spark-imbalance

• Stratified cross validation

  val randomForestCrossValidation = new RandomForestCrossValidation(labelIndexer, indexerPipeline, trainingData, testData)
  randomForestCrossValidation.avgMetricsParamGrid
  randomForestCrossValidation.combined

• Tree printer

TODO

  private lazy val continuousFeatures = Seq()
  private lazy val categoricalFeatures = Seq()

  val labelNames = new Labels(labelIndexer)
  val featuresNames = new Labels(
    (continuousFeatures ++ categoricalFeatures).toArray
  )
  val categories = categoricalFeatures.toArray.indices
    .map(index => index -> new Labels(categoricalFeaturesIndexer.labels)).toMap

  val treePrinter: TreePrinter = new TreePrinter(bestRandomForestModel.trees(0), labelNames, featuresNames, categories)
  val tree: String = treePrinter.toJson

• SMOTE-NC

TODO

• Useful UDFs (ensemble)

TODO