Github user mengxr commented on a diff in the pull request:
https://github.com/apache/spark/pull/11136#discussion_r53722634
--- Diff:
mllib/src/main/scala/org/apache/spark/ml/regression/GeneralizedLinearRegression.scala
---
@@ -0,0 +1,547 @@
+/*
+ * 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 GeneralizedLinearRegressionParams 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".
+ * @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",
+
ParamValidators.inArray[String](GeneralizedLinearRegression.supportedFamilies.toArray))
+
+ /** @group getParam */
+ @Since("2.0.0")
+ def getFamily: String = $(family)
+
+ /**
+ * Param for the name of the model link 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 the
model link function",
+
ParamValidators.inArray[String](GeneralizedLinearRegression.supportedLinks.toArray))
+
+ /** @group getParam */
+ @Since("2.0.0")
+ def getLink: String = $(link)
+
+ @Since("2.0.0")
+ override def validateParams(): Unit = {
+ if (isDefined(link)) {
+
require(GeneralizedLinearRegression.supportedFamilyLinkPairs.contains($(family)
-> $(link)),
+ s"Generalized Linear Regression with ${$(family)} family does not
support ${$(link)} " +
+ s"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 and
+ * a description of the error distribution.
+ */
+@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 GeneralizedLinearRegressionParams with Logging {
+
+ @Since("2.0.0")
+ def this() = this(Identifiable.randomUID("genLinReg"))
+
+ /**
+ * Set the name of family which is a description of the error
distribution
+ * to be used in the model.
+ * @group setParam
+ */
+ @Since("2.0.0")
+ def setFamily(value: String): this.type = set(family, value)
+
+ /**
+ * Set the name of the model link function.
+ * @group setParam
+ */
+ @Since("2.0.0")
+ def setLink(value: String): this.type = set(link, value)
+
+ /**
+ * Set if we should fit the intercept.
+ * Default is true.
+ * @group setParam
+ */
+ @Since("2.0.0")
+ def setFitIntercept(value: Boolean): this.type = set(fitIntercept, value)
+ setDefault(fitIntercept -> true)
+
+ /**
+ * Set the maximum number of iterations.
+ * Default is 100.
+ * @group setParam
+ */
+ @Since("2.0.0")
+ def setMaxIter(value: Int): this.type = set(maxIter, value)
+ setDefault(maxIter -> 100)
+
+ /**
+ * Set 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)
+
+ /**
+ * Set 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)
+
+ /**
+ * Whether to over-/under-sample training instances according to the
given weights in weightCol.
+ * If empty, all instances are treated equally (weight 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 -> "")
+
+ /**
+ * Set 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 familyLink = $(family) match {
+ case "gaussian" => if (isDefined(link)) Gaussian($(link)) else
Gaussian("identity")
+ case "binomial" => if (isDefined(link)) Binomial($(link)) else
Binomial("logit")
+ case "poisson" => if (isDefined(link)) Poisson($(link)) else
Poisson("log")
+ case "gamma" => if (isDefined(link)) Gamma($(link)) else
Gamma("inverse")
+ }
+
+ val numFeatures = dataset.select(col($(featuresCol))).limit(1).map {
+ case Row(features: Vector) => features.size
+ }.first()
+ if (numFeatures > 4096) {
+ val msg = "Currently, GeneralizedLinearRegression only supports
number of features" +
+ s" <= 4096. 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($(featuresCol))).map {
+ case Row(label: Double, weight: Double, features: Vector) =>
+ Instance(label, weight, features)
+ }
+
+ if ($(family) == "gaussian" && $(link) == "identity") {
+ val optimizer = new WeightedLeastSquares($(fitIntercept),
$(regParam),
+ standardizeFeatures = true, standardizeLabel = true)
+ val wlsModel = optimizer.fit(instances)
+ val model = copyValues(new GeneralizedLinearRegressionModel(uid,
+ wlsModel.coefficients, wlsModel.intercept).setParent(this))
+ return model
+ }
+
+ val newInstances = instances.map { instance =>
+ val mu = familyLink.initialize(instance.label, instance.weight)
+ val eta = familyLink.predict(mu)
+ Instance(eta, instance.weight, instance.features)
+ }
+
+ val initialModel = new WeightedLeastSquares($(fitIntercept),
$(regParam),
+ standardizeFeatures = true, standardizeLabel =
true).fit(newInstances)
+
+ val reweightFunc: (Instance, WeightedLeastSquaresModel) => (Double,
Double) = {
+ (instance: Instance, model: WeightedLeastSquaresModel) => {
+ val eta = model.predict(instance.features)
+ val mu = familyLink.fitted(eta)
+ val z = familyLink.adjusted(instance.label, mu, eta)
+ val w = familyLink.weights(mu) * instance.weight
+ (z, w)
+ }
+ }
+
+ val optimizer = new IterativelyReweightedLeastSquares(initialModel,
reweightFunc,
+ $(fitIntercept), $(regParam), $(maxIter), $(tol))
+
+ val irlsModel = optimizer.fit(instances)
+
+ val model = copyValues(new GeneralizedLinearRegressionModel(uid,
+ irlsModel.coefficients, irlsModel.intercept).setParent(this))
--- End diff --
ditto on code alignment
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
