Schema compliance checks whether incoming data still matches the structure expected by a machine-learning pipeline: required features exist, their types match, and no unexpected columns appear. In production monitoring, it acts as an early data-quality guard before feature values reach the model. section 14.5.1
Retail forecasting pipeline
A retail forecasting monitor checks each incoming dataset for missing features, type mismatches, and newly added columns. It runs alongside missing-data checks, feature-range checks, correlation checks, and Evidently AI data-quality and stability presets. section 14.5.1
