Databricks Certified Machine Learning Professional Practice Test 98
Question 970
Consider the following fictional Databricks production case: A streaming recommender's request-time feature function calls a slow third-party API synchronously on every prediction. Which result is the strongest acceptance criterion after remediation?
- Replay reconstructs only feature values visible by each original decision time.
- Duplicated input fixtures contribute exactly once to the computed feature total.
- Load tests meet request-time latency while documented fallback values remain semantically valid.
- P95 inference latency and feature freshness meet service targets under realistic load.
Explanation
Correct Answer
C:This verification measures the required result rather than just process liveness. Load tests meet request-time latency while documented fallback values remain semantically valid. It confirms whether the targeted action succeeded: use a supported cached or precomputed feature strategy or bounded fallback consistent with freshness needs.
Incorrect Answers
A:That acceptance test is directed at a different advanced feature store concepts situation: A retailer's features arrive two hours late, and offline training uses event time alone even though the online system could not have seen those late arrivals at prediction time. Here the actual success criterion is load tests meet request-time latency while documented fallback values remain semantically valid.
B:That acceptance test is directed at a different advanced feature store concepts situation: A real-time feature job updates user purchase totals without checking whether upstream transaction events are delivered twice. Here the actual success criterion is load tests meet request-time latency while documented fallback values remain semantically valid.
D:That acceptance test is directed at a different advanced feature store concepts situation: A pricing endpoint queries a large offline Delta feature table for every user request, and P95 response time reaches several seconds. Here the actual success criterion is load tests meet request-time latency while documented fallback values remain semantically valid.
Question 971
A production ML review contains this situation: A bundle references a model registry name without qualifying its environment, and a development deployment accidentally updates production registered-model aliases. What is the most material immediate risk in this situation?
- Critical retraining stops when a human account or notebook workspace changes.
- A developer's experiment can alter production feature tables or serving configuration.
- Sensitive records cross an unapproved residency boundary through convenience caching.
- A test release mutates production model aliases despite otherwise successful deployment.
Explanation
Correct Answer
D:This consequence follows from the specific unresolved condition. A test release mutates production model aliases despite otherwise successful deployment. The causal constraint is: Bundle target parameters must resolve to explicitly authorized environment-specific resources.
Incorrect Answers
A:That direct consequence is directed at a different environment architectures situation: A distributed retraining workflow uses ephemeral personal compute and fails when the original engineer's account is disabled. The more immediate risk here is a test release mutates production model aliases despite otherwise successful deployment.
B:That direct consequence is directed at a different environment architectures situation: A company operates development and production models in one workspace under the same broad service account and frequently overwrites production tables during experiments. The more immediate risk here is a test release mutates production model aliases despite otherwise successful deployment.
C:That direct consequence is directed at a different environment architectures situation: A multilingual prediction service must handle regional data residency, but the team copies a shared feature cache across regions without governance. The more immediate risk here is a test release mutates production model aliases despite otherwise successful deployment.
Question 972
Consider the following fictional Databricks production case: A model's feature generation step overwrites all feature-table history daily even though time-aware training requires earlier snapshots. What underlying principle determines the most appropriate interpretation?
- Online feature access should use least-privilege identity and scoped permissions rather than universal shared credentials.
- On-demand feature functions must explicitly handle invalid input and output type constraints.
- Streaming feature windows should use appropriate event-time semantics and a stated lateness policy.
- Historical feature timelines must be retained when reproducible point-in-time joins depend on prior values.
Explanation
Correct Answer
D:The technical constraint is specific to the reported failure. Historical feature timelines must be retained when reproducible point-in-time joins depend on prior values. To test it, compare current feature-table history and prior Delta versions against the training-set timestamp requirements.
Incorrect Answers
A:That underlying principle is directed at a different advanced feature store concepts situation: A model serving service uses a broadly privileged shared token to read every online feature table when it needs only one approved feature set. The current incident instead turns on historical feature timelines must be retained when reproducible point-in-time joins depend on prior values.
B:That underlying principle is directed at a different advanced feature store concepts situation: A payment classifier's on-demand feature division returns infinity for zero denominators and fails serialization in the serving endpoint. The current incident instead turns on historical feature timelines must be retained when reproducible point-in-time joins depend on prior values.
