Risk classification

As you have learned, model governance never follows a one-size-fits-all approach.

As each additional control can slow down the project and increase the final cost of the ML service, you should apply them only where they are really necessary.

One of the first steps in this process is to classify the model in question into the appropriate risk category.

Whether you use three risk categories (low, medium, high), five, or ten -- we usually assign the models to them based on two questions, which are: