Databricks Certified Machine Learning Professional Practice Test 94
Question 930
Consider the following fictional Databricks production case: A demographic fairness review uses a tiny sensitive subgroup with three labeled examples and reports its observed error as definitive. Which result is the strongest acceptance criterion after remediation?
- All denominator boundary fixtures produce defined permitted outputs and appropriate alert behavior.
- A cohort with large cumulative distance from its training baseline raises a distinct signal.
- Undersized cohorts are appropriately suppressed while adequately powered authorized slices remain measurable.
- A replayed morning surge remains within throughput and availability limits with timely scale behavior.
Explanation
Correct Answer
C:This verification measures the required result rather than just process liveness. Undersized cohorts are appropriately suppressed while adequately powered authorized slices remain measurable. It confirms whether the targeted action succeeded: require approved aggregation, privacy safeguards and sample sufficiency for sensitive slices.
Incorrect Answers
A:That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A custom drift metric is expressed as today-minus-yesterday percentage, but yesterday's eligible count can be zero. Here the actual success criterion is undersized cohorts are appropriately suppressed while adequately powered authorized slices remain measurable.
B:That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A model trained on last year's customer mix is monitored only against yesterday's already shifted cohort, so its original training-population mismatch is hidden. Here the actual success criterion is undersized cohorts are appropriately suppressed while adequately powered authorized slices remain measurable.
D:That acceptance test is directed at a different drift detection and lakehouse monitoring situation: An endpoint receives a large request burst every morning and its monitoring alert reacts only to 5xx errors after autoscaling cannot keep up. Here the actual success criterion is undersized cohorts are appropriately suppressed while adequately powered authorized slices remain measurable.
Question 931
The engineering team encounters the following issue: An engineer logs a customer-specific API token inside the pickled PythonModel object before registering it to Unity Catalog. What investigation would most effectively isolate this behavior?
- Compare current endpoint settings with source-controlled deployment configuration and documented backup state.
- Inspect deployment operation status, served entity health and known inference fixture after API completion.
- Inspect actual test response JSON and registered output schema for a representative batch.
- Scan the serialized custom model artifact and logs for token material and audit its access history.
Explanation
Correct Answer
D:This is the most discriminating check for the observed symptoms. Scan the serialized custom model artifact and logs for token material and audit its access history. The underlying technical issue is that model artifacts should not serialize secrets; serving identities should resolve credentials securely at runtime.
Incorrect Answers
A:That diagnostic evidence is directed at a different custom model serving situation: A platform engineer creates a custom serving endpoint manually and assumes the exact settings can be recovered after an accidental deletion. For the stated incident, the decisive observation is scan the serialized custom model artifact and logs for token material and audit its access history.
B:That diagnostic evidence is directed at a different custom model serving situation: A Databricks SDK deployment script sends a model-version update but does not wait for the endpoint to reach a usable ready state before announcing success. For the stated incident, the decisive observation is scan the serialized custom model artifact and logs for token material and audit its access history.
C:That diagnostic evidence is directed at a different custom model serving situation: A serving client calls the REST endpoint expecting one number but the custom PyFunc returns a structured list of prediction and confidence fields. For the stated incident, the decisive observation is scan the serialized custom model artifact and logs for token material and audit its access history.
Question 932
Consider the following fictional Databricks production case: An ML project stores test files directly in each production notebook and execution of a deployed notebook inadvertently reruns expensive tests. What consequence most specifically follows from leaving the issue unchanged?
- Production jobs waste resources or mutate test artifacts during business scoring.
- A valid-looking endpoint reverses class interpretation for downstream automation.
- Predictions drift between restarts because identical labels occupy different vector positions.
- A simple deterministic transform bug reaches users before it is caught by a heavy live workflow.
Explanation
Correct Answer
A:This consequence follows from the specific unresolved condition. Production jobs waste resources or mutate test artifacts during business scoring. The causal constraint is: Test organization should separate production orchestration from the test runner and control execution stages.
Incorrect Answers
B:That direct consequence is directed at a different validation testing situation: An end-to-end validation uses only a synthetic positive label and fails to detect a swapped positive/negative class mapping in inference. The more immediate risk here is production jobs waste resources or mutate test artifacts during business scoring.
