[SPARK-12811][ML] Estimator for Generalized Linear Models(GLMs)

Estimator for Generalized Linear Models(GLMs) which will be solved by IRLS.

cc mengxr

Author: Yanbo Liang <ybliang8@gmail.com>

Closes #11136 from yanboliang/spark-12811.

mllib/src/main/scala/org/apache/spark/ml/optim/WeightedLeastSquares.scala

@@ -156,6 +156,12 @@ private[ml] class WeightedLeastSquares(
 
 private[ml] object WeightedLeastSquares {
 
+  /**
+   * In order to take the normal equation approach efficiently, [[WeightedLeastSquares]]
+   * only supports the number of features is no more than 4096.
+   */
+  val MAX_NUM_FEATURES: Int = 4096
+
   /**
    * Aggregator to provide necessary summary statistics for solving [[WeightedLeastSquares]].
    */
@@ -174,8 +180,8 @@ private[ml] object WeightedLeastSquares {
     private var aaSum: DenseVector = _
 
     private def init(k: Int): Unit = {
-      require(k <= 4096, "In order to take the normal equation approach efficiently, " +
+      require(k <= MAX_NUM_FEATURES, "In order to take the normal equation approach efficiently, " +
-        s"we set the max number of features to 4096 but got $k.")
+        s"we set the max number of features to $MAX_NUM_FEATURES but got $k.")
       this.k = k
       triK = k * (k + 1) / 2
       count = 0L