C:That underlying principle is directed at a different advanced feature store concepts situation: A sensor feature pipeline computes rolling 30-minute averages by arrival time, although delayed events carry trustworthy event timestamps. The current incident instead turns on historical feature timelines must be retained when reproducible point-in-time joins depend on prior values.
Question 973
The engineering team encounters the following issue: A preprocessing function divides by a standard deviation and returns infinity for a constant numeric column. Which experiment most directly demonstrates a successful fix?
- A constant-vector unit fixture returns a documented safe result with no NaN or infinity.
- Known positive and negative synthetic inputs map to the intended serving output labels.
- Test runs require no unauthorized identifiers and negative access probes fail as intended.
- Normal production execution does not run tests, while CI executes the full isolated suite.
Explanation
Correct Answer
A:This verification measures the required result rather than just process liveness. A constant-vector unit fixture returns a documented safe result with no NaN or infinity. It confirms whether the targeted action succeeded: guard zero-variance inputs and assert expected fallback or explicit failure semantics in unit tests.
Incorrect Answers
B:That acceptance test 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. Here the actual success criterion is a constant-vector unit fixture returns a documented safe result with no NaN or infinity.
C:That acceptance test is directed at a different validation testing situation: An engineer uses only production replay data to validate a code change and exposes customer events in an unrestricted test workspace. Here the actual success criterion is a constant-vector unit fixture returns a documented safe result with no NaN or infinity.
D:That acceptance test is directed at a different validation testing situation: An ML project stores test files directly in each production notebook and execution of a deployed notebook inadvertently reruns expensive tests. Here the actual success criterion is a constant-vector unit fixture returns a documented safe result with no NaN or infinity.
Question 974
The engineering team encounters the following issue: A streaming anomaly team uses a static read of a message directory once, believing the trained SparkML model will continuously produce scores for future arriving events. Which technical principle most directly explains the root of the issue?
- Hyperparameter selection uses validation evidence; the final unbiased estimate requires an independent untouched test split.
- Ranking quality and operating threshold quality are distinct; precision and recall depend on the chosen cutoff.
- A fitted model can transform streaming DataFrames in a supported Structured Streaming pipeline, but a one-time static read is not a continuing source.
- Keeping fitted preprocessing and the estimator in one saved PipelineModel prevents training-serving transformation drift.
Explanation
Correct Answer
C:The technical constraint is specific to the reported failure. A fitted model can transform streaming DataFrames in a supported Structured Streaming pipeline, but a one-time static read is not a continuing source. To test it, inspect read versus readStream setup and the active streaming query's progress and sink contract.
Incorrect Answers
A:That underlying principle is directed at a different using spark ml situation: After choosing the best parameter map, a data scientist reports the same cross-validation score as an independent final test result without any untouched data. The current incident instead turns on a fitted model can transform streaming DataFrames in a supported Structured Streaming pipeline, but a one-time static read is not a continuing source.
B:That underlying principle is directed at a different using spark ml situation: A binary classifier ranks risky cases well but sends too many alerts at a default 0.5 decision threshold. The current incident instead turns on a fitted model can transform streaming DataFrames in a supported Structured Streaming pipeline, but a one-time static read is not a continuing source.
D:That underlying principle is directed at a different using spark ml situation: A churn experiment separately scripts feature scaling in training but omits that scaling when the deployed prediction job loads the classifier. The current incident instead turns on a fitted model can transform streaming DataFrames in a supported Structured Streaming pipeline, but a one-time static read is not a continuing source.
Question 975
Consider the following fictional Databricks production case: A loan-risk feature usually has median debt-to-income 0.34, but a new cohort has median 0.61 while the average stays almost unchanged due to outliers. Which observation most directly tests the suspected failure boundary?
- Inspect the table schema for presence of model, prediction and timestamp columns versus the requested snapshot statistics.
- Check label availability, label delay and prediction-to-label matching IDs in the inference data.
- Inspect inference-record completeness for request, model version, inputs, outputs and timestamps.
- Compare recent and baseline quantiles plus a supported numeric distribution-difference metric at equivalent sample sizes.
Explanation
Correct Answer
D:This is the most discriminating check for the observed symptoms. Compare recent and baseline quantiles plus a supported numeric distribution-difference metric at equivalent sample sizes. The underlying technical issue is that distributional drift cannot be ruled out by one stable mean; numeric quantiles and supported statistical comparisons can expose shape changes.
