QuoraQuestionsPairsPipeline.scala

val logisticRegression: Param[LogisticRegression] =
    new Param(this, "logisticRegression", "Combine question vectors pairs into a predicted probability.")
  setDefault(logisticRegression, new LogisticRegression())
private def logisticRegressionPipeline(): Array[PipelineStage] = {
  val labelCol = "isDuplicateLabel"
  val assembler = new VectorAssembler().setInputCols(questions("lda")).setOutputCol("mergedlda")
  val labeler = new SQLTransformer().setStatement(
    s"SELECT *, cast(isDuplicate as int) $labelCol from __THIS__")
  val lr = $(logisticRegression)
    .setFeaturesCol("mergedlda").setProbabilityCol("p").setRawPredictionCol("raw")
    .setLabelCol(labelCol)
  Array(assembler, labeler, lr)
}