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

@@ -0,0 +1,577 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.ml.regression
+
+import breeze.stats.distributions.{Gaussian => GD}
+
+import org.apache.spark.{Logging, SparkException}
+import org.apache.spark.annotation.{Experimental, Since}
+import org.apache.spark.ml.PredictorParams
+import org.apache.spark.ml.feature.Instance
+import org.apache.spark.ml.optim._
+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.{BLAS, Vector}
+import org.apache.spark.rdd.RDD
+import org.apache.spark.sql.{DataFrame, Row}
+import org.apache.spark.sql.functions._
+
+/**
+ * Params for Generalized Linear Regression.
+ */
+private[regression] trait GeneralizedLinearRegressionBase extends PredictorParams
+  with HasFitIntercept with HasMaxIter with HasTol with HasRegParam with HasWeightCol
+  with HasSolver with Logging {
+
+  /**
+   * Param for the name of family which is a description of the error distribution
+   * to be used in the model.
+   * Supported options: "gaussian", "binomial", "poisson" and "gamma".
+   * Default is "gaussian".
+   * @group param
+   */
+  @Since("2.0.0")
+  final val family: Param[String] = new Param(this, "family",
+    "The name of family which is a description of the error distribution to be used in the " +
+      "model. Supported options: gaussian(default), binomial, poisson and gamma.",
+    ParamValidators.inArray[String](GeneralizedLinearRegression.supportedFamilyNames.toArray))
+
+  /** @group getParam */
+  @Since("2.0.0")
+  def getFamily: String = $(family)
+
+  /**
+   * Param for the name of link function which provides the relationship
+   * between the linear predictor and the mean of the distribution function.
+   * Supported options: "identity", "log", "inverse", "logit", "probit", "cloglog" and "sqrt".
+   * @group param
+   */
+  @Since("2.0.0")
+  final val link: Param[String] = new Param(this, "link", "The name of link function " +
+    "which provides the relationship between the linear predictor and the mean of the " +
+    "distribution function. Supported options: identity, log, inverse, logit, probit, " +
+    "cloglog and sqrt.",
+    ParamValidators.inArray[String](GeneralizedLinearRegression.supportedLinkNames.toArray))
+
+  /** @group getParam */
+  @Since("2.0.0")
+  def getLink: String = $(link)
+
+  import GeneralizedLinearRegression._
+
+  @Since("2.0.0")
+  override def validateParams(): Unit = {
+    if ($(solver) == "irls") {
+      setDefault(maxIter -> 25)
+    }
+    if (isDefined(link)) {
+      require(supportedFamilyAndLinkPairs.contains(
+        Family.fromName($(family)) -> Link.fromName($(link))), "Generalized Linear Regression " +
+        s"with ${$(family)} family does not support ${$(link)} link function.")
+    }
+  }
+}
+
+/**
+ * :: Experimental ::
+ *
+ * Fit a Generalized Linear Model ([[https://en.wikipedia.org/wiki/Generalized_linear_model]])
+ * specified by giving a symbolic description of the linear predictor (link function) and
+ * a description of the error distribution (family).
+ * It supports "gaussian", "binomial", "poisson" and "gamma" as family.
+ * Valid link functions for each family is listed below. The first link function of each family
+ * is the default one.
+ *  - "gaussian" -> "identity", "log", "inverse"
+ *  - "binomial" -> "logit", "probit", "cloglog"
+ *  - "poisson"  -> "log", "identity", "sqrt"
+ *  - "gamma"    -> "inverse", "identity", "log"
+ */
+@Experimental
+@Since("2.0.0")
+class GeneralizedLinearRegression @Since("2.0.0") (@Since("2.0.0") override val uid: String)
+  extends Regressor[Vector, GeneralizedLinearRegression, GeneralizedLinearRegressionModel]
+  with GeneralizedLinearRegressionBase with Logging {
+
+  import GeneralizedLinearRegression._
+
+  @Since("2.0.0")
+  def this() = this(Identifiable.randomUID("glm"))
+
+  /**
+   * Sets the value of param [[family]].
+   * Default is "gaussian".
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setFamily(value: String): this.type = set(family, value)
+  setDefault(family -> Gaussian.name)
+
+  /**
+   * Sets the value of param [[link]].
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setLink(value: String): this.type = set(link, value)
+
+  /**
+   * Sets if we should fit the intercept.
+   * Default is true.
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setFitIntercept(value: Boolean): this.type = set(fitIntercept, value)
+
+  /**
+   * Sets the maximum number of iterations.
+   * Default is 25 if the solver algorithm is "irls".
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setMaxIter(value: Int): this.type = set(maxIter, value)
+
+  /**
+   * Sets the convergence tolerance of iterations.
+   * Smaller value will lead to higher accuracy with the cost of more iterations.
+   * Default is 1E-6.
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setTol(value: Double): this.type = set(tol, value)
+  setDefault(tol -> 1E-6)
+
+  /**
+   * Sets the regularization parameter.
+   * Default is 0.0.
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setRegParam(value: Double): this.type = set(regParam, value)
+  setDefault(regParam -> 0.0)
+
+  /**
+   * Sets the value of param [[weightCol]].
+   * If this is not set or empty, we treat all instance weights as 1.0.
+   * Default is empty, so all instances have weight one.
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setWeightCol(value: String): this.type = set(weightCol, value)
+  setDefault(weightCol -> "")
+
+  /**
+   * Sets the solver algorithm used for optimization.
+   * Currently only support "irls" which is also the default solver.
+   * @group setParam
+   */
+  @Since("2.0.0")
+  def setSolver(value: String): this.type = set(solver, value)
+  setDefault(solver -> "irls")
+
+  override protected def train(dataset: DataFrame): GeneralizedLinearRegressionModel = {
+    val familyObj = Family.fromName($(family))
+    val linkObj = if (isDefined(link)) {
+      Link.fromName($(link))
+    } else {
+      familyObj.defaultLink
+    }
+    val familyAndLink = new FamilyAndLink(familyObj, linkObj)
+
+    val numFeatures = dataset.select(col($(featuresCol))).limit(1).rdd
+      .map { case Row(features: Vector) =>
+        features.size
+      }.first()
+    if (numFeatures > WeightedLeastSquares.MAX_NUM_FEATURES) {
+      val msg = "Currently, GeneralizedLinearRegression only supports number of features" +
+        s" <= ${WeightedLeastSquares.MAX_NUM_FEATURES}. Found $numFeatures in the input dataset."
+      throw new SparkException(msg)
+    }
+
+    val w = if ($(weightCol).isEmpty) lit(1.0) else col($(weightCol))
+    val instances: RDD[Instance] = dataset.select(col($(labelCol)), w, col($(featuresCol))).rdd
+      .map { case Row(label: Double, weight: Double, features: Vector) =>
+        Instance(label, weight, features)
+      }
+
+    if (familyObj == Gaussian && linkObj == Identity) {
+      // TODO: Make standardizeFeatures and standardizeLabel configurable.
+      val optimizer = new WeightedLeastSquares($(fitIntercept), $(regParam),
+        standardizeFeatures = true, standardizeLabel = true)
+      val wlsModel = optimizer.fit(instances)
+      val model = copyValues(
+        new GeneralizedLinearRegressionModel(uid, wlsModel.coefficients, wlsModel.intercept)
+          .setParent(this))
+      return model
+    }
+
+    // Fit Generalized Linear Model by iteratively reweighted least squares (IRLS).
+    val initialModel = familyAndLink.initialize(instances, $(fitIntercept), $(regParam))
+    val optimizer = new IterativelyReweightedLeastSquares(initialModel, familyAndLink.reweightFunc,
+      $(fitIntercept), $(regParam), $(maxIter), $(tol))
+    val irlsModel = optimizer.fit(instances)
+
+    val model = copyValues(
+      new GeneralizedLinearRegressionModel(uid, irlsModel.coefficients, irlsModel.intercept)
+        .setParent(this))
+    model
+  }
+
+  @Since("2.0.0")
+  override def copy(extra: ParamMap): GeneralizedLinearRegression = defaultCopy(extra)
+}
+
+@Since("2.0.0")
+private[ml] object GeneralizedLinearRegression {
+
+  /** Set of family and link pairs that GeneralizedLinearRegression supports. */
+  lazy val supportedFamilyAndLinkPairs = Set(
+    Gaussian -> Identity, Gaussian -> Log, Gaussian -> Inverse,
+    Binomial -> Logit, Binomial -> Probit, Binomial -> CLogLog,
+    Poisson -> Log, Poisson -> Identity, Poisson -> Sqrt,
+    Gamma -> Inverse, Gamma -> Identity, Gamma -> Log
+  )
+
+  /** Set of family names that GeneralizedLinearRegression supports. */
+  lazy val supportedFamilyNames = supportedFamilyAndLinkPairs.map(_._1.name)
+
+  /** Set of link names that GeneralizedLinearRegression supports. */
+  lazy val supportedLinkNames = supportedFamilyAndLinkPairs.map(_._2.name)
+
+  val epsilon: Double = 1E-16
+
+  /**
+   * Wrapper of family and link combination used in the model.
+   */
+  private[ml] class FamilyAndLink(val family: Family, val link: Link) extends Serializable {
+
+    /** Linear predictor based on given mu. */
+    def predict(mu: Double): Double = link.link(family.project(mu))
+
+    /** Fitted value based on linear predictor eta. */
+    def fitted(eta: Double): Double = family.project(link.unlink(eta))
+
+    /**
+     * Get the initial guess model for [[IterativelyReweightedLeastSquares]].
+     */
+    def initialize(
+        instances: RDD[Instance],
+        fitIntercept: Boolean,
+        regParam: Double): WeightedLeastSquaresModel = {
+      val newInstances = instances.map { instance =>
+        val mu = family.initialize(instance.label, instance.weight)
+        val eta = predict(mu)
+        Instance(eta, instance.weight, instance.features)
+      }
+      // TODO: Make standardizeFeatures and standardizeLabel configurable.
+      val initialModel = new WeightedLeastSquares(fitIntercept, regParam,
+        standardizeFeatures = true, standardizeLabel = true)
+        .fit(newInstances)
+      initialModel
+    }
+
+    /**
+     * The reweight function used to update offsets and weights
+     * at each iteration of [[IterativelyReweightedLeastSquares]].
+     */
+    val reweightFunc: (Instance, WeightedLeastSquaresModel) => (Double, Double) = {
+      (instance: Instance, model: WeightedLeastSquaresModel) => {
+        val eta = model.predict(instance.features)
+        val mu = fitted(eta)
+        val offset = eta + (instance.label - mu) * link.deriv(mu)
+        val weight = instance.weight / (math.pow(this.link.deriv(mu), 2.0) * family.variance(mu))
+        (offset, weight)
+      }
+    }
+  }
+
+  /**
+   * A description of the error distribution to be used in the model.
+   * @param name the name of the family.
+   */
+  private[ml] abstract class Family(val name: String) extends Serializable {
+
+    /** The default link instance of this family. */
+    val defaultLink: Link
+
+    /** Initialize the starting value for mu. */
+    def initialize(y: Double, weight: Double): Double
+
+    /** The variance of the endogenous variable's mean, given the value mu. */
+    def variance(mu: Double): Double
+
+    /** Trim the fitted value so that it will be in valid range. */
+    def project(mu: Double): Double = mu
+  }
+
+  private[ml] object Family {
+
+    /**
+     * Gets the [[Family]] object from its name.
+     * @param name family name: "gaussian", "binomial", "poisson" or "gamma".
+     */
+    def fromName(name: String): Family = {
+      name match {
+        case Gaussian.name => Gaussian
+        case Binomial.name => Binomial
+        case Poisson.name => Poisson
+        case Gamma.name => Gamma
+      }
+    }
+  }
+
+  /**
+   * Gaussian exponential family distribution.
+   * The default link for the Gaussian family is the identity link.
+   */
+  private[ml] object Gaussian extends Family("gaussian") {
+
+    val defaultLink: Link = Identity
+
+    override def initialize(y: Double, weight: Double): Double = y
+
+    def variance(mu: Double): Double = 1.0
+
+    override def project(mu: Double): Double = {
+      if (mu.isNegInfinity) {
+        Double.MinValue
+      } else if (mu.isPosInfinity) {
+        Double.MaxValue
+      } else {
+        mu
+      }
+    }
+  }
+
+  /**
+   * Binomial exponential family distribution.