Incorrect Answers
A:That diagnostic evidence is directed at a different drift detection and lakehouse monitoring 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. For the stated incident, the decisive observation is compare recent and baseline quantiles plus a supported numeric distribution-difference metric at equivalent sample sizes.
B:That diagnostic evidence is directed at a different drift detection and lakehouse monitoring situation: A model's inference table records output probabilities but omits ground-truth labels, yet reviewers expect immediate precision and recall metrics. For the stated incident, the decisive observation is compare recent and baseline quantiles plus a supported numeric distribution-difference metric at equivalent sample sizes.
C:That diagnostic evidence is directed at a different drift detection and lakehouse monitoring situation: A serving endpoint appears healthy but nobody records model version, request identifier or prediction value, so a drift incident cannot be reconstructed. For the stated incident, the decisive observation is compare recent and baseline quantiles plus a supported numeric distribution-difference metric at equivalent sample sizes.
Question 976
In an enterprise ML workflow, the reported issue is: A regulated company changes deployed Python notebook code directly in production to correct a model transform while its test environment runs a different untracked revision. How should the engineer validate the correction?
- The same signed release revision and dependencies run in test and production with matching fixture output.
- Negative permission tests deny unapproved endpoint and registry changes while allowing permitted experiment logging.
- Production loads a pinned registered version and access audits identify its owner and version history.
- A replay from retained lineage recreates an example prediction within documented tolerance.
Explanation
Correct Answer
A:This verification measures the required result rather than just process liveness. The same signed release revision and dependencies run in test and production with matching fixture output. It confirms whether the targeted action succeeded: build and validate one immutable release artifact before promoting it through controlled environments.
Incorrect Answers
B:That acceptance test is directed at a different model lifecycle management situation: A platform team proposes using MLflow Tracking alone to enforce production endpoint authorization, ignoring Unity Catalog and service permissions. Here the actual success criterion is the same signed release revision and dependencies run in test and production with matching fixture output.
C:That acceptance test is directed at a different model lifecycle management situation: An engineer stores model weights in a personal workspace folder and treats that path as a production artifact registry with audit history. Here the actual success criterion is the same signed release revision and dependencies run in test and production with matching fixture output.
D:That acceptance test is directed at a different model lifecycle management situation: An organization has reliable model serving but never archives feature definitions or experiment parameters and cannot reproduce predictions after drift. Here the actual success criterion is the same signed release revision and dependencies run in test and production with matching fixture output.
Question 977
Consider the following fictional Databricks production case: A date feature function treats all incoming timestamp strings as UTC even though some include explicit offsets and a unit test uses only Z-suffixed inputs. What underlying principle determines the most appropriate interpretation?
- Cross-component tests must validate shared time semantics and feature availability throughout the ML pipeline.
- Test organization should separate production orchestration from the test runner and control execution stages.
- Release-stage testing must validate the actual artifact and dependency environment being promoted.
- Date transformation tests need timezone and offset boundary fixtures aligned with documented semantics.
Explanation
Correct Answer
D:The technical constraint is specific to the reported failure. Date transformation tests need timezone and offset boundary fixtures aligned with documented semantics. To test it, compare expected and actual normalized timestamps for UTC, offset and daylight-transition examples.
Incorrect Answers
A:That underlying principle is directed at a different validation testing situation: A feature-refresh task writes rows with event time in local timezone while the training join assumes UTC; unit tests of each component pass. The current incident instead turns on date transformation tests need timezone and offset boundary fixtures aligned with documented semantics.
B:That underlying principle is directed at a different validation testing situation: An ML project stores test files directly in each production notebook and execution of a deployed notebook inadvertently reruns expensive tests. The current incident instead turns on date transformation tests need timezone and offset boundary fixtures aligned with documented semantics.
C:That underlying principle is directed at a different validation testing situation: A release pipeline runs development unit tests against a different dependency image from the one deployed to production. The current incident instead turns on date transformation tests need timezone and offset boundary fixtures aligned with documented semantics.
Question 978
A production ML review contains this situation: A hyperparameter search operates on large Spark-managed feature tables but each candidate model is small and independent, and workers repeatedly convert the entire table for each Ray task. Which statement best explains the failure mode?
- Data parallelism duplicates model parameters, while model parallelism or sharding distributes parameters when required.
- Ray Tune can schedule independent hyperparameter trials with explicit per-trial resources rather than serial execution.
