[[email protected] aas]# ls ch02 ch03 spark-1.2.1-bin-hadoop2.4 spark-1.2.1-bin-hadoop2.4.tgz [[email protected] aas]# cd spark-1.2.1-bin-hadoop2.4 [[email protected] spark-1.2.1-bin-hadoop2.4]# cd .. [[email protected] aas]# mkdir ch04 [[email protected] aas]# cd ch04 [[email protected] ch04]# ls [[email protected] ch04]# wget https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/covtype.data.gz --2015-12-06 08:52:34-- https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/covtype.data.gz Resolving archive.ics.uci.edu... 128.195.10.249 Connecting to archive.ics.uci.edu|128.195.10.249|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 11240707 (11M) [application/x-gzip] Saving to: ?.ovtype.data.gz? 100%[===============================================================================================================================================================>] 11,240,707 2.62M/s in 4.2s 2015-12-06 08:52:39 (2.53 MB/s) - ?.ovtype.data.gz?.saved [11240707/11240707] [[email protected] ch04]# wget https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/covtype.info --2015-12-06 08:53:00-- https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/covtype.info Resolving archive.ics.uci.edu... 128.195.10.249 Connecting to archive.ics.uci.edu|128.195.10.249|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 14610 (14K) [text/plain] Saving to: ?.ovtype.info? 100%[===============================================================================================================================================================>] 14,610 --.-K/s in 0.001s 2015-12-06 08:53:01 (15.6 MB/s) - ?.ovtype.info?.saved [14610/14610] [[email protected] ch04]# wget https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/old_covtype.info --2015-12-06 08:53:25-- https://archive.ics.uci.edu/ml/machine-learning-databases/covtype/old_covtype.info Resolving archive.ics.uci.edu... 128.195.10.249 Connecting to archive.ics.uci.edu|128.195.10.249|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 4847 (4.7K) [text/plain] Saving to: ?.ld_covtype.info? 100%[===============================================================================================================================================================>] 4,847 --.-K/s in 0s 2015-12-06 08:53:26 (12.7 MB/s) - ?.ld_covtype.info?.saved [4847/4847]
将数据放到HDFS上
[[email protected] ch04]# ls covtype.data.gz covtype.info old_covtype.info [[email protected] ch04]# gunzip -d covtype.data.gz [[email protected] ch04]# ll total 73432 -rw-r--r-- 1 root root 75169317 Sep 1 1998 covtype.data -rw-r--r-- 1 root root 14610 Apr 18 2010 covtype.info -rw-r--r-- 1 root root 4847 Sep 1 1998 old_covtype.info [[email protected] ch04]# hdfs dfs -mkdir /user/root/covtype [[email protected] ch04]# hdfs dfs -put * /user/root/covtype [[email protected] ch04]# hdfs dfs -ls /user/root/covtype Found 3 items -rw-r--r-- 3 root supergroup 75169317 2015-12-06 09:02 /user/root/covtype/covtype.data -rw-r--r-- 3 root supergroup 14610 2015-12-06 09:02 /user/root/covtype/covtype.info -rw-r--r-- 3 root supergroup 4847 2015-12-06 09:02 /user/root/covtype/old_covtype.info
启动spark-shell
[[email protected] ch04]# ../spark-1.2.1-bin-hadoop2.4/bin/spark-shell --master yarn-client Welcome to ____ __ / __/__ ___ _____/ /__ _\ \/ _ \/ _ `/ __/ ‘_/ /___/ .__/\_,_/_/ /_/\_\ version 1.2.1 /_/ Using Scala version 2.10.4 (OpenJDK 64-Bit Server VM, Java 1.7.0_09-icedtea) Type in expressions to have them evaluated. Type :help for more information. 15/12/06 09:08:05 WARN NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable Spark context available as sc.
