Spark 机器学习

将Mahout on Spark 中的机器学习算法和MLlib中支持的算法统计如下：

主要针对MLlib进行总结

分类与回归

分类和回归是监督式学习;

监督式学习是指使用有标签的数据（LabeledPoint）进行训练，得到模型后，使用测试数据预测结果。其中标签数据是指已知结果的特征数据。

分类和回归的区别：预测结果的变量类型

　　分类预测出来的变量是离散的（比如对邮件的分类，垃圾邮件和非垃圾邮件），对于二元分类的标签是0和1，对于多元分类标签范围是0~C-1,C表示类别数目；

　　回归预测出来的变量是连续的（比如根据年龄和体重预测身高）

线性回归

　　线性回归是回归中最常用的方法之一，是指用特征的线性组合来预测输出值。

　　线性回归算法可以使用的类有:

　　　　LinearRegressionWithSGD
　　　　RidgeRegressionWithSGD
　　　　LassoWithSGD

　　　　ridge regression 使用 L2 正规化;
　　　　Lasso 使用 L1 正规化;

　　参数：

　　　　stepSize:梯度下降的步数

　　　　numIterations:迭代次数

　　　　设置intercept:是否给数据加上一个干扰特征或者偏差特征，一个始终值为1的特征，默认不增加false

　　{stepSize: 1.0, numIterations: 100, miniBatchFraction: 1.0}

　　模型的使用：

　　　　1、对数据进行预测,使用model.predict()

　　　　2、获取数据特征的权重model.weights()

　　模型的评估：

　　　　均方误差

例子：

import org.apache.spark.{SparkContext, SparkConf}
import org.apache.spark.mllib.regression.LabeledPoint
import org.apache.spark.mllib.regression.LinearRegressionModel
import org.apache.spark.mllib.regression.LinearRegressionWithSGD
import org.apache.spark.mllib.linalg.Vectors

/**
  * Created by Edward on 2016/9/21.
  */
object LinearRegression {
  def main(args: Array[String]) {
    val conf: SparkConf = new SparkConf().setAppName("LinearRegression").setMaster("local")
    val sc = new SparkContext(conf)

    // Load and parse the data
    val data = sc.textFile("data/mllib/ridge-data/lpsa.data")
    val parsedData = data.map { line =>
      val parts = line.split(',')
      LabeledPoint(parts(0).toDouble, Vectors.dense(parts(1).split(' ').map(_.toDouble)))
    }.cache()

    // Building the model
    val numIterations = 100
    val model = LinearRegressionWithSGD.train(parsedData, numIterations)
//    var lr = new LinearRegressionWithSGD().setIntercept(true)
//    val model = lr.run(parsedData)

    //获取特征权重，及干扰特征
    println("weights:%s, intercept:%s".format(model.weights,model.intercept))

    // Evaluate model on training examples and compute training error
    val valuesAndPreds = parsedData.map { point =>
      val prediction = model.predict(point.features)
      (point.label, prediction)
    }

    //计算 均方误差
    val MSE = valuesAndPreds.map{case(v, p) => math.pow((v - p), 2)}.mean()
    println("training Mean Squared Error = " + MSE)

    // Save and load model
    model.save(sc, "myModelPath")
    val sameModel = LinearRegressionModel.load(sc, "myModelPath")

  }
}

数据：

-0.4307829,-1.63735562648104 -2.00621178480549 -1.86242597251066 -1.02470580167082 -0.522940888712441 -0.863171185425945 -1.04215728919298 -0.864466507337306
-0.1625189,-1.98898046126935 -0.722008756122123 -0.787896192088153 -1.02470580167082 -0.522940888712441 -0.863171185425945 -1.04215728919298 -0.864466507337306
-0.1625189,-1.57881887548545 -2.1887840293994 1.36116336875686 -1.02470580167082 -0.522940888712441 -0.863171185425945 0.342627053981254 -0.155348103855541
-0.1625189,-2.16691708463163 -0.807993896938655 -0.787896192088153 -1.02470580167082 -0.522940888712441 -0.863171185425945 -1.04215728919298 -0.864466507337306
0.3715636,-0.507874475300631 -0.458834049396776 -0.250631301876899 -1.02470580167082 -0.522940888712441 -0.863171185425945 -1.04215728919298 -0.864466507337306
0.7654678,-2.03612849966376 -0.933954647105133 -1.86242597251066 -1.02470580167082 -0.522940888712441 -0.863171185425945 -1.04215728919298 -0.864466507337306
...

