snapshot

mllib/src/main/scala/org/apache/spark/ml/Estimator.scala

@@ -78,7 +78,5 @@ abstract class Estimator[M <: Model[M]] extends PipelineStage {
     paramMaps.map(fit(dataset, _))
   }
 
-  override def copy(extra: ParamMap): Estimator[M] = {
-    super.copy(extra).asInstanceOf[Estimator[M]]
-  }
+  override def copy(extra: ParamMap): Estimator[M]
 }

mllib/src/main/scala/org/apache/spark/ml/Model.scala

@@ -45,8 +45,5 @@ abstract class Model[M <: Model[M]] extends Transformer {
   /** Indicates whether this [[Model]] has a corresponding parent. */
   def hasParent: Boolean = parent != null
 
-  override def copy(extra: ParamMap): M = {
-    // The default implementation of Params.copy doesn't work for models.
-    throw new NotImplementedError(s"${this.getClass} doesn't implement copy(extra: ParamMap)")
-  }
+  override def copy(extra: ParamMap): M
 }

mllib/src/main/scala/org/apache/spark/ml/Pipeline.scala

@@ -63,9 +63,7 @@ abstract class PipelineStage extends Params with Logging {
     outputSchema
   }
 
-  override def copy(extra: ParamMap): PipelineStage = {
-    super.copy(extra).asInstanceOf[PipelineStage]
-  }
+  override def copy(extra: ParamMap): PipelineStage
 }
 
 /**

mllib/src/main/scala/org/apache/spark/ml/Predictor.scala

@@ -90,9 +90,7 @@ abstract class Predictor[
     copyValues(train(dataset).setParent(this))
   }
 
-  override def copy(extra: ParamMap): Learner = {
-    super.copy(extra).asInstanceOf[Learner]
-  }
+  override def copy(extra: ParamMap): Learner
 
   /**
    * Train a model using the given dataset and parameters.

mllib/src/main/scala/org/apache/spark/ml/Transformer.scala

@@ -67,9 +67,7 @@ abstract class Transformer extends PipelineStage {
    */
   def transform(dataset: DataFrame): DataFrame
 
-  override def copy(extra: ParamMap): Transformer = {
-    super.copy(extra).asInstanceOf[Transformer]
-  }
+  override def copy(extra: ParamMap): Transformer
 }
 
 /**
@@ -120,4 +118,6 @@ abstract class UnaryTransformer[IN, OUT, T <: UnaryTransformer[IN, OUT, T]]
     dataset.withColumn($(outputCol),
       callUDF(this.createTransformFunc, outputDataType, dataset($(inputCol))))
   }
+
+  override def copy(extra: ParamMap): T = defaultCopyWithParams(extra)
 }

mllib/src/main/scala/org/apache/spark/ml/classification/Classifier.scala

@@ -18,6 +18,7 @@
 package org.apache.spark.ml.classification
 
 import org.apache.spark.annotation.DeveloperApi
+import org.apache.spark.ml.param.ParamMap
 import org.apache.spark.ml.{PredictionModel, PredictorParams, Predictor}
 import org.apache.spark.ml.param.shared.HasRawPredictionCol
 import org.apache.spark.ml.util.SchemaUtils

mllib/src/main/scala/org/apache/spark/ml/classification/DecisionTreeClassifier.scala

@@ -86,6 +86,8 @@ final class DecisionTreeClassifier(override val uid: String)
     super.getOldStrategy(categoricalFeatures, numClasses, OldAlgo.Classification, getOldImpurity,
       subsamplingRate = 1.0)
   }
+
+  override def copy(extra: ParamMap): DecisionTreeClassifier = defaultCopyWithParams(extra)
 }
 
 @Experimental

mllib/src/main/scala/org/apache/spark/ml/classification/GBTClassifier.scala

@@ -141,6 +141,8 @@ final class GBTClassifier(override val uid: String)
     val oldModel = oldGBT.run(oldDataset)
     GBTClassificationModel.fromOld(oldModel, this, categoricalFeatures)
   }
+
+  override def copy(extra: ParamMap): GBTClassifier = defaultCopyWithParams(extra)
 }
 
 @Experimental

mllib/src/main/scala/org/apache/spark/ml/classification/LogisticRegression.scala

@@ -220,6 +220,8 @@ class LogisticRegression(override val uid: String)
 
     new LogisticRegressionModel(uid, weights.compressed, intercept)
   }
+
+  override def copy(extra: ParamMap): LogisticRegression = defaultCopyWithParams(extra)
 }
 
 /**

mllib/src/main/scala/org/apache/spark/ml/classification/OneVsRest.scala

@@ -24,7 +24,7 @@ import scala.language.existentials
 import org.apache.spark.annotation.Experimental
 import org.apache.spark.ml._
 import org.apache.spark.ml.attribute._
-import org.apache.spark.ml.param.{ParamMap, Param}
+import org.apache.spark.ml.param.{Param, ParamMap}
 import org.apache.spark.ml.util.{Identifiable, MetadataUtils}
 import org.apache.spark.mllib.linalg.Vector
 import org.apache.spark.sql.{DataFrame, Row}
@@ -217,7 +217,7 @@ final class OneVsRest(override val uid: String)
   }
 
   override def copy(extra: ParamMap): OneVsRest = {
-    val copied = super.copy(extra).asInstanceOf[OneVsRest]
+    val copied = defaultCopyWithParams(extra).asInstanceOf[OneVsRest]
     if (isDefined(classifier)) {
       copied.setClassifier($(classifier).copy(extra))
     }

mllib/src/main/scala/org/apache/spark/ml/classification/RandomForestClassifier.scala

@@ -97,6 +97,8 @@ final class RandomForestClassifier(override val uid: String)
       oldDataset, strategy, getNumTrees, getFeatureSubsetStrategy, getSeed.toInt)
     RandomForestClassificationModel.fromOld(oldModel, this, categoricalFeatures)
   }
+
+  override def copy(extra: ParamMap): RandomForestClassifier = defaultCopyWithParams(extra)
 }
 
 @Experimental

mllib/src/main/scala/org/apache/spark/ml/evaluation/BinaryClassificationEvaluator.scala

@@ -79,4 +79,6 @@ class BinaryClassificationEvaluator(override val uid: String)
     metrics.unpersist()
     metric
   }
+
+  override def copy(extra: ParamMap): BinaryClassificationEvaluator = defaultCopyWithParams(extra)
 }

mllib/src/main/scala/org/apache/spark/ml/evaluation/Evaluator.scala

@@ -46,7 +46,5 @@ abstract class Evaluator extends Params {
    */
   def evaluate(dataset: DataFrame): Double
 
-  override def copy(extra: ParamMap): Evaluator = {
-    super.copy(extra).asInstanceOf[Evaluator]
-  }
+  override def copy(extra: ParamMap): Evaluator
 }

mllib/src/main/scala/org/apache/spark/ml/evaluation/RegressionEvaluator.scala

@@ -18,7 +18,7 @@
 package org.apache.spark.ml.evaluation
 
 import org.apache.spark.annotation.Experimental
-import org.apache.spark.ml.param.{Param, ParamValidators}
+import org.apache.spark.ml.param.{Param, ParamMap, ParamValidators}
 import org.apache.spark.ml.param.shared.{HasLabelCol, HasPredictionCol}
 import org.apache.spark.ml.util.{Identifiable, SchemaUtils}
 import org.apache.spark.mllib.evaluation.RegressionMetrics
@@ -80,4 +80,6 @@ final class RegressionEvaluator(override val uid: String)
     }
     metric
   }
+
+  override def copy(extra: ParamMap): RegressionEvaluator = defaultCopyWithParams(extra)
 }

mllib/src/main/scala/org/apache/spark/ml/feature/Binarizer.scala

@@ -83,4 +83,6 @@ final class Binarizer(override val uid: String)
     val outputFields = inputFields :+ attr.toStructField()
     StructType(outputFields)
   }
+
+  override def copy(extra: ParamMap): Binarizer = defaultCopyWithParams(extra)
 }

mllib/src/main/scala/org/apache/spark/ml/feature/Bucketizer.scala

@@ -89,6 +89,8 @@ final class Bucketizer(override val uid: String)
     SchemaUtils.checkColumnType(schema, $(inputCol), DoubleType)
     SchemaUtils.appendColumn(schema, prepOutputField(schema))
   }
+
+  override def copy(extra: ParamMap): Bucketizer = defaultCopyWithParams(extra)
 }
 
 private[feature] object Bucketizer {

mllib/src/main/scala/org/apache/spark/ml/feature/ElementwiseProduct.scala

@@ -19,7 +19,7 @@ package org.apache.spark.ml.feature
 
 import org.apache.spark.annotation.Experimental
 import org.apache.spark.ml.UnaryTransformer
-import org.apache.spark.ml.param.Param
+import org.apache.spark.ml.param.{ParamMap, Param}
 import org.apache.spark.ml.util.Identifiable
 import org.apache.spark.mllib.feature
 import org.apache.spark.mllib.linalg.{Vector, VectorUDT}

