When the statistical properties of input data change over time compared to the training data, potentially degrading model performance. Like a recipe not working well when ingredient quality changes.
A shopping recommendation model experiences data drift when customer behavior shifts during holidays, requiring monitoring and potential retraining.
All major clouds support monitoring for changes in production data compared to training/baseline data. AWS SageMaker Model Monitor, Azure ML data drift monitoring, and Vertex AI Model Monitoring provide managed drift detection and alerting. OCI Data Science supports model deployment, but drift detection is typically implemented by logging features/predictions and using OCI Monitoring/Logging (or custom code) to compute drift metrics.