A model needs a baseline dataset for comparing its performance and identifying any degradation. A baseline is a set of reference data that is used to compare with our current data.
The dataset that was used to train the model is often a good starting point for a baseline. For more in-depth analysis, we may want to use a specific time period or a rolling window of production events.
In Fiddler, the default baseline for all monitoring metrics is the training dataset that was associated with the model during registration. Use this default baseline if you do not anticipate any differences between training and production. New baselines can be added to existing models using the Python client APIs.
Updated 4 months ago