+   * The default link for the Binomial family is the logit link.
+   */
+  private[ml] object Binomial extends Family("binomial") {
+
+    val defaultLink: Link = Logit
+
+    override def initialize(y: Double, weight: Double): Double = {
+      val mu = (weight * y + 0.5) / (weight + 1.0)
+      require(mu > 0.0 && mu < 1.0, "The response variable of Binomial family" +
+        s"should be in range (0, 1), but got $mu")
+      mu
+    }
+
+    override def variance(mu: Double): Double = mu * (1.0 - mu)
+
+    override def project(mu: Double): Double = {
+      if (mu < epsilon) {
+        epsilon
+      } else if (mu > 1.0 - epsilon) {
+        1.0 - epsilon
+      } else {
+        mu
+      }
+    }
+  }
+
+  /**
+   * Poisson exponential family distribution.
+   * The default link for the Poisson family is the log link.
+   */
+  private[ml] object Poisson extends Family("poisson") {
+
+    val defaultLink: Link = Log
+
+    override def initialize(y: Double, weight: Double): Double = {
+      require(y > 0.0, "The response variable of Poisson family " +
+        s"should be positive, but got $y")
+      y
+    }
+
+    override def variance(mu: Double): Double = mu
+
+    override def project(mu: Double): Double = {
+      if (mu < epsilon) {
+        epsilon
+      } else if (mu.isInfinity) {
+        Double.MaxValue
+      } else {
+        mu
+      }
+    }
+  }
+
+  /**
+   * Gamma exponential family distribution.
+   * The default link for the Gamma family is the inverse link.
+   */
+  private[ml] object Gamma extends Family("gamma") {
+
+    val defaultLink: Link = Inverse
+
+    override def initialize(y: Double, weight: Double): Double = {
+      require(y > 0.0, "The response variable of Gamma family " +
+        s"should be positive, but got $y")
+      y
+    }
+
+    override def variance(mu: Double): Double = math.pow(mu, 2.0)
+
+    override def project(mu: Double): Double = {
+      if (mu < epsilon) {
+        epsilon
+      } else if (mu.isInfinity) {
+        Double.MaxValue
+      } else {
+        mu
+      }
+    }
+  }
+
+  /**
+   * A description of the link function to be used in the model.
+   * The link function provides the relationship between the linear predictor
+   * and the mean of the distribution function.
+   * @param name the name of link function.
+   */
+  private[ml] abstract class Link(val name: String) extends Serializable {
+
+    /** The link function. */
+    def link(mu: Double): Double
+
+    /** Derivative of the link function. */
+    def deriv(mu: Double): Double
+
+    /** The inverse link function. */
+    def unlink(eta: Double): Double
+  }
+
+  private[ml] object Link {
+
+    /**
+     * Gets the [[Link]] object from its name.
+     * @param name link name: "identity", "logit", "log",
+     *             "inverse", "probit", "cloglog" or "sqrt".
+     */
+    def fromName(name: String): Link = {
+      name match {
+        case Identity.name => Identity
+        case Logit.name => Logit
+        case Log.name => Log
+        case Inverse.name => Inverse
+        case Probit.name => Probit
+        case CLogLog.name => CLogLog
+        case Sqrt.name => Sqrt
+      }
+    }
+  }
+
+  private[ml] object Identity extends Link("identity") {
+
+    override def link(mu: Double): Double = mu
+
+    override def deriv(mu: Double): Double = 1.0
+
+    override def unlink(eta: Double): Double = eta
+  }
+
+  private[ml] object Logit extends Link("logit") {
+
+    override def link(mu: Double): Double = math.log(mu / (1.0 - mu))
+
+    override def deriv(mu: Double): Double = 1.0 / (mu * (1.0 - mu))
+
+    override def unlink(eta: Double): Double = 1.0 / (1.0 + math.exp(-1.0 * eta))
+  }
+
+  private[ml] object Log extends Link("log") {
+
+    override def link(mu: Double): Double = math.log(mu)
+
+    override def deriv(mu: Double): Double = 1.0 / mu
+
+    override def unlink(eta: Double): Double = math.exp(eta)
+  }
+
+  private[ml] object Inverse extends Link("inverse") {
+
+    override def link(mu: Double): Double = 1.0 / mu
+
+    override def deriv(mu: Double): Double = -1.0 * math.pow(mu, -2.0)
+
+    override def unlink(eta: Double): Double = 1.0 / eta
+  }
+
+  private[ml] object Probit extends Link("probit") {
+
+    override def link(mu: Double): Double = GD(0.0, 1.0).icdf(mu)
+
+    override def deriv(mu: Double): Double = 1.0 / GD(0.0, 1.0).pdf(GD(0.0, 1.0).icdf(mu))
+
+    override def unlink(eta: Double): Double = GD(0.0, 1.0).cdf(eta)
+  }
+
+  private[ml] object CLogLog extends Link("cloglog") {
+
+    override def link(mu: Double): Double = math.log(-1.0 * math.log(1 - mu))
+
+    override def deriv(mu: Double): Double = 1.0 / ((mu - 1.0) * math.log(1.0 - mu))
+
+    override def unlink(eta: Double): Double = 1.0 - math.exp(-1.0 * math.exp(eta))
+  }
+
+  private[ml] object Sqrt extends Link("sqrt") {
+
+    override def link(mu: Double): Double = math.sqrt(mu)
+
+    override def deriv(mu: Double): Double = 1.0 / (2.0 * math.sqrt(mu))
+
+    override def unlink(eta: Double): Double = math.pow(eta, 2.0)
+  }
+}
+
+/**
+ * :: Experimental ::
+ * Model produced by [[GeneralizedLinearRegression]].
+ */
+@Experimental
+@Since("2.0.0")
+class GeneralizedLinearRegressionModel private[ml] (
+    @Since("2.0.0") override val uid: String,
+    @Since("2.0.0") val coefficients: Vector,
+    @Since("2.0.0") val intercept: Double)
+  extends RegressionModel[Vector, GeneralizedLinearRegressionModel]
+  with GeneralizedLinearRegressionBase {
+
+  import GeneralizedLinearRegression._
+
+  lazy val familyObj = Family.fromName($(family))
+  lazy val linkObj = if (isDefined(link)) {
+    Link.fromName($(link))
+  } else {
+    familyObj.defaultLink
+  }
+  lazy val familyAndLink = new FamilyAndLink(familyObj, linkObj)
+
+  override protected def predict(features: Vector): Double = {
+    val eta = BLAS.dot(features, coefficients) + intercept
+    familyAndLink.fitted(eta)
+  }
+
+  @Since("2.0.0")
+  override def copy(extra: ParamMap): GeneralizedLinearRegressionModel = {
+    copyValues(new GeneralizedLinearRegressionModel(uid, coefficients, intercept), extra)
+      .setParent(parent)
+  }
+}

