有没有办法从模型中提取特征重要性，并附加featureCols名称以便于分析？
我有类似这样的代码：
val featureCols = Array("a","b","c".......... 还有67个)val assembler = new VectorAssembler().setInputCols(featureCols).setOutputCol("features")val df2 = assembler.transform(modeling_db)val labelIndexer = new StringIndexer().setInputCol("def").setOutputCol("label")val df3 = labelIndexer.fit(df2).transform(df2)val splitSeed = 5043val Array(trainingData, testDataCE) = df3.randomSplit(Array(0.7, 0.3), splitSeed)val classifier = new RandomForestClassifier().setImpurity("gini").setMaxDepth(19).setNumTrees(57).setFeatureSubsetStrategy("auto").setSeed(5043)val model = classifier.fit(trainingData)之后，我们尝试提取重要性，使用：
model.featureImportances但结果很难分析：
res14: org.apache.spark.mllib.linalg.Vector = (71,[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,20,23,25,27,33,34,35,38,39,41,42,45,47,48,49,50,51,52,53,54,55,56,57,58,60,61,62,63,64,65,66,67,68,69,70],[0.22362951804309808,0.1830148359365108,0.10246542303449771,0.1699399958851977,0.06486419413350401,0.05187244974385025,0.02627047699833213,0.014498050071723645,0.026182513062665076,0.007126662761055224,0.012375060477018274,0.004354513006816487,0.004361008357237427,0.008435852744278544,0.003195472326415685,0.0023071401643885753,0.004602370417578224,0.0030394399903992345,6.92408316823549E-4,0.011207695216651398,7.609910745572573E-4,8.316382113306638E-4,0.0021506289318167916,0.0013468620354363688,0.006968754359778437,0.018796331618729723,0.0024516591941419444,0.005980997035580654,0.0027983...有没有办法“解包”这个结果并将其附加到原始标签名称上？
回答：
您在featureCols中有原始列名，看起来没有涉及任何向量，因此您可以简单地将这两个数组zip在一起。对于这样的输入数据：
val featureCols = Array("a", "b", "c", "d", "e")val featureImportance = Vectors.dense(Array(0.15, 0.25, 0.1, 0.35, 0.15)).toSparse只需执行
val res = featureCols.zip(featureImportance.toArray).sortBy(-_._2)打印结果将是
(d,0.35)(b,0.25)(a,0.15)(e,0.15)(c,0.1)