- Combining Spark feature preparation with Ray trial execution requires explicit materialization and data locality planning.
- Concurrent study workers need a supported shared trial-storage design and correct coordination, not a fragile local-only backend.
Explanation
Correct Answer
C:The technical constraint is specific to the reported failure. Combining Spark feature preparation with Ray trial execution requires explicit materialization and data locality planning. To test it, measure repeated dataset exports and per-trial deserialization relative to model-fitting time.
Incorrect Answers
A:That underlying principle is directed at a different scaling and tuning situation: A language model's weights cannot fit on one accelerator yet a team launches pure data-parallel replicas, each loading the entire model. The current incident instead turns on combining Spark feature preparation with Ray trial execution requires explicit materialization and data locality planning.
B:That underlying principle is directed at a different scaling and tuning situation: A team runs independent GPU-heavy neural trials sequentially, despite having several isolated GPU workers that can fit each model. The current incident instead turns on combining Spark feature preparation with Ray trial execution requires explicit materialization and data locality planning.
D:That underlying principle is directed at a different scaling and tuning situation: A researcher parallelizes Optuna trials while sharing one writable SQLite file on a fragile network mount and observes duplicate reservations and storage locks. The current incident instead turns on combining Spark feature preparation with Ray trial execution requires explicit materialization and data locality planning.
Question 979
In an enterprise ML workflow, the reported issue is: An engineer logs a customer-specific API token inside the pickled PythonModel object before registering it to Unity Catalog. Which change addresses the real cause without undermining other requirements?
- Use a properly scoped authorized credential and supported authenticated inference request.
- Format requests according to the supported MLflow Deployments client and registered model input schema.
- Remove embedded secrets and configure an approved scoped secret retrieval mechanism for the serving process.
- Poll or otherwise verify documented readiness, then perform a permitted smoke prediction before marking the release ready.
Explanation
Correct Answer
C:This corrects the relevant failure boundary instead of merely hiding the symptom. Remove embedded secrets and configure an approved scoped secret retrieval mechanism for the serving process. The issue matters because anyone allowed to download the model artifact can extract credentials beyond the model's intended data access.
Incorrect Answers
A:That corrective action is directed at a different custom model serving situation: A client invokes a registered custom ML model using an unauthenticated HTTP POST to the serving endpoint and assumes network access implies authorization. The relevant corrective action here is remove embedded secrets and configure an approved scoped secret retrieval mechanism for the serving process.
B:That corrective action is directed at a different custom model serving situation: A notebook uses the MLflow Deployments SDK to send a column-major payload to a custom model endpoint expecting a different documented dataframe schema. The relevant corrective action here is remove embedded secrets and configure an approved scoped secret retrieval mechanism for the serving process.
D:That corrective action 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. The relevant corrective action here is remove embedded secrets and configure an approved scoped secret retrieval mechanism for the serving process.
Question 980
In an enterprise ML workflow, the reported issue is: A recommender experiment computes all evaluation metrics on training interactions and reports the result as generalization to new customers. What likely operational or model-quality harm does this issue create?
- Pipeline fitting fails with missing-column or mismatched-schema errors.
- The workflow fails or includes unnecessary invalid fit calls that cannot create a tokenizer model.
- Deployment performance may collapse on customers never seen during optimization.
- A rare new category can fail an entire scoring batch rather than produce a controlled result.
Explanation
Correct Answer
C:This consequence follows from the specific unresolved condition. Deployment performance may collapse on customers never seen during optimization. The causal constraint is: Training-set fit does not establish out-of-sample predictive quality or cold-start behavior.
Incorrect Answers
A:That direct consequence is directed at a different using spark ml situation: A pipeline engineer fits a vectorizer after an estimator in the stage list and then tries to use the estimator's prediction as the vectorizer input. The more immediate risk here is deployment performance may collapse on customers never seen during optimization.
B:That direct consequence is directed at a different using spark ml situation: A text pipeline repeatedly fits a Tokenizer stage even though tokenization simply converts strings into token arrays without learning a model. The more immediate risk here is deployment performance may collapse on customers never seen during optimization.
D:That direct consequence is directed at a different using spark ml situation: An inference pipeline sends new categorical labels unseen during training to StringIndexerModel with error handling configured to reject unknown categories. The more immediate risk here is deployment performance may collapse on customers never seen during optimization.