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

@@ -163,8 +163,8 @@ class LinearRegression @Since("1.3.0") (@Since("1.3.0") override val uid: String
     }.first()
     val w = if ($(weightCol).isEmpty) lit(1.0) else col($(weightCol))
 
-    if (($(solver) == "auto" && $(elasticNetParam) == 0.0 && numFeatures <= 4096) ||
+    if (($(solver) == "auto" && $(elasticNetParam) == 0.0 &&
-      $(solver) == "normal") {
+      numFeatures <= WeightedLeastSquares.MAX_NUM_FEATURES) || $(solver) == "normal") {
       require($(elasticNetParam) == 0.0, "Only L2 regularization can be used when normal " +
         "solver is used.'")
       // For low dimensional data, WeightedLeastSquares is more efficiently since the

mllib/src/test/scala/org/apache/spark/ml/regression/GeneralizedLinearRegressionSuite.scala

@@ -0,0 +1,507 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.ml.regression
+
+import scala.util.Random
+
+import org.apache.spark.SparkFunSuite
+import org.apache.spark.ml.param.ParamsSuite
+import org.apache.spark.ml.util.MLTestingUtils
+import org.apache.spark.mllib.classification.LogisticRegressionSuite._
+import org.apache.spark.mllib.linalg.{BLAS, DenseVector, Vectors}
+import org.apache.spark.mllib.random._
+import org.apache.spark.mllib.regression.LabeledPoint
+import org.apache.spark.mllib.util.MLlibTestSparkContext
+import org.apache.spark.mllib.util.TestingUtils._
+import org.apache.spark.sql.{DataFrame, Row}
+
+class GeneralizedLinearRegressionSuite extends SparkFunSuite with MLlibTestSparkContext {
+
+  private val seed: Int = 42
+  @transient var datasetGaussianIdentity: DataFrame = _
+  @transient var datasetGaussianLog: DataFrame = _
+  @transient var datasetGaussianInverse: DataFrame = _
+  @transient var datasetBinomial: DataFrame = _
+  @transient var datasetPoissonLog: DataFrame = _
+  @transient var datasetPoissonIdentity: DataFrame = _
+  @transient var datasetPoissonSqrt: DataFrame = _
+  @transient var datasetGammaInverse: DataFrame = _
+  @transient var datasetGammaIdentity: DataFrame = _
+  @transient var datasetGammaLog: DataFrame = _
+
+  override def beforeAll(): Unit = {
+    super.beforeAll()
+
+    import GeneralizedLinearRegressionSuite._
+
+    datasetGaussianIdentity = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 2.5, coefficients = Array(2.2, 0.6), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "gaussian", link = "identity"), 2))
+
+    datasetGaussianLog = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 0.25, coefficients = Array(0.22, 0.06), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "gaussian", link = "log"), 2))
+
+    datasetGaussianInverse = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 2.5, coefficients = Array(2.2, 0.6), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "gaussian", link = "inverse"), 2))
+
+    datasetBinomial = {
+      val nPoints = 10000
+      val coefficients = Array(-0.57997, 0.912083, -0.371077, -0.819866, 2.688191)
+      val xMean = Array(5.843, 3.057, 3.758, 1.199)
+      val xVariance = Array(0.6856, 0.1899, 3.116, 0.581)
+
+      val testData =
+        generateMultinomialLogisticInput(coefficients, xMean, xVariance,
+          addIntercept = true, nPoints, seed)
+
+      sqlContext.createDataFrame(sc.parallelize(testData, 2))
+    }
+
+    datasetPoissonLog = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 0.25, coefficients = Array(0.22, 0.06), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "poisson", link = "log"), 2))
+
+    datasetPoissonIdentity = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 2.5, coefficients = Array(2.2, 0.6), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "poisson", link = "identity"), 2))
+
+    datasetPoissonSqrt = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 2.5, coefficients = Array(2.2, 0.6), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "poisson", link = "sqrt"), 2))
+
+    datasetGammaInverse = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 2.5, coefficients = Array(2.2, 0.6), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "gamma", link = "inverse"), 2))
+
+    datasetGammaIdentity = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 2.5, coefficients = Array(2.2, 0.6), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "gamma", link = "identity"), 2))
+
+    datasetGammaLog = sqlContext.createDataFrame(
+      sc.parallelize(generateGeneralizedLinearRegressionInput(
+        intercept = 0.25, coefficients = Array(0.22, 0.06), xMean = Array(2.9, 10.5),
+        xVariance = Array(0.7, 1.2), nPoints = 10000, seed, noiseLevel = 0.01,
+        family = "gamma", link = "log"), 2))
+  }
+
+  test("params") {
+    ParamsSuite.checkParams(new GeneralizedLinearRegression)
