2020.02.15

1.数据导入

从文件中导入数据，并转化为 DataFrame。

import org.apache.spark.ml.feature.PCA 
import org.apache.spark.sql.Row 
import org.apache.spark.ml.linalg.{Vector,Vectors} 
import org.apache.spark.ml.evaluation.MulticlassClassificationEvaluator 
import org.apache.spark.ml.{Pipeline,PipelineModel} 
import org.apache.spark.ml.feature.{IndexToString, StringIndexer, VectorIndexer,HashingTF, Tokenizer} 
import org.apache.spark.ml.classification.LogisticRegression
import org.apache.spark.ml.classification.LogisticRegressionModel 
import org.apache.spark.ml.classification.{BinaryLogisticRegressionSummary, LogisticRegression} 
import org.apache.spark.sql.functions; 
scala> import spark.implicits._
import spark.implicits._
scala> case class Adult(features: org.apache.spark.ml.linalg.Vector, label: String) <br>defined class Adult 
scala> val df=sc.textFile("adult.data.txt").map(_.split(",")).map(p=>Adult(Vectors.dense(p(0).toDouble,p(2).toDouble,p(4).toDouble,p(10).toDouble,p(11).toDouble,p(12).toDouble),p(14).toString())).toDF()<br>df:org.apache.spark.sql.DataFrame = [features: vector, label: string] 
scala> val test = sc.textFile("adult.test.txt").map(_.split(",")).map(p=>Adult(Vectors.dense(p(0).toDouble,p(2).toDouble,p(4).toDouble, p(10).toDouble, p(11).toDouble, p(12).toDouble), p(14).toString())).toDF() <br>test:org.apache.spark.sql.DataFrame = [features: vector, label: string]

2.进行主成分分析（PCA）

对 6 个连续型的数值型变量进行主成分分析。PCA（主成分分析）是通过正交变换把一组相关变量的观测值转化成一组线性无关的变量值，即主成分的一种方法。PCA 通过使用主成分把特征向量投影到低维空间，实现对特征向量的降维。请通过 setK()方法将主成分数量设置为 3，把连续型的特征向量转化成一个 3 维的主成分。

scala> val pca = new PCA().setInputCol("features").setOutputCol("pcaFeatures").setK(3).fit(df) 
scala> val result = pca.transform(df)
scala> val testdata = pca.transform(test) 
scala> result.show(false)
scala> testdata.show(false)

3.训练分类模型并预测居民收入

在主成分分析的基础上，采用逻辑斯蒂回归，或者决策树模型预测居民收入是否超过 50K；对 Test 数据集进行验证。

scala> val labelIndexer = new StringIndexer().setInputCol("label").setOutputCol("indexedLabel").fit(result) 
scala> labelIndexer.labels.foreach(println) 
scala> val featureIndexer = new VectorIndexer().setInputCol("pcaFeatures").setOutputCol("indexedFeatures").fit(result) 
scala> println(featureIndexer.numFeatures) 
scala> val labelConverter = new IndexToString().setInputCol("prediction").setOutputCol("predictedLabel").setLabels(labelIndexer. labels)
scala> val lr = new LogisticRegression().setLabelCol("indexedLabel").setFeaturesCol("indexedFeatures").setMaxIter( 100) 
scala> val lrPipeline = new Pipeline().setStages(Array(labelIndexer, featureIndexer, lr, labelConverter)) 
scala> val lrPipelineModel = lrPipeline.fit(result) 
scala> val lrModel = lrPipelineModel.stages(2).asInstanceOf[LogisticRegressionModel] 
scala> println("Coefficients: " + lrModel.coefficientMatrix+"Intercept: "+lrModel.interceptVector+"numClasses: "+lrModel.numClasses+"numFeatures: "+lrModel.numFeatures)
scala> val lrPredictions = lrPipelineModel.transform(testdata) 
scala> val evaluator = new MulticlassClassificationEvaluator().setLabelCol("indexedLabel").setPredictionCol("prediction") 
scala> val lrAccuracy = evaluator.evaluate(lrPredictions)
scala> println("Test Error = " + (1.0 - lrAccuracy))

4.超参数调优

利用 CrossValidator 确定最优的参数，包括最优主成分 PCA 的维数、分类器自身的参数等。

scala> val pca = new PCA().setInputCol("features").setOutputCol("pcaFeatures") 
scala> val labelIndexer = new StringIndexer().setInputCol("label").setOutputCol("indexedLabel").fit(df)
scala> val featureIndexer = new VectorIndexer().setInputCol("pcaFeatures").setOutputCol("indexedFeatures") 
scala> val labelConverter = new IndexToString().setInputCol("prediction").setOutputCol("predictedLabel").setLabels(labelIndexer.l abels) 
scala> val lr = new LogisticRegression().setLabelCol("indexedLabel").setFeaturesCol("indexedFeatures").setMaxIter(1 00)
scala> val lrPipeline = new Pipeline().setStages(Array(pca, labelIndexer, featureIndexer, lr, labelConverter)) 
scala> val paramGrid = new ParamGridBuilder().addGrid(pca.k, Array(1,2,3,4,5,6)).addGrid(lr.elasticNetParam, Array(0.2,0.8)).addGrid(lr.regParam, Array(0.01, 0.1, 0.5)).build()
scala> val cv = new CrossValidator().setEstimator(lrPipeline).setEvaluator(new MulticlassClassificationEvaluator().setLabelCol("indexedLabel").setPredictionCol("prediction")).se tEstimatorParamMaps(paramGrid).setNumFolds(3) 
scala> val cvModel = cv.fit(df) 
scala> val lrPredictions=cvModel.transform(test) 
scala> val evaluator = new MulticlassClassificationEvaluator().setLabelCol("indexedLabel").setPredictionCol("prediction") 
scala> val lrAccuracy = evaluator.evaluate(lrPredictions) 
scala> println("准确率为"+lrAccuracy) 
scala> val bestModel= cvModel.bestModel.asInstanceOf[PipelineModel] 
scala> val lrModel = bestModel.stages(3).asInstanceOf[LogisticRegressionModel] 
scala> println("Coefficients: " + lrModel.coefficientMatrix + "Intercept: "+lrModel.interceptVector+ "numClasses: "+lrModel.numClasses+"numFeatures: "+lrModel.numFeatures) 
scala> val pcaModel = bestModel.stages(0).asInstanceOf[PCAModel] 
scala> println("Primary Component: " + pcaModel.pc)