Question 16
UnclassifiedWhat does model drift refer to?
Correct answer: B
Explanation
Model drift refers to when a model’s performance worsens over time because the underlying data patterns change. It is the “degradation of model performance over time as the data or relationships change,” which means the model no longer matches the current environment.
Why each option is right or wrong
A. Hardware failure of the inference server causing predictions to stop entirely
B. Degradation of model performance over time as the data or relationships change
Model drift is the observed decline in predictive accuracy when the real-world data distribution or the relationship between inputs and outputs changes after deployment. In practice, this is diagnosed by comparing current performance metrics against the model’s original validation or production baseline over time; no fixed statutory threshold applies, but the defining feature is a measurable deterioration caused by changing data patterns rather than a coding error.
C. A model running too slowly because of inefficient code or limited compute resources
D. Overfitting caused by choosing hyperparameters that are too complex for the training data