C:That direct consequence is directed at a different validation testing situation: A notebook's label encoder produces a different category order depending on an unordered Python set, but tests assert only output length. The more immediate risk here is production jobs waste resources or mutate test artifacts during business scoring.
D:That direct consequence is directed at a different validation testing situation: A developer changes a feature transform in a notebook and directly uses production data as the only test case. The more immediate risk here is production jobs waste resources or mutate test artifacts during business scoring.
Question 933
Consider the following fictional Databricks production case: An ML engineer manually compares two logistic-regression regularization values on a single holdout but has no reusable SparkML parameter-search plan. Which implementation is most appropriate for the stated incident?
- Create an appropriate holdout design, including relevant new-user or time-split cohorts, and evaluate there.
- Apply VectorAssembler to produce the features vector before fitting the chosen learning algorithm.
- Build an explicit parameter grid with a compatible CrossValidator or TrainValidationSplit and preserve a final untouched test set.
- Order categorical indexing and encoding before assembling numeric and encoded features into a vector.
Explanation
Correct Answer
C:This corrects the relevant failure boundary instead of merely hiding the symptom. Build an explicit parameter grid with a compatible CrossValidator or TrainValidationSplit and preserve a final untouched test set. The issue matters because the selected hyperparameter reflects an unrepeatable or misreported single holdout trial.
Incorrect Answers
A:That corrective action is directed at a different using spark ml situation: A recommender experiment computes all evaluation metrics on training interactions and reports the result as generalization to new customers. The relevant corrective action here is build an explicit parameter grid with a compatible CrossValidator or TrainValidationSplit and preserve a final untouched test set.
B:That corrective action is directed at a different using spark ml situation: A team wants to convert several numeric inputs into one features vector but treats VectorAssembler as a clustering Estimator that should output predictions. The relevant corrective action here is build an explicit parameter grid with a compatible CrossValidator or TrainValidationSplit and preserve a final untouched test set.
D:That corrective action is directed at a different using spark ml situation: An IoT classifier passes raw text categories directly into VectorAssembler ahead of StringIndexer and OneHotEncoder, generating incompatible vector inputs. The relevant corrective action here is build an explicit parameter grid with a compatible CrossValidator or TrainValidationSplit and preserve a final untouched test set.
Question 934
Consider the following fictional Databricks production case: A training notebook reports F1 for several models but only logs the final model's parameters in MLflow, leaving metrics and hyperparameters disconnected. Which result is the strongest acceptance criterion after remediation?
- Each worker's fit results appear under its own child run without overwritten artifacts.
- Concurrent cross-user fixtures return isolated correct outputs with no cache-key collision.
- An audit enumerates every unique trial across resume boundaries and traces its exact run parentage.
- Re-executing the recorded configuration reproduces the tracked validation metric within expected variance.
Explanation
Correct Answer
D:This verification measures the required result rather than just process liveness. Re-executing the recorded configuration reproduces the tracked validation metric within expected variance. It confirms whether the targeted action succeeded: log each candidate's complete parameter map and named metrics in the same versioned run.
Incorrect Answers
A:That acceptance test is directed at a different advanced mlflow usage situation: A training orchestrator starts parallel child fits but all workers accidentally log into one active MLflow run and overwrite each other's metrics. Here the actual success criterion is re-executing the recorded configuration reproduces the tracked validation metric within expected variance.
B:That acceptance test is directed at a different advanced mlflow usage situation: A fraud PyFunc uses a global mutable feature cache initialized by the first request and mixes data between subsequent users. Here the actual success criterion is re-executing the recorded configuration reproduces the tracked validation metric within expected variance.
C:That acceptance test is directed at a different advanced mlflow usage situation: A research team restarts a long hyperparameter sweep and merges all newly executed trials into a new experiment without connecting them to the original parent. Here the actual success criterion is re-executing the recorded configuration reproduces the tracked validation metric within expected variance.
Question 935
Consider the following fictional Databricks production case: A notebook uses the MLflow Deployments SDK to send a column-major payload to a custom model endpoint expecting a different documented dataframe schema. Which implementation is most appropriate for the stated incident?
- Route changes through a single approved configuration source and verify the active endpoint after each change.
- Log the tokenizer as a referenced MLflow artifact and load it from the model context in the PyFunc implementation.
