Drift & Model Monitor — MLA-C01 Practice Question
A representative AWS Machine Learning Engineer Associate (MLA-C01) exam question on Drift & Model Monitor. Work through it below, then read why each option is right or wrong.
Short answer
The correct answer is B. Data drift.
Data drift (also called covariate shift) occurs when the statistical distribution of the input features changes while the underlying relationship between inputs and the target stays the same. A shift in applicant income/DTI distributions is exactly this.
The Question
A retail bank retrained its loan-default model six months ago. Recently, the *distribution of incoming applicant income and debt-to-income values* has shifted compared to the training data, even though the relationship between those features and default outcomes has not changed. Which type of drift is the bank observing?
Why B is correct
Data drift (also called covariate shift) occurs when the statistical distribution of the input features changes while the underlying relationship between inputs and the target stays the same. A shift in applicant income/DTI distributions is exactly this.
Why the other options are wrong
Concept drift means the relationship between inputs and the target label itself changes (e.g., the same income/DTI profile now defaults at a different rate). That is not described here.
Label leakage is a training-time data issue where information about the target leaks into features; it is unrelated to a shift observed in production input distributions.
Overfitting is a training-time generalization problem, not a description of a change in the live input distribution.
Key idea: Drift & Model Monitor
Why B is correct: Data drift (also called covariate shift) occurs when the statistical distribution of the input features changes while the underlying relationship between inputs and the target stays the same. A shift in applicant income/DTI distributions is exactly this. Why A is wrong: Concept drift means the relationship between inputs and the target label itself changes (e.g., the same income/DTI profile now defaults at a different rate). That is not described here. Why C is wrong: Label leakage is a training-time data issue where information about the target leaks into features; it is unrelated to a shift observed in production input distributions. Why D is wrong: Overfitting is a training-time generalization problem, not a description of a change in the live input distribution. On the MLA-C01 exam, questions in the "ML Solution Monitoring, Maintenance, and Security" domain test whether you can map a scenario's constraints to the right choice. Read the requirement carefully, eliminate options that violate any single constraint, and pick the one that satisfies all of them with the least operational overhead.
Ready to see how you'd score?
Take the free practice quiz and find out which AWS Certified Machine Learning Engineer - Associate domains you need to focus on. No signup required.
Practice 5 similar questions
Same cert, same or adjacent domain. Use these after reviewing the explanation.