ParamMap): GBTRegressionModel = { - copyValues(new GBTRegressionModel(parent, _trees, _treeWeights), extra) + copyValues(new GBTRegressionModel(uid, _trees, _treeWeights), extra) } override def toString: String = { @@ -199,6 +201,7 @@ private[ml] object GBTRegressionModel { // parent, fittingParamMap for each tree is null since there are no good ways to set these. DecisionTreeRegressionModel.fromOld(tree, null, categoricalFeatures) } - new GBTRegressionModel(parent, newTrees, oldModel.treeWeights) + val uid = if (parent != null) parent.uid else Identifiable.randomUID("gbtr") + new GBTRegressionModel(parent.uid, newTrees, oldModel.treeWeights) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala b/mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala index 36c242b..3ebb78f 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala @@ -20,14 +20,14 @@ package org.apache.spark.ml.regression import scala.collection.mutable import breeze.linalg.{DenseVector => BDV, norm => brzNorm} -import breeze.optimize.{CachedDiffFunction, DiffFunction, LBFGS => BreezeLBFGS, - OWLQN => BreezeOWLQN} +import breeze.optimize.{CachedDiffFunction, DiffFunction, LBFGS => BreezeLBFGS, OWLQN => BreezeOWLQN} import org.apache.spark.Logging import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.PredictorParams import org.apache.spark.ml.param.ParamMap import org.apache.spark.ml.param.shared.{HasElasticNetParam, HasMaxIter, HasRegParam, HasTol} +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.linalg.{Vector, Vectors} import org.apache.spark.mllib.linalg.BLAS._ import org.apache.spark.mllib.regression.LabeledPoint @@ -59,9 +59,12 @@ private[regression] trait LinearRegressionParams extends PredictorParams * - L2 + L1 (elastic net) */ @AlphaComponent -class LinearRegression extends Regressor[Vector, LinearRegression, LinearRegressionModel] +class LinearRegression(override val uid: String) + extends Regressor[Vector, LinearRegression, LinearRegressionModel] with LinearRegressionParams with Logging { + def this() = this(Identifiable.randomUID("linReg")) + /** * Set the regularization parameter. * Default is 0.0. @@ -128,7 +131,7 @@ class LinearRegression extends Regressor[Vector, LinearRegression, LinearRegress logWarning(s"The standard deviation of the label is zero, so the weights will be zeros " + s"and the intercept will be the mean of the label; as a result, training is not needed.") if (handlePersistence) instances.unpersist() - return new LinearRegressionModel(this, Vectors.sparse(numFeatures, Seq()), yMean) + return new LinearRegressionModel(uid, Vectors.sparse(numFeatures, Seq()), yMean) } val featuresMean = summarizer.mean.toArray @@ -182,7 +185,7 @@ class LinearRegression extends Regressor[Vector, LinearRegression, LinearRegress if (handlePersistence) instances.unpersist() // TODO: Converts to sparse format based on the storage, but may base on the scoring speed. - new LinearRegressionModel(this, weights.compressed, intercept) + copyValues(new LinearRegressionModel(uid, weights.compressed, intercept)) } } @@ -193,7 +196,7 @@ class LinearRegression extends Regressor[Vector, LinearRegression, LinearRegress */ @AlphaComponent class LinearRegressionModel private[ml] ( - override val parent: LinearRegression, + override val uid: String, val weights: Vector, val intercept: Double) extends RegressionModel[Vector, LinearRegressionModel] @@ -204,7 +207,7 @@ class LinearRegressionModel private[ml] ( } override def copy(extra: ParamMap): LinearRegressionModel = { - copyValues(new LinearRegressionModel(parent, weights, intercept), extra) + copyValues(new LinearRegressionModel(uid, weights, intercept), extra) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/regression/RandomForestRegressor.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/regression/RandomForestRegressor.scala b/mllib/src/main/scala/org/apache/spark/ml/regression/RandomForestRegressor.scala index dbc6289..82437aa 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/regression/RandomForestRegressor.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/regression/RandomForestRegressor.scala @@ -21,7 +21,7 @@ import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml.{PredictionModel, Predictor} import org.apache.spark.ml.param.ParamMap import org.apache.spark.ml.tree.{RandomForestParams, TreeRegressorParams, DecisionTreeModel, TreeEnsembleModel} -import org.apache.spark.ml.util.MetadataUtils +import org.apache.spark.ml.util.{Identifiable, MetadataUtils} import org.apache.spark.mllib.linalg.Vector import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.tree.