mllib/src/main/scala/org/apache/spark/ml/feature/HashingTF.scala

@@ -20,7 +20,7 @@ package org.apache.spark.ml.feature
 import org.apache.spark.annotation.Experimental
 import org.apache.spark.ml.Transformer
 import org.apache.spark.ml.attribute.AttributeGroup
-import org.apache.spark.ml.param.{IntParam, ParamValidators}
+import org.apache.spark.ml.param.{IntParam, ParamMap, ParamValidators}
 import org.apache.spark.ml.param.shared.{HasInputCol, HasOutputCol}
 import org.apache.spark.ml.util.{Identifiable, SchemaUtils}
 import org.apache.spark.mllib.feature
@@ -74,4 +74,6 @@ class HashingTF(override val uid: String) extends Transformer with HasInputCol w
     val attrGroup = new AttributeGroup($(outputCol), $(numFeatures))
     SchemaUtils.appendColumn(schema, attrGroup.toStructField())
   }
+
+  override def copy(extra: ParamMap): HashingTF = defaultCopyWithParams(extra)
 }

mllib/src/main/scala/org/apache/spark/ml/feature/IDF.scala

@@ -82,6 +82,8 @@ final class IDF(override val uid: String) extends Estimator[IDFModel] with IDFBa
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): IDF = defaultCopyWithParams(extra)
 }
 
 /**
@@ -109,4 +111,9 @@ class IDFModel private[ml] (
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): IDFModel = {
+    val copied = new IDFModel(uid, idfModel)
+    copyValues(copied, extra)
+  }
 }

mllib/src/main/scala/org/apache/spark/ml/feature/OneHotEncoder.scala

@@ -165,4 +165,6 @@ class OneHotEncoder(override val uid: String) extends Transformer
 
     dataset.select(col("*"), encode(col(inputColName).cast(DoubleType)).as(outputColName, metadata))
   }
+
+  override def copy(extra: ParamMap): OneHotEncoder = defaultCopyWithParams(extra)
 }

mllib/src/main/scala/org/apache/spark/ml/feature/StandardScaler.scala

@@ -92,6 +92,8 @@ class StandardScaler(override val uid: String) extends Estimator[StandardScalerM
     val outputFields = schema.fields :+ StructField($(outputCol), new VectorUDT, false)
     StructType(outputFields)
   }
+
+  override def copy(extra: ParamMap): StandardScaler = defaultCopyWithParams(extra)
 }
 
 /**
@@ -125,4 +127,9 @@ class StandardScalerModel private[ml] (
     val outputFields = schema.fields :+ StructField($(outputCol), new VectorUDT, false)
     StructType(outputFields)
   }
+
+  override def copy(extra: ParamMap): StandardScalerModel = {
+    val copied = new StandardScalerModel(uid, scaler)
+    copyValues(copied, extra)
+  }
 }

mllib/src/main/scala/org/apache/spark/ml/feature/StringIndexer.scala

@@ -83,6 +83,8 @@ class StringIndexer(override val uid: String) extends Estimator[StringIndexerMod
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): StringIndexer = defaultCopyWithParams(extra)
 }
 
 /**
@@ -130,4 +132,9 @@ class StringIndexerModel private[ml] (
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): StringIndexerModel = {
+    val copied = new StringIndexerModel(uid, labels)
+    copyValues(copied, extra)
+  }
 }

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.Experimental
 import org.apache.spark.ml.Transformer
 import org.apache.spark.ml.attribute.{Attribute, AttributeGroup, NumericAttribute, UnresolvedAttribute}
+import org.apache.spark.ml.param.ParamMap
 import org.apache.spark.ml.param.shared._
 import org.apache.spark.ml.util.Identifiable
 import org.apache.spark.mllib.linalg.{Vector, VectorUDT, Vectors}
@@ -117,6 +118,8 @@ class VectorAssembler(override val uid: String)
     }
     StructType(schema.fields :+ new StructField(outputColName, new VectorUDT, false))
   }
+
+  override def copy(extra: ParamMap): VectorAssembler = defaultCopyWithParams(extra)
 }
 
