# What is Concept Drift? Definition & Technical Mechanics

> **Source:** [SPIDITS AI AI Glossary](https://spidits.com/ai-glossary)  
> **Category:** Model Operations  
> **Canonical Citation:** [https://spidits.com/ai-glossary/concept-drift](https://spidits.com/ai-glossary/concept-drift)

## Definition
Concept Drift is the phenomenon where the statistical properties of the target variable that a model is predicting change over time in an unforeseen way, causing the model to become inaccurate.

## Architectural Mechanics & Deep Dive
Concept drift is a phenomenon in machine learning where the statistical properties of the target variable change over time in unforeseen ways, causing model performance to degrade. This occurs when real-world relationships evolve, such as changes in consumer behavior or macroeconomic shifts. Mitigating concept drift requires continuous model monitoring, anomaly detection, and periodic retraining on fresh, representative datasets.

## Industry Applications & Real-World Use Cases
Financial fraud detection, demand forecasting models, and long-term MLOps model monitoring.

## Frequently Asked Questions

### How does concept drift differ from data drift?
Data drift is when the distribution of the input features shifts, but the underlying relationships stay the same. Concept drift is when the relationship between the inputs and outputs itself changes.

### How do you handle concept drift in production systems?
By setting up continuous monitoring to detect model performance degradation, and scheduling automated retraining runs with more recent data.

## Related Terminology
- [model-drift](https://spidits.com/ai-glossary/model-drift)
- [mlops](https://spidits.com/ai-glossary/mlops)
- [observability](https://spidits.com/ai-glossary/observability)

---
*From the SPIDITS AI Knowledge Base (300+ terms indexed). Explore full technical definitions: [https://spidits.com/ai-glossary](https://spidits.com/ai-glossary)*