+    val model = new GeneralizedLinearRegressionModel("genLinReg", Vectors.dense(0.0), 0.0)
+    ParamsSuite.checkParams(model)
+  }
+
+  test("generalized linear regression: default params") {
+    val glr = new GeneralizedLinearRegression
+    assert(glr.getLabelCol === "label")
+    assert(glr.getFeaturesCol === "features")
+    assert(glr.getPredictionCol === "prediction")
+    assert(glr.getFitIntercept)
+    assert(glr.getTol === 1E-6)
+    assert(glr.getWeightCol === "")
+    assert(glr.getRegParam === 0.0)
+    assert(glr.getSolver == "irls")
+    // TODO: Construct model directly instead of via fitting.
+    val model = glr.setFamily("gaussian").setLink("identity")
+      .fit(datasetGaussianIdentity)
+
+    // copied model must have the same parent.
+    MLTestingUtils.checkCopy(model)
+
+    assert(model.getFeaturesCol === "features")
+    assert(model.getPredictionCol === "prediction")
+    assert(model.intercept !== 0.0)
+    assert(model.hasParent)
+    assert(model.getFamily === "gaussian")
+    assert(model.getLink === "identity")
+  }
+
+  test("generalized linear regression: gaussian family against glm") {
+    /*
+       R code:
+       f1 <- data$V1 ~ data$V2 + data$V3 - 1
+       f2 <- data$V1 ~ data$V2 + data$V3
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family="gaussian", data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 2.2960999 0.8087933
+       [1] 2.5002642 2.2000403 0.5999485
+
+       data <- read.csv("path", header=FALSE)
+       model1 <- glm(f1, family=gaussian(link=log), data=data, start=c(0,0))
+       model2 <- glm(f2, family=gaussian(link=log), data=data, start=c(0,0,0))
+       print(as.vector(coef(model1)))
+       print(as.vector(coef(model2)))
+
+       [1] 0.23069326 0.07993778
+       [1] 0.25001858 0.22002452 0.05998789
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=gaussian(link=inverse), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 2.3010179 0.8198976
+       [1] 2.4108902 2.2130248 0.6086152
+     */
+
+    val expected = Seq(
+      Vectors.dense(0.0, 2.2960999, 0.8087933),
+      Vectors.dense(2.5002642, 2.2000403, 0.5999485),
+      Vectors.dense(0.0, 0.23069326, 0.07993778),
+      Vectors.dense(0.25001858, 0.22002452, 0.05998789),
+      Vectors.dense(0.0, 2.3010179, 0.8198976),
+      Vectors.dense(2.4108902, 2.2130248, 0.6086152))
+
+    import GeneralizedLinearRegression._
+
+    var idx = 0
+    for ((link, dataset) <- Seq(("identity", datasetGaussianIdentity), ("log", datasetGaussianLog),
+      ("inverse", datasetGaussianInverse))) {
+      for (fitIntercept <- Seq(false, true)) {
+        val trainer = new GeneralizedLinearRegression().setFamily("gaussian").setLink(link)
+          .setFitIntercept(fitIntercept)
+        val model = trainer.fit(dataset)
+        val actual = Vectors.dense(model.intercept, model.coefficients(0), model.coefficients(1))
+        assert(actual ~= expected(idx) absTol 1e-4, "Model mismatch: GLM with gaussian family, " +
+          s"$link link and fitIntercept = $fitIntercept.")
+
+        val familyLink = new FamilyAndLink(Gaussian, Link.fromName(link))
+        model.transform(dataset).select("features", "prediction").collect().foreach {
+          case Row(features: DenseVector, prediction1: Double) =>
+            val eta = BLAS.dot(features, model.coefficients) + model.intercept
+            val prediction2 = familyLink.fitted(eta)
+            assert(prediction1 ~= prediction2 relTol 1E-5, "Prediction mismatch: GLM with " +
+              s"gaussian family, $link link and fitIntercept = $fitIntercept.")
+        }
+
+        idx += 1
+      }
+    }
+  }
+
+  test("generalized linear regression: gaussian family against glmnet") {
+    /*
+       R code:
+       library(glmnet)
+       data <- read.csv("path", header=FALSE)
+       label = data$V1
+       features = as.matrix(data.frame(data$V2, data$V3))
+       for (intercept in c(FALSE, TRUE)) {
+         for (lambda in c(0.0, 0.1, 1.0)) {
+           model <- glmnet(features, label, family="gaussian", intercept=intercept,
+                           lambda=lambda, alpha=0, thresh=1E-14)
+           print(as.vector(coef(model)))
+         }
+       }
+
+       [1] 0.0000000 2.2961005 0.8087932
+       [1] 0.0000000 2.2130368 0.8309556
+       [1] 0.0000000 1.7176137 0.9610657
+       [1] 2.5002642 2.2000403 0.5999485
+       [1] 3.1106389 2.0935142 0.5712711
+       [1] 6.7597127 1.4581054 0.3994266
+     */
+
+    val expected = Seq(
+      Vectors.dense(0.0, 2.2961005, 0.8087932),
+      Vectors.dense(0.0, 2.2130368, 0.8309556),
+      Vectors.dense(0.0, 1.7176137, 0.9610657),
+      Vectors.dense(2.5002642, 2.2000403, 0.5999485),
+      Vectors.dense(3.1106389, 2.0935142, 0.5712711),
+      Vectors.dense(6.7597127, 1.4581054, 0.3994266))
+
+    var idx = 0