 private object VectorAssembler {

mllib/src/main/scala/org/apache/spark/ml/feature/VectorIndexer.scala

@@ -25,7 +25,7 @@ import scala.collection.JavaConverters._
 import org.apache.spark.annotation.Experimental
 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.{IntParam, ParamMap, ParamValidators, Params}
 import org.apache.spark.ml.param.shared._
 import org.apache.spark.ml.util.{Identifiable, SchemaUtils}
 import org.apache.spark.mllib.linalg.{DenseVector, SparseVector, Vector, VectorUDT}
@@ -131,6 +131,8 @@ class VectorIndexer(override val uid: String) extends Estimator[VectorIndexerMod
     SchemaUtils.checkColumnType(schema, $(inputCol), dataType)
     SchemaUtils.appendColumn(schema, $(outputCol), dataType)
   }
+
+  override def copy(extra: ParamMap): VectorIndexer = defaultCopyWithParams(extra)
 }
 
 private object VectorIndexer {
@@ -399,4 +401,9 @@ class VectorIndexerModel private[ml] (
     val newAttributeGroup = new AttributeGroup($(outputCol), featureAttributes)
     newAttributeGroup.toStructField()
   }
+
+  override def copy(extra: ParamMap): VectorIndexerModel = {
+    val copied = new VectorIndexerModel(uid, numFeatures, categoryMaps)
+    copyValues(copied, extra)
+  }
 }

mllib/src/main/scala/org/apache/spark/ml/feature/Word2Vec.scala

@@ -132,6 +132,8 @@ final class Word2Vec(override val uid: String) extends Estimator[Word2VecModel]
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): Word2Vec = defaultCopyWithParams(extra)
 }
 
 /**
@@ -180,4 +182,9 @@ class Word2VecModel private[ml] (
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): Word2VecModel = {
+    val copied = new Word2VecModel(uid, wordVectors)
+    copyValues(copied, extra)
+  }
 }

mllib/src/main/scala/org/apache/spark/ml/param/params.scala

@@ -494,13 +494,20 @@ trait Params extends Identifiable with Serializable {
 
   /**
    * 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.
+   * Subclasses should implement this method and set the return type properly.
+   *
+   * @see [[defaultCopyWithParams()]]
+   */
+  def copy(extra: ParamMap): Params
+
+  /**
+   * Default implementation of copy with extra params.
+   * It 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 = {
+  protected final def defaultCopyWithParams[T <: Params](extra: ParamMap): T = {
     val that = this.getClass.getConstructor(classOf[String]).newInstance(uid)
-    copyValues(that, extra)
+    copyValues(that, extra).asInstanceOf[T]
   }
 
   /**

mllib/src/main/scala/org/apache/spark/ml/recommendation/ALS.scala

@@ -216,6 +216,11 @@ class ALSModel private[ml] (
     SchemaUtils.checkColumnType(schema, $(itemCol), IntegerType)
     SchemaUtils.appendColumn(schema, $(predictionCol), FloatType)
   }
+
+  override def copy(extra: ParamMap): ALSModel = {
+    val copied = new ALSModel(uid, rank, userFactors, itemFactors)
+    copyValues(copied, extra)
+  }
 }
 
 
@@ -330,6 +335,8 @@ class ALS(override val uid: String) extends Estimator[ALSModel] with ALSParams {
   override def transformSchema(schema: StructType): StructType = {
     validateAndTransformSchema(schema)
   }
+
+  override def copy(extra: ParamMap): ALS = defaultCopyWithParams(extra)
 }
 
 /**

mllib/src/main/scala/org/apache/spark/ml/regression/DecisionTreeRegressor.scala

@@ -76,6 +76,8 @@ final class DecisionTreeRegressor(override val uid: String)
     super.getOldStrategy(categoricalFeatures, numClasses = 0, OldAlgo.Regression, getOldImpurity,
       subsamplingRate = 1.0)
   }
+
+  override def copy(extra: ParamMap): DecisionTreeRegressor = defaultCopyWithParams(extra)
 }
 
 @Experimental

mllib/src/main/scala/org/apache/spark/ml/regression/GBTRegressor.scala

@@ -131,6 +131,8 @@ final class GBTRegressor(override val uid: String)
     val oldModel = oldGBT.run(oldDataset)
     GBTRegressionModel.fromOld(oldModel, this, categoricalFeatures)
   }
+
+  override def copy(extra: ParamMap): GBTRegressor = defaultCopyWithParams(extra)
 }
 
 @Experimental

mllib/src/main/scala/org/apache/spark/ml/regression/LinearRegression.scala

@@ -186,6 +186,8 @@ class LinearRegression(override val uid: String)
     // TODO: Converts to sparse format based on the storage, but may base on the scoring speed.
     copyValues(new LinearRegressionModel(uid, weights.compressed, intercept))
   }
+
+  override def copy(extra: ParamMap): LinearRegression = defaultCopyWithParams(extra)
 }
 
 /**