运行代码4.7节代码:
scala> import org.apache.spark.mllib.linalg._ import org.apache.spark.mllib.linalg._ scala> import org.apache.spark.mllib.regression._ import org.apache.spark.mllib.regression._ scala> scala> val rawData = sc.textFile("hdfs:///user/ds/covtype.data" ) rawData: org.apache.spark.rdd.RDD[String] = hdfs:///user/ds/covtype.data MappedRDD[1] at textFile at <console>:18 scala> val data = rawData.map { line => | val values = line.split(‘,‘ ).map(_. toDouble) | val featureVector = Vectors.dense(values. init) | val label = values.last - 1 | LabeledPoint(label, featureVector) | } data: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint] = MappedRDD[2] at map at <console>:20 scala> val Array(trainData, cvData, testData) = data. randomSplit(Array(0.8, 0.1, 0.1)) trainData: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint] = PartitionwiseSampledRDD[3] at randomSplit at <console>:22 cvData: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint] = PartitionwiseSampledRDD[4] at randomSplit at <console>:22 testData: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint] = PartitionwiseSampledRDD[5] at randomSplit at <console>:22 scala> trainData.cache() res0: trainData.type = PartitionwiseSampledRDD[3] at randomSplit at <console>:22 scala> cvData.cache() res1: cvData.type = PartitionwiseSampledRDD[4] at randomSplit at <console>:22 scala> testData.cache() res2: testData.type = PartitionwiseSampledRDD[5] at randomSplit at <console>:22
scala> import org.apache.spark.mllib.evaluation._
import org.apache.spark.mllib.evaluation._
scala> import org.apache.spark.mllib.tree._
import org.apache.spark.mllib.tree._
scala> import org.apache.spark.mllib.tree.model._
import org.apache.spark.mllib.tree.model._
scala> import org.apache.spark.rdd._
import org.apache.spark.rdd._
scala> def getMetrics(model: DecisionTreeModel, data: RDD[LabeledPoint]):
| MulticlassMetrics = {
| val predictionsAndLabels = data. map(example =>(model. predict(example. features), example. label))
| new MulticlassMetrics(predictionsAndLabels)
| }
getMetrics: (model: org.apache.spark.mllib.tree.model.DecisionTreeModel, data: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint])org.apache.spark.mllib.evaluation.MulticlassMetrics
scala> val model = DecisionTree.trainClassifier(trainData, 7, Map[Int, Int](), "gini" , 4, 100)
model: org.apache.spark.mllib.tree.model.DecisionTreeModel = DecisionTreeModel classifier of depth 4 with 31 nodes
scala> val metrics = getMetrics(model, cvData)
metrics: org.apache.spark.mllib.evaluation.MulticlassMetrics = [email protected]
scala> metrics. confusionMatrix
res6: org.apache.spark.mllib.linalg.Matrix =
15535.0 5345.0 21.0 0.0 0.0 0.0 392.0
6669.0 20855.0 688.0 0.0 5.0 0.0 47.0
0.0 610.0 2942.0 0.0 0.0 0.0 0.0
0.0 0.0 274.0 0.0 0.0 0.0 0.0
12.0 874.0 57.0 0.0 15.0 0.0 0.0
0.0 446.0 1318.0 0.0 0.0 0.0 0.0
1150.0 19.0 8.0 0.0 0.0 0.0 905.0
scala> metrics.precision
res7: Double = 0.6917696392665028
scala> metrics.precision
res7: Double = 0.6917696392665028
scala> (0 until 7).map(
| cat => (metrics.precision(cat), metrics.recall(cat))
| ).foreach(println)
(0.6648549174013524,0.729582491898746)
(0.7408788944545099,0.7378644211718086)
(0.554257724189902,0.8282657657657657)
(0.0,0.0)
(0.75,0.015657620041753653)
(0.0,0.0)
(0.6733630952380952,0.4346781940441883)
scala> import org.apache.spark.rdd._
import org.apache.spark.rdd._
scala> def classProbabilities(data: RDD[LabeledPoint]): Array[Double] = {
| val countsByCategory = data.map(_.label).countByValue()
| val counts = countsByCategory.toArray.sortBy(_. _1).map(_. _2)
| counts.map(_.toDouble / counts.sum)
| }
classProbabilities: (data: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint])Array[Double]
