spark LocalALS 源码

spark LocalALS 代码

文件路径：/examples/src/main/scala/org/apache/spark/examples/LocalALS.scala

/*
 * 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.
 */

// scalastyle:off println
package org.apache.spark.examples

import org.apache.commons.math3.linear._

/**
 * Alternating least squares matrix factorization.
 *
 * This is an example implementation for learning how to use Spark. For more conventional use,
 * please refer to org.apache.spark.ml.recommendation.ALS.
 */
object LocalALS {

  // Parameters set through command line arguments
  var M = 0 // Number of movies
  var U = 0 // Number of users
  var F = 0 // Number of features
  var ITERATIONS = 0
  val LAMBDA = 0.01 // Regularization coefficient

  def generateR(): RealMatrix = {
    val mh = randomMatrix(M, F)
    val uh = randomMatrix(U, F)
    mh.multiply(uh.transpose())
  }

  def rmse(targetR: RealMatrix, ms: Array[RealVector], us: Array[RealVector]): Double = {
    val r = new Array2DRowRealMatrix(M, U)
    for (i <- 0 until M; j <- 0 until U) {
      r.setEntry(i, j, ms(i).dotProduct(us(j)))
    }
    val diffs = r.subtract(targetR)
    var sumSqs = 0.0
    for (i <- 0 until M; j <- 0 until U) {
      val diff = diffs.getEntry(i, j)
      sumSqs += diff * diff
    }
    math.sqrt(sumSqs / (M.toDouble * U.toDouble))
  }

  def updateMovie(i: Int, m: RealVector, us: Array[RealVector], R: RealMatrix) : RealVector = {
    var XtX: RealMatrix = new Array2DRowRealMatrix(F, F)
    var Xty: RealVector = new ArrayRealVector(F)
    // For each user that rated the movie
    for (j <- 0 until U) {
      val u = us(j)
      // Add u * u^t to XtX
      XtX = XtX.add(u.outerProduct(u))
      // Add u * rating to Xty
      Xty = Xty.add(u.mapMultiply(R.getEntry(i, j)))
    }
    // Add regularization coefficients to diagonal terms
    for (d <- 0 until F) {
      XtX.addToEntry(d, d, LAMBDA * U)
    }
    // Solve it with Cholesky
    new CholeskyDecomposition(XtX).getSolver.solve(Xty)
  }

  def updateUser(j: Int, u: RealVector, ms: Array[RealVector], R: RealMatrix) : RealVector = {
    var XtX: RealMatrix = new Array2DRowRealMatrix(F, F)
    var Xty: RealVector = new ArrayRealVector(F)
    // For each movie that the user rated
    for (i <- 0 until M) {
      val m = ms(i)
      // Add m * m^t to XtX
      XtX = XtX.add(m.outerProduct(m))
      // Add m * rating to Xty
      Xty = Xty.add(m.mapMultiply(R.getEntry(i, j)))
    }
    // Add regularization coefficients to diagonal terms
    for (d <- 0 until F) {
      XtX.addToEntry(d, d, LAMBDA * M)
    }
    // Solve it with Cholesky
    new CholeskyDecomposition(XtX).getSolver.solve(Xty)
  }

  def showWarning(): Unit = {
    System.err.println(
      """WARN: This is a naive implementation of ALS and is given as an example!
        |Please use org.apache.spark.ml.recommendation.ALS
        |for more conventional use.
      """.stripMargin)
  }

  def main(args: Array[String]): Unit = {

    args match {
      case Array(m, u, f, iters) =>
        M = m.toInt
        U = u.toInt
        F = f.toInt
        ITERATIONS = iters.toInt
      case _ =>
        System.err.println("Usage: LocalALS <M> <U> <F> <iters>")
        System.exit(1)
    }

    showWarning()

    println(s"Running with M=$M, U=$U, F=$F, iters=$ITERATIONS")

    val R = generateR()

    // Initialize m and u randomly
    var ms = Array.fill(M)(randomVector(F))
    var us = Array.fill(U)(randomVector(F))

    // Iteratively update movies then users
    for (iter <- 1 to ITERATIONS) {
      println(s"Iteration $iter:")
      ms = (0 until M).map(i => updateMovie(i, ms(i), us, R)).toArray
      us = (0 until U).map(j => updateUser(j, us(j), ms, R)).toArray
      println(s"RMSE = ${rmse(R, ms, us)}")
    }
  }

  private def randomVector(n: Int): RealVector =
    new ArrayRealVector(Array.fill(n)(math.random))

  private def randomMatrix(rows: Int, cols: Int): RealMatrix =
    new Array2DRowRealMatrix(Array.fill(rows, cols)(math.random))

}
// scalastyle:on println

相关信息

spark 源码目录

相关文章

spark AccumulatorMetricsTest 源码

spark BroadcastTest 源码

spark DFSReadWriteTest 源码

spark DriverSubmissionTest 源码

spark ExceptionHandlingTest 源码

spark GroupByTest 源码

spark HdfsTest 源码

spark LocalFileLR 源码

spark LocalKMeans 源码

spark LocalLR 源码