You have been receiving customer data for the past six months. However recently you notice that this data has drastically changed due to the upcoming holiday season. What seems to be taking place?

Data drift is the shift in the statistical distribution of incoming data relative to past data. Holiday-season purchasing behaviors alter feature distributions, so the change observed before any model degradation is data drift.

A Stakeholder support concerns governance, unrelated to data statistics. B Milestone completeness is a project-planning issue, not data quality. D Model drift refers to declining model accuracy; no model behaviour is discussed.

1. Widmer, G., Kubat, M. “Learning in the Presence of Concept Drift,” Machine Learning 23(1), 1996, pp. 69-70, DOI:10.1007/BF00116900.

2. Baier, L. et al., “Handling Data and Concept Drift in ML Systems,” arXiv:2102.11129, §2 (2021).