QuoraQuestionsPairsPipeline.scala

private def assemblePipeline(): Pipeline = {
  val stages = Array(
    cleanerPipeline(),
    tokenizePipeline(),
    vectorizePipeline(),
    ldaPipeline(),
    logisticRegressionPipeline()
  ).flatten
  new Pipeline().setStages(stages)
}
override def fit(dataset: Dataset[_]): PipelineModel = {
  logger.info(s"Preparing to fit quora question pipeline with params:\n${explainParams()}")
  val model = assemblePipeline().fit(dataset)
  logMetrics(model, dataset)
  logTopics(dataset.sparkSession, model)
  model
}