- Log the supported custom PyFunc model and register it to an authorized governed model name.
- Format requests according to the supported MLflow Deployments client and registered model input schema.
Explanation
Correct Answer
D:This corrects the relevant failure boundary instead of merely hiding the symptom. Format requests according to the supported MLflow Deployments client and registered model input schema. The issue matters because the endpoint rejects requests or silently associates values with incorrect features.
Incorrect Answers
A:That corrective action is directed at a different custom model serving situation: A team edits a model endpoint through the UI while its CI pipeline concurrently applies an older configuration and reverts the intended traffic split. The relevant corrective action here is format requests according to the supported MLflow Deployments client and registered model input schema.
B:That corrective action is directed at a different custom model serving situation: A custom PyFunc estimator loads a tokenizer file from an engineer's laptop path when called in a serving container. The relevant corrective action here is format requests according to the supported MLflow Deployments client and registered model input schema.
C:That corrective action is directed at a different custom model serving situation: A research group wants to register a custom classification model in Unity Catalog but uses a local path as its complete model identity and has no governed model name. The relevant corrective action here is format requests according to the supported MLflow Deployments client and registered model input schema.
Question 936
Consider the following fictional Databricks production case: A monitor owner writes a custom metric that logs unaggregated account-level health data in a broadly shared results table. Which result is the strongest acceptance criterion after remediation?
- A deliberate new-customer regression is detected with sufficient labeled sample support.
- Known labeled fixtures join one-to-one with their source predictions and yield the expected error rate.
- A privacy scan confirms only approved aggregate values appear in the shared metric output.
- A synthetic region-only failure triggers the scoped alert without misclassifying healthy regions.
Explanation
Correct Answer
C:This verification measures the required result rather than just process liveness. A privacy scan confirms only approved aggregate values appear in the shared metric output. It confirms whether the targeted action succeeded: aggregate or minimize the custom metric and apply authorized monitoring-table permissions.
Incorrect Answers
A:That acceptance test is directed at a different drift detection and lakehouse monitoring situation: An account-risk model's fraud recall is stable overall but unexpectedly falls for new customers in their first 30 days. Here the actual success criterion is a privacy scan confirms only approved aggregate values appear in the shared metric output.
B:That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A model performance dashboard compares this week's labels to last week's predictions without joining by stable request ID or consistent outcome horizon. Here the actual success criterion is a privacy scan confirms only approved aggregate values appear in the shared metric output.
D:That acceptance test is directed at a different drift detection and lakehouse monitoring situation: An alert query uses one global transaction-failure count and masks a complete outage in a small high-value region. Here the actual success criterion is a privacy scan confirms only approved aggregate values appear in the shared metric output.
Question 937
A production ML review contains this situation: A Databricks Model Serving endpoint hosts champion and candidate versions, but a release config accidentally routes 100% of traffic to the candidate instead of 10%. Which check shows that this particular problem has been resolved?
- All synthetic protected-region requests remain on their authorized version during canary traffic.
- A controlled failback restores the previous version without rebuilding model artifacts.
- An endpoint configuration readback and sampled test traffic confirm the intended distribution.
- A simulated candidate failure affects only the canary share and triggers a stop or rollback gate.
Explanation
Correct Answer
C:This verification measures the required result rather than just process liveness. An endpoint configuration readback and sampled test traffic confirm the intended distribution. It confirms whether the targeted action succeeded: set and verify the intended approved candidate-versus-champion traffic split before exposure.
Incorrect Answers
A:That acceptance test is directed at a different deployment strategies situation: A model release requires all clients from a restricted region to keep the old version, but a simple percentage split sends some of those clients to the candidate. Here the actual success criterion is an endpoint configuration readback and sampled test traffic confirm the intended distribution.
B:That acceptance test is directed at a different deployment strategies situation: A bank requires an immediate rollback after deployment and wants the prior serving environment intact until the new environment passes tests. Here the actual success criterion is an endpoint configuration readback and sampled test traffic confirm the intended distribution.
D:That acceptance test is directed at a different deployment strategies situation: A travel site serves 30,000 bookings per minute and wants to evaluate a new fraud model while limiting customer impact during its first hour. Here the actual success criterion is an endpoint configuration readback and sampled test traffic confirm the intended distribution.
