The Brains.VOS outputs can be used to quantify and track model accuracy. Depending on the model/application type some variables which can be tracked over time are:

  • Root Mean Square Error (RMSE)
  • Absolute Delta Between Predicted and Actual
  • Uncertainty (%)
  • etc..

This quantification is deployment type dependent. The example below shows a multi-line widget over time

  • Absolute Model Delta
  • Model Uncertainty

Models are updated to ensure the highest possible accuracy in the following situations:

  • Models are updated quarterly as a standard part of the subscription package
  • If the models uncertainty drifts outside of agreed boundaries
  • If there has been a change in configuration of the process itself (i.e. new pipe diameter)
Did this answer your question?