Getting Started with Streaming Analytics
Also available as:
PDF
loading table of contents...

Score the Model Using the PMML Processor and Alert

Now you are ready to score the logistical regression model.

  1. Drag the PMML processor to the canvas and connect it to the Normalize processor.

  2. Configure the PMML processor like the following by selecting the DriverViolationPredictionModel that you uploaded earlier to the Model Registry.

    After this processor executes, a new field called ViolationPredicted is added to stream for the result of the prediction. In output fields, select all the contextual fields you want to pass on including the model value result.

  3. Determine if the model predicted if the driver will commit a violation by dragging a rule processor to the canvas and configuring a rule like the following:

  4. If a violation is predicted, send it to a Druid to display on a dashboard. Drag the Druid processor to canvas and configure.
  5. Stream the events into a cube called alerts-violation-predictions-cube.

The final flow looks like the following: