scala - 如何访问由 RandomForestClassifier (spark.ml-version) 创建的模型中的单个树？

Question

如何访问 Spark ML 生成的模型中的单个树 RandomForestClassifier ?我正在使用 Scala 版本的 RandomForestClassifier。

Answer 1

其实它有trees属性:

import org.apache.spark.ml.attribute.NominalAttribute
import org.apache.spark.ml.classification.{
  RandomForestClassificationModel, RandomForestClassifier, 
  DecisionTreeClassificationModel
}

val meta = NominalAttribute
  .defaultAttr
  .withName("label")
  .withValues("0.0", "1.0")
  .toMetadata

val data = sqlContext.read.format("libsvm")
  .load("data/mllib/sample_libsvm_data.txt")
  .withColumn("label", $"label".as("label", meta))

val rf: RandomForestClassifier = new RandomForestClassifier()
  .setLabelCol("label")
  .setFeaturesCol("features")

val trees: Array[DecisionTreeClassificationModel] = rf.fit(data).trees.collect {
  case t: DecisionTreeClassificationModel => t
}

正如你所看到的，唯一的问题是让类型正确，这样我们就可以实际使用这些:

trees.head.transform(data).show(3)
// +-----+--------------------+-------------+-----------+----------+
// |label|            features|rawPrediction|probability|prediction|
// +-----+--------------------+-------------+-----------+----------+
// |  0.0|(692,[127,128,129...|   [33.0,0.0]|  [1.0,0.0]|       0.0|
// |  1.0|(692,[158,159,160...|   [0.0,59.0]|  [0.0,1.0]|       1.0|
// |  1.0|(692,[124,125,126...|   [0.0,59.0]|  [0.0,1.0]|       1.0|
// +-----+--------------------+-------------+-----------+----------+
// only showing top 3 rows

备注 :

如果您使用管道，您也可以提取单个树:

import org.apache.spark.ml.Pipeline

val model = new Pipeline().setStages(Array(rf)).fit(data)

// There is only one stage and know its type 
// but lets be thorough
val rfModelOption = model.stages.headOption match {
  case Some(m: RandomForestClassificationModel) => Some(m)
  case _ => None
}

val trees = rfModelOption.map {
  _.trees //  ... as before
}.getOrElse(Array())