Which data concept describes changes in input data distribution over time?

Which data concept describes changes in input data distribution over time?

• A. Data drift
• B. Concept drift
• C. Training data
• D. Validation data

Answer: (A)

Data drift describes changes in input data distribution over time. When the statistical properties of the features you feed a model shift after training—like their means, variances, or the overall distribution changing—your model may see inputs that look different than what it learned on. This can cause predictions to become less accurate even if the underlying concept hasn’t changed. This is distinct from concept drift, which is about changes in the relationship between inputs and the target output, not just the inputs themselves. Training data and validation data are simply datasets used for learning and evaluation, not the phenomenon of distribution change. So, the term that captures changes in input data distribution over time is data drift.

Data drift describes changes in input data distribution over time. When the statistical properties of the features you feed a model shift after training—like their means, variances, or the overall distribution changing—your model may see inputs that look different than what it learned on. This can cause predictions to become less accurate even if the underlying concept hasn’t changed. This is distinct from concept drift, which is about changes in the relationship between inputs and the target output, not just the inputs themselves. Training data and validation data are simply datasets used for learning and evaluation, not the phenomenon of distribution change. So, the term that captures changes in input data distribution over time is data drift.