hardMultiple Choice
PMLE Practice Question: After setting up model monitoring on Vertex AI…
After setting up model monitoring on Vertex AI for a classification model, the engineer sees a high number of anomaly alerts for the "age" feature. Upon investigation, the age distribution in recent predictions is similar to training data. What might be the cause?
⚠ Common exam trap
Candidates often confuse 'anomaly alerts' with 'model performance degradation' or 'data drift,' but the question specifically states the distribution is similar, so the root cause is a misconfigured sensitivity threshold, not a genuine distribution shift.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
The monitoring threshold for age is too low
The high number of anomaly alerts despite the age distribution being similar to training data indicates that the monitoring threshold for the 'age' feature is set too low. In Vertex AI Model Monitoring, anomaly detection compares recent prediction distributions against a baseline using statistical tests (e.g., the Kolmogorov-Smirnov test for numerical features). If the threshold is too sensitive, even minor, statistically insignificant deviations can trigger alerts, leading to false positives even when the distribution is essentially unchanged.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
The feature importance of age has changed
Why it's wrong here
Feature importance describes how much a trained model relies on age when producing predictions; it is not an input to Vertex AI's distribution-based drift or skew detection. It is tempting because importance shifts accompany concept drift, but monitoring compares feature value distributions, not attribution scores, so it cannot explain these alerts.
- ✗
The monitoring baseline was incorrectly set
Why it's wrong here
A misconfigured baseline would shift the reference distribution itself, yet the stem states recent predictions resemble training data, so the comparison should not fire. It is tempting because baseline setup genuinely causes false alerts, but that would be the answer only if the baseline differed from the actual training distribution.
- ✓
The monitoring threshold for age is too low
Why this is correct
Alerts fire when monitored values breach the configured threshold. Since the recent age distribution matches training data, no genuine drift exists; an excessively low threshold triggers alerts on normal statistical variation, so raising it eliminates the false positives.
- ✗
The model is overfitting to age
Why it's wrong here
Overfitting concerns training-time generalisation, not prediction-time input distributions, so it cannot trigger drift alerts when recent age values already match training data. It is tempting because overfitting does affect model quality, but it would be diagnosed through validation metrics or regularisation, not Vertex AI feature monitoring skew or drift detection.
Go deeper
Related to this question
About these practice questions
One of 775 original PMLE practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This PMLE practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the PMLE exam.