数据第一列表示标签数据，也就是结果数据，其他列表示特征数据；

预测就是再给一组特征数据，预测结果；

结果：

weights:[0.5808575763272221,0.18930001482946976,0.2803086929991066,0.1110834181777876,0.4010473965597895,-0.5603061626684255,-0.5804740464000981,0.8742741176970946], intercept:0.0
training Mean Squared Error = 6.207597210613579

逻辑回归 

是一种二元分类方法，也是多类分类方法；

　　逻辑回归可以使用的方法：

　　LogisticRegressionWithLBFGS (建议使用这个)

　　LogisticRegressionWithSGD

　　参数：

　　与线性回归类似

　　模型的使用：

　　1、对数据进行预测,使用model.predict()
　　2、获取数据特征的权重model.weights()

　　模型的评估：

　　二元分类：AUC(Area Under roc Curve)

　　

import org.apache.spark.{SparkContext, SparkConf}

import org.apache.spark.SparkContext
import org.apache.spark.mllib.classification.{LogisticRegressionWithLBFGS, LogisticRegressionModel}
import org.apache.spark.mllib.evaluation.{BinaryClassificationMetrics, MulticlassMetrics}
import org.apache.spark.mllib.regression.LabeledPoint
import org.apache.spark.mllib.util.MLUtils

/**
  * Created by Edward on 2016/9/21.
  */
object LogisticRegression {
  def main(args: Array[String]) {
    val conf: SparkConf = new SparkConf().setAppName("LogisticRegression").setMaster("local")
    val sc: SparkContext = new SparkContext(conf)

    // Load training data in LIBSVM format.
    val data = MLUtils.loadLibSVMFile(sc, "data/mllib/sample_libsvm_data.txt")

    // Split data into training (60%) and test (40%).
    val splits = data.randomSplit(Array(0.6, 0.4), seed = 11L)
    val training = splits(0).cache()
    val test = splits(1)

    // Run training algorithm to build the model
    val model = new LogisticRegressionWithLBFGS()
      .setNumClasses(10)
      .run(training)

    model.setThreshold(0.8)

    // Compute raw scores on the test set.
    val predictionAndLabels = test.map { case LabeledPoint(label, features) =>
      val prediction = model.predict(features)
      (prediction, label)
    }

    //多元矩阵
    // Get evaluation metrics.
    //val metrics = new MulticlassMetrics(predictionAndLabels)
    //val precision = metrics.precision
    //println("Precision = " + precision)

    //二元矩阵
    val metrics = new BinaryClassificationMetrics(predictionAndLabels)
    //通过ROC对模型进行评估,值趋近于1 receiver operating characteristic (ROC), 接受者操作特征 曲线下面积
    val auROC: Double = metrics.areaUnderROC()
    println("Area under ROC = " + auROC)
    //通过PR对模型进行评估，值趋近于1 precision-recall (PR), 精确率
    val underPR: Double = metrics.areaUnderPR()
    println("Area under PR = " + underPR)

    // Save and load model
    model.save(sc, "myModelPath")
    val sameModel = LogisticRegressionModel.load(sc, "myModelPath")

  }

}

支持向量机 Support Vector Machines (SVMs) 

　　分类算法，二元分类算法

　　和逻辑回归二元分类相似

import org.apache.spark.{SparkContext, SparkConf}
import org.apache.spark.mllib.classification.{SVMModel, SVMWithSGD}
import org.apache.spark.mllib.evaluation.BinaryClassificationMetrics
import org.apache.spark.mllib.util.MLUtils

