What Is Covariate Shift?
Covariate Shift is a form of distribution shift where the statistical distribution of input features changes between training and deployment, while the underlying relationship between inputs and outputs remains the same.
Covariate Shift — a form of distribution shift where the statistical distribution of input features changes between training and deployment, while the underlying relationship between inputs and outputs remains the same.
Covariate shift is one of three main types of model drift. It is common when a model is trained in one environment and deployed in another — for example, a fraud detection model trained on historical transaction patterns may underperform when payment behaviours change after a pandemic, economic shock, or regulatory change. The model's learned relationships may still be valid in principle, but its training data no longer represents the actual distribution of inputs. Monitoring for covariate shift requires tracking input feature distributions in production.
Source: NIST AI 100-1; ISO/IEC 42001:2023, Clause 9.1
Plain-language explanation
Covariate shift is one of three main types of model drift. It is common when a model is trained in one environment and deployed in another — for example, a fraud detection model trained on historical transaction patterns may underperform when payment behaviours change after a pandemic, economic shock, or regulatory change. The model's learned relationships may still be valid in principle, but its training data no longer represents the actual distribution of inputs. Monitoring for covariate shift requires tracking input feature distributions in production.
See where you stand on AI governance
Take the free 7-question maturity assessment and get a personalised action plan.
Free assessment — 3 minutes →