Scala名前（ラベル）を使用したランダムフォレストの重要度抽出

Question

モデルから機能の重要度を抽出し、より簡単な分析のためにfeatureColsという名前を追加する方法はありますか？

私が何かのように持っている：その後

val featureCols = Array("a","b","c".......... like 67 more) 

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 = 5043 
val 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) 

、我々は持つ重要性抽出しよう：

model.featureImportances 

をし、答えは分析するのは本当に難しいです：

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.,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... 

この答えを「盛り上げて」元のラベル名に付け加える方法はありますか？

Answer 1

元の列名はfeatureColsであり、関連するベクターは存在しないように見えるので、単純にzipの2つの配列を組み合わせることができます。入力については、このようなデータ：

val featureCols = Array("a", "b", "c", "d", "e") 
val featureImportance = Vectors.dense(Array(0.15, 0.25, 0.1, 0.35, 0.15)).toSparse 

は、単に印刷により

(d,0.35) 
(b,0.25) 
(a,0.15) 
(e,0.15) 
(c,0.1) 

になります

val res = featureCols.zip(featureImportance.toArray).sortBy(-_._2) 

を行います