When configuring a Lakehouse Monitoring inference profile, what is the recommended choice for the baseline table?

Question

Accepted Answer

B. The training or validation dataset used to train and evaluate the model being monitored.. For inference profiles, Databricks recommends using the training or validation dataset as the baseline table. This ensures that drift is measured relative to the data distribution the model was trained and validated on. Using the training data as a baseline allows the system to alert when inference data has drifted from what the model was designed to handle.

When configuring a Lakehouse Monitoring inference profile, what is the recommended choice for the baseline table?

Related Questions

Discussion