Question 938
In an enterprise ML workflow, the reported issue is: A churn model logs requests, predictions, model versions and later ground-truth labels, but monitoring is configured as a generic feature snapshot. Which change addresses the real cause without undermining other requirements?
- Choose a compatible inference profile and map the correct prediction, version and label columns.
- Configure a documented threshold with the responsible on-call team and a tested delivery route.
- Configure model-identity-aware inference monitoring with version-specific comparisons.
- Compare to a justified weekend reference or adjust baseline design with documented change control.
Explanation
Correct Answer
A:This corrects the relevant failure boundary instead of merely hiding the symptom. Choose a compatible inference profile and map the correct prediction, version and label columns. The issue matters because performance comparisons by deployed model version cannot be calculated from the generic snapshot configuration.
Incorrect Answers
B:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A fraud-feature drift monitor crosses its threshold but the alert has no assigned recipient and no incident ownership. The relevant corrective action here is choose a compatible inference profile and map the correct prediction, version and label columns.
C:That corrective action is directed at a different drift detection and lakehouse monitoring situation: An inference profile groups two active model versions into one aggregate and hides a regression introduced by the new release. The relevant corrective action here is choose a compatible inference profile and map the correct prediction, version and label columns.
D:That corrective action is directed at a different drift detection and lakehouse monitoring situation: An online feature monitor compares weekend shopping to weekday reference data and treats expected weekend behavior as a model failure. The relevant corrective action here is choose a compatible inference profile and map the correct prediction, version and label columns.
Question 939
A production ML review contains this situation: An analyst wants to inspect today’s static customer table for missing values and profile changes but chooses an inference monitor requiring model IDs and predictions. What production-safe change best matches the evidence?
- Use a supported snapshot data profile for the governed table and its quality metrics.
- Maintain a model-version-linked training baseline alongside rolling comparisons.
- Add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.
- Normalize and validate a supported timestamp field before applying the configured temporal profile.
Explanation
Correct Answer
A:This corrects the relevant failure boundary instead of merely hiding the symptom. Use a supported snapshot data profile for the governed table and its quality metrics. The issue matters because configuration fails on nonexistent inference fields and delays basic table monitoring.
Incorrect Answers
B:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A model trained on last year's customer mix is monitored only against yesterday's already shifted cohort, so its original training-population mismatch is hidden. The relevant corrective action here is use a supported snapshot data profile for the governed table and its quality metrics.
C:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A monitor alerts whenever a ratio exceeds 0.3 for a single minute, although normal traffic produces brief spikes and operations require ten-minute persistence. The relevant corrective action here is use a supported snapshot data profile for the governed table and its quality metrics.
D:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A time-series monitor uses a string-formatted date that cannot be reliably converted to timestamps and silently groups most rows into an unknown bucket. The relevant corrective action here is use a supported snapshot data profile for the governed table and its quality metrics.
Question 940
A production ML review contains this situation: A real-time feature job updates user purchase totals without checking whether upstream transaction events are delivered twice. What is the most material immediate risk in this situation?
- Scores pull the wrong customer's features or fail after name changes.
- Historical training datasets cannot be faithfully reconstructed after the overwrite.
- Several models change their effective input distributions without deliberate retraining or review.
- One purchase increases the spending feature twice, affecting downstream risk scores.
Explanation
Correct Answer
D:This consequence follows from the specific unresolved condition. One purchase increases the spending feature twice, affecting downstream risk scores. The causal constraint is: Streaming derived features should deduplicate at a stable business-event key before non-idempotent aggregation.
Incorrect Answers
A:That direct consequence is directed at a different advanced feature store concepts situation: A low-latency feature table uses external customer names as lookup keys; names change and are not unique. The more immediate risk here is one purchase increases the spending feature twice, affecting downstream risk scores.
B:That direct consequence is directed at a different advanced feature store concepts situation: A model's feature generation step overwrites all feature-table history daily even though time-aware training requires earlier snapshots. The more immediate risk here is one purchase increases the spending feature twice, affecting downstream risk scores.
C:That direct consequence is directed at a different advanced feature store concepts situation: A team updates a shared on-demand feature function without version control and unexpectedly changes both batch retraining and online serving behavior. The more immediate risk here is one purchase increases the spending feature twice, affecting downstream risk scores.