+    for (fitIntercept <- Seq(false, true);
+         regParam <- Seq(0.0, 0.1, 1.0)) {
+      val trainer = new GeneralizedLinearRegression().setFamily("gaussian")
+        .setFitIntercept(fitIntercept).setRegParam(regParam)
+      val model = trainer.fit(datasetGaussianIdentity)
+      val actual = Vectors.dense(model.intercept, model.coefficients(0), model.coefficients(1))
+      assert(actual ~= expected(idx) absTol 1e-4, "Model mismatch: GLM with gaussian family, " +
+        s"fitIntercept = $fitIntercept and regParam = $regParam.")
+
+      idx += 1
+    }
+  }
+
+  test("generalized linear regression: binomial family against glm") {
+    /*
+       R code:
+       f1 <- data$V1 ~ data$V2 + data$V3 + data$V4 + data$V5 - 1
+       f2 <- data$V1 ~ data$V2 + data$V3 + data$V4 + data$V5
+       data <- read.csv("path", header=FALSE)
+
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family="binomial", data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] -0.3560284  1.3010002 -0.3570805 -0.7406762
+       [1]  2.8367406 -0.5896187  0.8931655 -0.3925169 -0.7996989
+
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=binomial(link=probit), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] -0.2134390  0.7800646 -0.2144267 -0.4438358
+       [1]  1.6995366 -0.3524694  0.5332651 -0.2352985 -0.4780850
+
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=binomial(link=cloglog), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] -0.2832198  0.8434144 -0.2524727 -0.5293452
+       [1]  1.5063590 -0.4038015  0.6133664 -0.2687882 -0.5541758
+     */
+    val expected = Seq(
+      Vectors.dense(0.0, -0.3560284, 1.3010002, -0.3570805, -0.7406762),
+      Vectors.dense(2.8367406, -0.5896187, 0.8931655, -0.3925169, -0.7996989),
+      Vectors.dense(0.0, -0.2134390, 0.7800646, -0.2144267, -0.4438358),
+      Vectors.dense(1.6995366, -0.3524694, 0.5332651, -0.2352985, -0.4780850),
+      Vectors.dense(0.0, -0.2832198, 0.8434144, -0.2524727, -0.5293452),
+      Vectors.dense(1.5063590, -0.4038015, 0.6133664, -0.2687882, -0.5541758))
+
+    import GeneralizedLinearRegression._
+
+    var idx = 0
+    for ((link, dataset) <- Seq(("logit", datasetBinomial), ("probit", datasetBinomial),
+      ("cloglog", datasetBinomial))) {
+      for (fitIntercept <- Seq(false, true)) {
+        val trainer = new GeneralizedLinearRegression().setFamily("binomial").setLink(link)
+          .setFitIntercept(fitIntercept)
+        val model = trainer.fit(dataset)
+        val actual = Vectors.dense(model.intercept, model.coefficients(0), model.coefficients(1),
+          model.coefficients(2), model.coefficients(3))
+        assert(actual ~= expected(idx) absTol 1e-4, "Model mismatch: GLM with binomial family, " +
+          s"$link link and fitIntercept = $fitIntercept.")
+
+        val familyLink = new FamilyAndLink(Binomial, Link.fromName(link))
+        model.transform(dataset).select("features", "prediction").collect().foreach {
+          case Row(features: DenseVector, prediction1: Double) =>
+            val eta = BLAS.dot(features, model.coefficients) + model.intercept
+            val prediction2 = familyLink.fitted(eta)
+            assert(prediction1 ~= prediction2 relTol 1E-5, "Prediction mismatch: GLM with " +
+              s"binomial family, $link link and fitIntercept = $fitIntercept.")
+        }
+
+        idx += 1
+      }
+    }
+  }
+
+  test("generalized linear regression: poisson family against glm") {
+    /*
+       R code:
+       f1 <- data$V1 ~ data$V2 + data$V3 - 1
+       f2 <- data$V1 ~ data$V2 + data$V3
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family="poisson", data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 0.22999393 0.08047088
+       [1] 0.25022353 0.21998599 0.05998621
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=poisson(link=identity), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 2.2929501 0.8119415
+       [1] 2.5012730 2.1999407 0.5999107
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=poisson(link=sqrt), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 2.2958947 0.8090515
+       [1] 2.5000480 2.1999972 0.5999968
+     */
+    val expected = Seq(
+      Vectors.dense(0.0, 0.22999393, 0.08047088),
+      Vectors.dense(0.25022353, 0.21998599, 0.05998621),
+      Vectors.dense(0.0, 2.2929501, 0.8119415),
+      Vectors.dense(2.5012730, 2.1999407, 0.5999107),
+      Vectors.dense(0.0, 2.2958947, 0.8090515),
+      Vectors.dense(2.5000480, 2.1999972, 0.5999968))
+
+    import GeneralizedLinearRegression._
+
+    var idx = 0
+    for ((link, dataset) <- Seq(("log", datasetPoissonLog), ("identity", datasetPoissonIdentity),
+      ("sqrt", datasetPoissonSqrt))) {
+      for (fitIntercept <- Seq(false, true)) {
