AI & Machine Learning

Model Drift (Data and Concept Drift)

Model drift is the decline in a deployed model's performance over time because the data it sees changes, known as data drift, or because the relationship between inputs and the correct output changes, known as concept drift.

Data drift occurs when input distributions shift, for example with new suppliers or raw materials, camera replacements, sensor recalibration, seasonal temperatures or new product variants. Concept drift occurs when the meaning of the data changes, for example when a process change alters which vibration patterns indicate a fault, or when fraud tactics evolve. Drift can be sudden, gradual or recurring.

Industrial AI is especially exposed because processes, equipment and products change continually. An inspection model may start rejecting good parts after a lighting change, or a predictive maintenance model may miss failures after a machine is rebuilt.

Detection compares live input distributions with the training data using statistical tests or measures such as the population stability index, and tracks performance against ground-truth labels when they become available. Responses include investigating root causes, retraining with recent data and updating validation sets. Monitoring and retraining pipelines are a core part of MLOps.

Key points

• Performance declines as data or input-output relationships change
• Data drift is a shift in inputs; concept drift is a change in meaning
• Common after process, equipment, material or product changes
• Detected by distribution monitoring and performance tracking

Related terms

• MLOpsAI & Machine Learning
• Model EvaluationAI & Machine Learning
• Anomaly DetectionAI & Machine Learning
• Overfitting and UnderfittingAI & Machine Learning
• Predictive Maintenance (PdM)Quality, Reliability & Maintenance

Terms that refer to Model Drift (Data and Concept Drift)

• AI BiasAI & Machine Learning