Transformer

Transformers

A transformer is used to turn one or multiple datasets
into another one. A Transformer accept a configuration class as a parameter a configuration class of type TransformerConfig. This configuration file control the behaviour of the Transformer through its methods and values.

transform is the main and the unique public available method of Transformer, this method accepts one or several datasets as input and after the application of the transformation logic, returns a dataset of type Event(Events) with a new category.

If we take as an example ExposureTransformer, we pass an instance of the class ExposuresTransformerConfig as parameter, it extends a TransformerConfig.

/**
 * A tag to improve readability of subclasses and to allow type binding
 */

trait TransformerConfig 

The config class is used to control the Transformer's behaviour.

class ExposuresTransformerConfig(
 val exposurePeriodAdder: ExposurePeriodAdder) extends TransformerConfig with Serializable

This main method take as parameter two datasets of type Event and subtypes FollowUp and Drug respectively and return a dataset of type Event and subtype Exposure.

  def transform(followUps: Dataset[Event[FollowUp]])(drugs: Dataset[Event[Drug]]): Dataset[Event[Exposure]] = {
    drugs
      .transform(config.exposurePeriodAdder.toExposure(followUps))
      .transform(regulateWithFollowUps(followUps))
  }

The objective of the transform method is thus to combine, filter, check relations between multiples Events to form a new Event. In the example of the ExposureTransformer, it combines multiples Drug Events based on the logic of the exposurePeriodAdder of the ExposuresTransformerConfig to form an Exposure while making sure that it is contained within the period defined in the FollowUp Event.