Industrial Internet Application Development

Model performance

In practice, due to various factors such as operating settings and environment conditions, model performance starts to degrade. It's always good practice to monitor the performance of the model by tracking the predictions of the model and the real observed data over a period of time. This can provide an insight into model performance to take the necessary actions on the acceptance of the model. In some IIoT use case scenarios, it is critical to refrain from accepting model predictions if the prediction goes out of range with an offset limit. In such cases, the performance metrics play a significant role