PMLE Practice Question: Refer to the exhibit

Exhibit

modelMonitoringConfig:
  skewDetection:
    defaultThreshold: 0.3
    featureThresholds:
      income: 0.2
  driftDetection:
    defaultThreshold: 0.5
  samplingRate: 0.5

Refer to the exhibit. A team configured Vertex AI Model Monitoring with skew detection for feature "income" with a threshold of 0.2. However, they have not received any alerts even though they suspect data drift. What is the most likely reason?

⚠ Common exam trap

PMLE often tests the assumption that configuring a monitor guarantees alerts — candidates overlook that skew detection requires the feature to be present in serving data, so a missing or renamed feature silently disables detection for that feature.

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 'income' feature is not present in the serving data

Vertex AI Model Monitoring skew detection compares the training data distribution against the serving (prediction) data distribution for each monitored feature. If the 'income' feature is missing from the serving requests — for example, the endpoint's input schema does not include it, or it is named differently — the monitor has no serving-side data to compare and cannot compute skew, so no alert fires even if drift is suspected. The feature must be present in the serving payload for skew to be evaluated.

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 monitoring is not enabled for the endpoint
Why it's wrong here
Skew detection compares serving traffic against the training baseline; it does not require the endpoint to be deployed with monitoring enabled, so drift in prediction requests is still measured. It tempts because endpoint monitoring is a real prerequisite for prediction drift, which is the correct choice when detecting drift in model outputs.
✓
The 'income' feature is not present in the serving data
Why this is correct
Skew detection compares training and serving feature distributions, so it only evaluates features present in both. If 'income' is absent from serving data, no comparison occurs and no alert fires regardless of the 0.2 threshold, explaining the silence despite suspected drift.
✗
The actual skew is below the threshold
Why it's wrong here
Skew compares training and serving feature distributions, so a genuine skew below 0.2 correctly produces no alert; the team's suspicion of drift does not override the measured value. It tempts because a low skew is a legitimate outcome, but the stem asks why alerts are absent despite suspected drift, which points to monitoring configuration rather than the computed metric.
✗
The drift detection threshold is set higher
Why it's wrong here
A higher threshold would suppress alerts, but the stem already fixes the threshold at 0.2, so this restates the given value rather than explaining the absence. It tempts because raising a threshold genuinely reduces alert volume, which is the right fix when thresholds are misconfigured, but here the threshold is specified and not the variable in question.

Go deeper

PMLE Practice Question: Refer to the exhibit

Exhibit

Why each option matters

Option-by-option breakdown

Related to this question