+        val trainer = new GeneralizedLinearRegression().setFamily("poisson").setLink(link)
+          .setFitIntercept(fitIntercept)
+        val model = trainer.fit(dataset)
+        val actual = Vectors.dense(model.intercept, model.coefficients(0), model.coefficients(1))
+        assert(actual ~= expected(idx) absTol 1e-4, "Model mismatch: GLM with poisson family, " +
+          s"$link link and fitIntercept = $fitIntercept.")
+
+        val familyLink = new FamilyAndLink(Poisson, Link.fromName(link))
+        model.transform(dataset).select("features", "prediction").collect().foreach {
+          case Row(features: DenseVector, prediction1: Double) =>
+            val eta = BLAS.dot(features, model.coefficients) + model.intercept
+            val prediction2 = familyLink.fitted(eta)
+            assert(prediction1 ~= prediction2 relTol 1E-5, "Prediction mismatch: GLM with " +
+              s"poisson family, $link link and fitIntercept = $fitIntercept.")
+        }
+
+        idx += 1
+      }
+    }
+  }
+
+  test("generalized linear regression: gamma family against glm") {
+    /*
+       R code:
+       f1 <- data$V1 ~ data$V2 + data$V3 - 1
+       f2 <- data$V1 ~ data$V2 + data$V3
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family="Gamma", data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 2.3392419 0.8058058
+       [1] 2.3507700 2.2533574 0.6042991
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=Gamma(link=identity), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 2.2908883 0.8147796
+       [1] 2.5002406 2.1998346 0.6000059
+
+       data <- read.csv("path", header=FALSE)
+       for (formula in c(f1, f2)) {
+         model <- glm(formula, family=Gamma(link=log), data=data)
+         print(as.vector(coef(model)))
+       }
+
+       [1] 0.22958970 0.08091066
+       [1] 0.25003210 0.21996957 0.06000215
+     */
+    val expected = Seq(
+      Vectors.dense(0.0, 2.3392419, 0.8058058),
+      Vectors.dense(2.3507700, 2.2533574, 0.6042991),
+      Vectors.dense(0.0, 2.2908883, 0.8147796),
+      Vectors.dense(2.5002406, 2.1998346, 0.6000059),
+      Vectors.dense(0.0, 0.22958970, 0.08091066),
+      Vectors.dense(0.25003210, 0.21996957, 0.06000215))
+
+    import GeneralizedLinearRegression._
+
+    var idx = 0
+    for ((link, dataset) <- Seq(("inverse", datasetGammaInverse),
+      ("identity", datasetGammaIdentity), ("log", datasetGammaLog))) {
+      for (fitIntercept <- Seq(false, true)) {
+        val trainer = new GeneralizedLinearRegression().setFamily("gamma").setLink(link)
+          .setFitIntercept(fitIntercept)
+        val model = trainer.fit(dataset)
+        val actual = Vectors.dense(model.intercept, model.coefficients(0), model.coefficients(1))
+        assert(actual ~= expected(idx) absTol 1e-4, "Model mismatch: GLM with gamma family, " +
+          s"$link link and fitIntercept = $fitIntercept.")
+
+        val familyLink = new FamilyAndLink(Gamma, Link.fromName(link))
+        model.transform(dataset).select("features", "prediction").collect().foreach {
+          case Row(features: DenseVector, prediction1: Double) =>
+            val eta = BLAS.dot(features, model.coefficients) + model.intercept
+            val prediction2 = familyLink.fitted(eta)
+            assert(prediction1 ~= prediction2 relTol 1E-5, "Prediction mismatch: GLM with " +
+              s"gamma family, $link link and fitIntercept = $fitIntercept.")
+        }
+
+        idx += 1
+      }
+    }
+  }
+}
+
+object GeneralizedLinearRegressionSuite {
+
+  def generateGeneralizedLinearRegressionInput(
+      intercept: Double,
+      coefficients: Array[Double],
+      xMean: Array[Double],
+      xVariance: Array[Double],
+      nPoints: Int,
+      seed: Int,
+      noiseLevel: Double,
+      family: String,
+      link: String): Seq[LabeledPoint] = {
+
+    val rnd = new Random(seed)
+    def rndElement(i: Int) = {
+      (rnd.nextDouble() - 0.5) * math.sqrt(12.0 * xVariance(i)) + xMean(i)
+    }
+    val (generator, mean) = family match {
+      case "gaussian" => (new StandardNormalGenerator, 0.0)
+      case "poisson" => (new PoissonGenerator(1.0), 1.0)
+      case "gamma" => (new GammaGenerator(1.0, 1.0), 1.0)
+    }
+    generator.setSeed(seed)
+
+    (0 until nPoints).map { _ =>
+      val features = Vectors.dense(coefficients.indices.map { rndElement(_) }.toArray)
+      val eta = BLAS.dot(Vectors.dense(coefficients), features) + intercept
+      val mu = link match {
+        case "identity" => eta
+        case "log" => math.exp(eta)
+        case "sqrt" => math.pow(eta, 2.0)
+        case "inverse" => 1.0 / eta
+      }
+      val label = mu + noiseLevel * (generator.nextValue() - mean)
+      // Return LabeledPoints with DenseVector
+      LabeledPoint(label, features)
+    }
+  }
+}