scala> val trainPriorProbabilities = classProbabilities(trainData)
trainPriorProbabilities: Array[Double] = Array(0.3644680841907762, 0.48778063233452534, 0.06163475731247069, 0.004682046846288574, 0.0163893156379504, 0.029860958700732, 0.035184204977256786)
scala> val cvPriorProbabilities = classProbabilities(cvData)
cvPriorProbabilities: Array[Double] = Array(0.36594084589341264, 0.4857442384037672, 0.061044563218588345, 0.004708955608641105, 0.016464158660869265, 0.03031604997679894, 0.03578118823792256)
scala> trainPriorProbabilities.zip(cvPriorProbabilities).map {
| case (trainProb, cvProb) => trainProb * cvProb
| }.sum
res9: Double = 0.3765289404519721
scala> val evaluations =
| for (impurity <- Array("gini" , "entropy" );
| depth <- Array(1, 20);
| bins <- Array(10, 300))
| yield {
| val model = DecisionTree. trainClassifier(trainData, 7, Map[Int, Int](), impurity, depth, bins)
| val predictionsAndLabels = cvData. map(example =>(model. predict(example. features), example. label))
| val accuracy = new MulticlassMetrics(predictionsAndLabels). precision
| ((impurity, depth, bins), accuracy)
| }
evaluations: Array[((String, Int, Int), Double)] = Array(((gini,1,10),0.6319968377816351), ((gini,1,300),0.6323577431385017), ((gini,20,10),0.889253613350061), ((gini,20,300),0.9074191829790159), ((entropy,1,10),0.4857442384037672), ((entropy,1,300),0.4857442384037672), ((entropy,20,10),0.8946500077336862), ((entropy,20,300),0.9099455204770825))
scala> evaluations.sortBy(_. _2).reverse.foreach(println)
((entropy,20,300),0.9099455204770825)
((gini,20,300),0.9074191829790159)
((entropy,20,10),0.8946500077336862)
((gini,20,10),0.889253613350061)
((gini,1,300),0.6323577431385017)
((gini,1,10),0.6319968377816351)
((entropy,1,300),0.4857442384037672)
((entropy,1,10),0.4857442384037672)
scala> val data = rawData.map { line =>
| val values = line.split(‘,‘ ).map(_.toDouble)
| val wilderness = values.slice(10, 14).indexOf(1.0).toDouble
| val soil = values.slice(14, 54).indexOf(1.0).toDouble
| val featureVector =
| Vectors.dense(values.slice(0, 10) :+ wilderness :+ soil)
| val label = values.last - 1
| LabeledPoint(label, featureVector)
| }
data: org.apache.spark.rdd.RDD[org.apache.spark.mllib.regression.LabeledPoint] = MappedRDD[391] at map at <console>:47
scala>
scala> val evaluations =
| for (impurity <- Array("gini" , "entropy" );
| depth <- Array(10, 20, 30);
| bins <- Array(40, 300))
| yield {
| val model = DecisionTree. trainClassifier(trainData, 7, Map(10 -> 4, 11 -> 40),impurity, depth, bins)
| val trainAccuracy = getMetrics(model, trainData). precision
| val cvAccuracy = getMetrics(model, cvData). precision
| ((impurity, depth, bins), (trainAccuracy, cvAccuracy))
| }
evaluations: Array[((String, Int, Int), (Double, Double))] = Array(((gini,10,40),(0.7772542032989496,0.7730420884389984)), ((gini,10,300),(0.7849615065174265,0.7793665251688522)), ((gini,20,40),(0.9393033733975393,0.904480382215959)), ((gini,20,300),(0.9421715574260792,0.904480382215959)), ((gini,30,40),(0.9972329447406585,0.9341089934177738)), ((gini,30,300),(0.9974352022790551,0.9347964321927579)), ((entropy,10,40),(0.7768755083334409,0.7716672108890302)), ((entropy,10,300),(0.7715307452975122,0.7655318198222971)), ((entropy,20,40),(0.9487578374796128,0.9103407977726984)), ((entropy,20,300),(0.9484781196073622,0.9088971763452317)), ((entropy,30,40),(0.998582045555283,0.9374430714764467)), ((entropy,30,300),(0.9990833860493938,0.9413786584632307)))
scala> val forest = RandomForest. trainClassifier(
| trainData, 7, Map(10 -> 4, 11 -> 40), 20,
| "auto" , "entropy" , 30, 300)
forest: org.apache.spark.mllib.tree.model.RandomForestModel =
TreeEnsembleModel classifier with 20 trees