/**
  * Created by Edward on 2016/9/21.
  */
object SVMs {
  def main(args: Array[String]) {

    val conf: SparkConf = new SparkConf().setAppName("SVM").setMaster("local")
    val sc: SparkContext = new SparkContext(conf)

    // Load training data in LIBSVM format.
    val data = MLUtils.loadLibSVMFile(sc, "data/mllib/sample_libsvm_data.txt")

    // Split data into training (60%) and test (40%).
    val splits = data.randomSplit(Array(0.6, 0.4), seed = 11L)
    val training = splits(0).cache()
    val test = splits(1)

    // Run training algorithm to build the model
    val numIterations = 100
    val model = SVMWithSGD.train(training, numIterations)

    // Clear the default threshold.
    model.clearThreshold()

    // Compute raw scores on the test set.
    val scoreAndLabels = test.map { point =>
      println("feature="+point.features)
      val score = model.predict(point.features)
      (score, point.label)
    }

    scoreAndLabels.foreach(println(_))

    // Get evaluation metrics.
    val metrics = new BinaryClassificationMetrics(scoreAndLabels)

    println("metrics="+metrics)

    val auROC = metrics.areaUnderROC()

    println("Area under ROC = " + auROC)

    // Save and load model
    model.save(sc, "myModelPath")
    val sameModel = SVMModel.load(sc, "myModelPath")

    sc.stop()

  }

}

数据：

0 128:51 129:159 130:253 131:159 132:50 155:48 156:238 157:252 158:252 159:252 160:237 182:54 183:227 184:253 185:252 186:239 187:233 188:252 189:57 190:6 208:10 209:60 210:224
1 159:124 160:253 161:255 162:63 186:96 187:244 188:251 189:253 190:62 214:127 215:251 216:251 217:253 218:62 
...

协同过滤 Collaborative Filtering

　　Spark中协同过滤算法主要由交替最小二乘法来实现 alternating least squares (ALS)

　　参数：

　　numBlocks block块的数量，用来控制并行度

　　rank 特征向量的大小

　　iterations 迭代数量

　　lambda 正规化参数

　　alpha 用来在隐式ALS中计算置信度的常量

　　方法：

　　ALS.train

　　模型的评估：

　　均方误差

例子：

　　

import org.apache.spark.{SparkContext, SparkConf}
import org.apache.spark.mllib.recommendation.ALS
import org.apache.spark.mllib.recommendation.MatrixFactorizationModel
import org.apache.spark.mllib.recommendation.Rating

/**
  * Created by Edward on 2016/9/22.
  */
object CollaborativeALS {
  def main(args: Array[String]) {

    val conf: SparkConf = new SparkConf().setAppName("CollaborativeALS").setMaster("local")
    val sc: SparkContext = new SparkContext(conf)
    // Load and parse the data
    val data = sc.textFile("data/mllib/als/test.data")
    val ratings = data.map(_.split(',') match { case Array(user, item, rate) =>
      Rating(user.toInt, item.toInt, rate.toDouble)
    })

    // Build the recommendation model using ALS
    val rank = 10
    val numIterations = 10
    val model = ALS.train(ratings, rank, numIterations, 0.01)

    // Evaluate the model on rating data
    val usersProducts = ratings.map { case Rating(user, product, rate) =>
      (user, product)
    }
    val predictions =
      model.predict(usersProducts).map { case Rating(user, product, rate) =>
        ((user, product), rate)
      }
    val ratesAndPreds = ratings.map { case Rating(user, product, rate) =>
      ((user, product), rate)
    }.join(predictions)
    val MSE = ratesAndPreds.map { case ((user, product), (r1, r2)) =>
      val err = (r1 - r2)
      err * err
    }.mean()

    //均方误差
    println("Mean Squared Error = " + MSE)

    // Save and load model
    model.save(sc, "target/tmp/myCollaborativeFilter")
    val sameModel = MatrixFactorizationModel.load(sc, "target/tmp/myCollaborativeFilter")

  }

}

持续更新中...