Deploying a model is not the end of the work. The world during training and the world during operation are constantly diverging. User behavior changes, sensors are replaced, economic conditions shift, and the distribution of data flowing into the model diverges from the training data. This is called drift, and if left unmanaged, model performance quietly deteriorates. SageMaker Model Monitor is a service that automatically monitors deployed endpoints to detect this drift.
The MLA-C01 exam covers "the monitoring domain (Domain 4)" where it asks about Model Monitor's four types of monitors and how they work. Today we cover the two most fundamental types: Data Quality and Model Quality drift