{RandomForest => OldRandomForest} @@ -37,10 +37,12 @@ import org.apache.spark.sql.DataFrame * It supports both continuous and categorical features. */ @AlphaComponent -final class RandomForestRegressor +final class RandomForestRegressor(override val uid: String) extends Predictor[Vector, RandomForestRegressor, RandomForestRegressionModel] with RandomForestParams with TreeRegressorParams { + def this() = this(Identifiable.randomUID("rfr")) + // Override parameter setters from parent trait for Java API compatibility. // Parameters from TreeRegressorParams: @@ -105,7 +107,7 @@ object RandomForestRegressor { */ @AlphaComponent final class RandomForestRegressionModel private[ml] ( - override val parent: RandomForestRegressor, + override val uid: String, private val _trees: Array[DecisionTreeRegressionModel]) extends PredictionModel[Vector, RandomForestRegressionModel] with TreeEnsembleModel with Serializable { @@ -128,7 +130,7 @@ final class RandomForestRegressionModel private[ml] ( } override def copy(extra: ParamMap): RandomForestRegressionModel = { - copyValues(new RandomForestRegressionModel(parent, _trees), extra) + copyValues(new RandomForestRegressionModel(uid, _trees), extra) } override def toString: String = { @@ -154,6 +156,6 @@ private[ml] object RandomForestRegressionModel { // parent, fittingParamMap for each tree is null since there are no good ways to set these. DecisionTreeRegressionModel.fromOld(tree, null, categoricalFeatures) } - new RandomForestRegressionModel(parent, newTrees) + new RandomForestRegressionModel(parent.uid, newTrees) } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/tuning/CrossValidator.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/tuning/CrossValidator.scala b/mllib/src/main/scala/org/apache/spark/ml/tuning/CrossValidator.scala index ac0d1fe..5c6ff2d 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/tuning/CrossValidator.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/tuning/CrossValidator.scala @@ -23,6 +23,7 @@ import org.apache.spark.Logging import org.apache.spark.annotation.AlphaComponent import org.apache.spark.ml._ import org.apache.spark.ml.param._ +import org.apache.spark.ml.util.Identifiable import org.apache.spark.mllib.util.MLUtils import org.apache.spark.sql.DataFrame import org.apache.spark.sql.types.StructType @@ -81,7 +82,10 @@ private[ml] trait CrossValidatorParams extends Params { * K-fold cross validation. */ @AlphaComponent -class CrossValidator extends Estimator[CrossValidatorModel] with CrossValidatorParams with Logging { +class CrossValidator(override val uid: String) extends Estimator[CrossValidatorModel] + with CrossValidatorParams with Logging { + + def this() = this(Identifiable.randomUID("cv")) private val f2jBLAS = new F2jBLAS @@ -136,7 +140,7 @@ class CrossValidator extends Estimator[CrossValidatorModel] with CrossValidatorP logInfo(s"Best set of parameters:\n${epm(bestIndex)}") logInfo(s"Best cross-validation metric: $bestMetric.") val bestModel = est.fit(dataset, epm(bestIndex)).asInstanceOf[Model[_]] - copyValues(new CrossValidatorModel(this, bestModel)) + copyValues(new CrossValidatorModel(uid, bestModel).setParent(this)) } override def transformSchema(schema: StructType): StructType = { @@ -150,7 +154,7 @@ class CrossValidator extends Estimator[CrossValidatorModel] with CrossValidatorP */ @AlphaComponent class CrossValidatorModel private[ml] ( - override val parent: CrossValidator, + override val uid: String, val bestModel: Model[_]) extends Model[CrossValidatorModel] with CrossValidatorParams { http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/main/scala/org/apache/spark/ml/util/Identifiable.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/util/Identifiable.scala b/mllib/src/main/scala/org/apache/spark/ml/util/Identifiable.scala index 8a56748..1466976 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/util/Identifiable.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/util/Identifiable.scala @@ -19,15 +19,24 @@ package org.apache.spark.ml.util import java.util.UUID + /** - * Object with a unique id. + * Trait for an object with an immutable unique ID that identifies itself and its derivatives. */ -private[ml] trait Identifiable extends Serializable { +trait Identifiable { + + /** + * An immutable unique ID for the object and its derivatives. + */ + val uid: String +} + +object Identifiable { /** - * A unique id for the object. The default implementation concatenates the class name, "_", and 8 - * random hex chars. + * Returns a random UID that concatenates the given prefix, "_", and 12 random hex chars. */ - private[ml] val uid: String = - this.getClass.getSimpleName + "_" + UUID.randomUUID().toString.take(8) + def randomUID(prefix: String): String = { + prefix + "_" + UUID.randomUUID().toString.takeRight(12) + } } http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/test/java/org/apache/spark/ml/classification/JavaLogisticRegressionSuite.java ---------------------------------------------------------------------- diff --git a/mllib/src/test/java/org/apache/spark/ml/classification/JavaLogisticRegressionSuite.java b/mllib/src/test/java/org/apache/spark/ml/classification/JavaLogisticRegressionSuite.java index 7e7189a..f75e024 100644 --- a/mllib/src/test/java/org/apache/spark/ml/classification/JavaLogisticRegressionSuite.java +++ b/mllib/src/test/java/org/apache/spark/ml/classification/JavaLogisticRegressionSuite.java @@ -84,7 +84,7 @@ public class JavaLogisticRegressionSuite implements Serializable { .setThreshold(0.6) .setProbabilityCol("myProbability"); LogisticRegressionModel model = lr.fit(dataset); - LogisticRegression parent = model.parent(); + LogisticRegression parent = (LogisticRegression) model.parent(); assert(parent.getMaxIter() == 10); assert(parent.getRegParam() == 1.0); assert(parent.getThreshold() == 0.6); @@ -110,7 +110,7 @@ public class JavaLogisticRegressionSuite implements Serializable { // Call fit() with new params, and check as many params as we can. LogisticRegressionModel model2 = lr.fit(dataset, lr.maxIter().w(5), lr.regParam().w(0.1), lr.threshold().w(0.4), lr.probabilityCol().w("theProb")); - LogisticRegression parent2 = model2.parent(); + LogisticRegression parent2 = (LogisticRegression) model2.parent(); assert(parent2.getMaxIter() == 5); assert(parent2.getRegParam() == 0.1); assert(parent2.getThreshold() == 0.4); http://git-wip-us.apache.org/repos/asf/spark/blob/e45cd9f7/mllib/src/test/java/org/apache/spark/ml/param/JavaTestParams.java ---------------------------------------------------------------------- diff --git a/mllib/src/test/java/org/apache/spark/ml/param/JavaTestParams.java b/mllib/src/test/java/org/apache/spark/ml/param/JavaTestParams.java index 8abe575..3a41890 100644 --- a/mllib/src/test/java/org/apache/spark/ml/param/JavaTestParams.java +++ b/mllib/src/test/java/org/apache/spark/ml/param/JavaTestParams.java @@ -21,43 +21,65 @@ import java.util.List; import com.google.common.collect.Lists; +import org.apache.spark.ml.util.Identifiable$; + /** * A subclass of Params for testing. */ public class JavaTestParams extends JavaParams { - public IntParam myIntParam; + public JavaTestParams() { + this.uid_ = Identifiable$.MODULE$.randomUID("javaTestParams"); + init(); + } + + public JavaTestParams(String uid) { + this.uid_ = uid; + init(); + } + + private String uid_; + + @Override + public String uid() { + return uid_; + } - public int getMyIntParam() { return (Integer)getOrDefault(myIntParam); } + private IntParam myIntParam_; + public IntParam myIntParam() { return myIntParam_; } + + public int getMyIntParam() { return (Integer)getOrDefault(myIntParam_); } public JavaTestParams setMyIntParam(int value) { - set(myIntParam, value); return this; + set(myIntParam_, value); return this; } - public DoubleParam myDoubleParam; + private DoubleParam myDoubleParam_; + public DoubleParam myDoubleParam() { return myDoubleParam_; } - public double getMyDoubleParam() { return (Double)getOrDefault(myDoubleParam); } + public double getMyDoubleParam() { return (Double)getOrDefault(myDoubleParam_); } public JavaTestParams setMyDoubleParam(double value) { - set(myDoubleParam, value); return this; + set(myDoubleParam_, value); return this; } - public Param<String> myStringParam; + private Param<String> myStringParam_; + public Param<String> myStringParam() { return myStringParam_; } - public String getMyStringParam() { return (String)getOrDefault(myStringParam); } + public String getMyStringParam() { return getOrDefault(myStringParam_); } public JavaTestParams setMyStringParam(String value) { - set(myStringParam, value); return this; + set(myStringParam_, value); return this; } - public JavaTestParams() { - myIntParam = new IntParam(this, "myIntParam", "this is an int param", ParamValidators.gt(0)); - myDoubleParam = new DoubleParam(this, "myDoubleParam", "this is a double param", + private void init() { + myIntParam_ = new IntParam(this, "myIntParam", "this is an int param", ParamValidators.gt(0)); + myDoubleParam_ = new DoubleParam(this, "myDoubleParam", "this is a double param", ParamValidators.inRange(0.0, 1.0)); List<String> validStrings = Lists.newArrayList("a", "b"); - myStringParam = new Param<String>(this, "myStringParam", "this is a string param", + myStringParam_ = new Param<String>(this, "myStringParam", "this is a string param", ParamValidators.inArray(validStrings)); - setDefault(myIntParam, 1); - setDefault(myDoubleParam, 0.5); + setDefault(myIntParam_, 1); + setDefault(myDoubleParam_, 0.5); } } --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
