cert
cert-1
cert-2

Pass Databricks Certified Machine Learning Professional Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers to Pass the Actual Exam!
30 Days Free Updates, Instant Download!

cert-5
cert-6
Certified Machine Learning Professional Exam - Verified By Experts
Certified Machine Learning Professional Premium File

Certified Machine Learning Professional Premium File

$69.99
$76.99
  • Premium File 216 Questions & Answers. Last Update: Nov 12, 2026

Whats Included:

  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates
 
$76.99
$69.99
accept 12 downloads in the last 7 days
block-screenshots
Databricks Certified Machine Learning Professional Practice Test 90

Question 890

In an enterprise ML workflow, the reported issue is: A binary model's average predicted probability remains 0.4, but outcomes show that 0.8 predictions are now correct only half the time. What likely operational or model-quality harm does this issue create?

  1. The metric fails or silently excludes malformed observations, distorting the reported trend.
  2. Downstream risk thresholds overtrust predicted probabilities and misprice decisions.
  3. Gradual cumulative drift becomes invisible despite material divergence from the trained population.
  4. Small-cell disclosure and unstable rates cause privacy risk and misleading fairness decisions.

Explanation

Correct Answer

B:

This consequence follows from the specific unresolved condition. Downstream risk thresholds overtrust predicted probabilities and misprice decisions. The causal constraint is: Calibration drift requires comparing predicted probabilities with observed outcome frequencies, not just prediction means.

Incorrect Answers

A:

That direct consequence is directed at a different drift detection and lakehouse monitoring situation: An ML owner adds a custom metric computed from the average of a string-typed currency column without validating conversion. The more immediate risk here is downstream risk thresholds overtrust predicted probabilities and misprice decisions.

C:

That direct consequence is directed at a different drift detection and lakehouse monitoring situation: A streaming payments monitor always compares today with the previous hour, and a slow week-long distribution drift escapes because adjacent hours differ only slightly. The more immediate risk here is downstream risk thresholds overtrust predicted probabilities and misprice decisions.

D:

That direct consequence is directed at a different drift detection and lakehouse monitoring situation: A demographic fairness review uses a tiny sensitive subgroup with three labeled examples and reports its observed error as definitive. The more immediate risk here is downstream risk thresholds overtrust predicted probabilities and misprice decisions.

Question 891

Consider the following fictional Databricks production case: A preprocessing function divides by a standard deviation and returns infinity for a constant numeric column. Which implementation is most appropriate for the stated incident?

  1. Require isolated unit tests in development followed by separate integration and production smoke gates.
  2. Guard zero-variance inputs and assert expected fallback or explicit failure semantics in unit tests.
  3. Add a full-path integration test with explicit time-zone and cutoff assertions.
  4. Centralize versioned validation functions and test behavior under all consuming environments.

Explanation

Correct Answer

B:

This corrects the relevant failure boundary instead of merely hiding the symptom. Guard zero-variance inputs and assert expected fallback or explicit failure semantics in unit tests. The issue matters because invalid standardized values contaminate model input vectors or crash scoring.

Incorrect Answers

A:

That corrective action 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 relevant corrective action here is guard zero-variance inputs and assert expected fallback or explicit failure semantics in unit tests.

C:

That corrective action 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 relevant corrective action here is guard zero-variance inputs and assert expected fallback or explicit failure semantics in unit tests.

D:

That corrective action is directed at a different validation testing situation: An engineering group duplicates a validation helper across six notebooks, and two copies enforce outdated threshold values. The relevant corrective action here is guard zero-variance inputs and assert expected fallback or explicit failure semantics in unit tests.

Question 892

A production ML review contains this situation: A logistics model must score 60 million daily routes already in Delta but the scoring script collects each route into Python before calling predict. Which statement best explains the failure mode?

  1. Ranking quality and operating threshold quality are distinct; precision and recall depend on the chosen cutoff.
  2. Distributed batch inference can apply a saved SparkML PipelineModel directly to a DataFrame without collecting every row.
  3. Parameter search must use the correct evaluator direction for the chosen loss or utility metric.
  4. A scheduled distributed batch inference job can better fit offline throughput and cost than always-on serving.

Explanation

Correct Answer

B:

The technical constraint is specific to the reported failure. Distributed batch inference can apply a saved SparkML PipelineModel directly to a DataFrame without collecting every row. To test it, inspect the scoring job for collect and driver memory growth versus distributed transform stages.

Incorrect Answers

A:

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 distributed batch inference can apply a saved SparkML PipelineModel directly to a DataFrame without collecting every row.

C:

That underlying principle is directed at a different using spark ml situation: A SparkML grid-search experiment selects the model with the largest error metric even though the evaluator reports RMSE to minimize. The current incident instead turns on distributed batch inference can apply a saved SparkML PipelineModel directly to a DataFrame without collecting every row.

D:

That underlying principle is directed at a different using spark ml situation: A credit reporting task needs one million records scored once per week for offline account review and requests a permanently running interactive endpoint. The current incident instead turns on distributed batch inference can apply a saved SparkML PipelineModel directly to a DataFrame without collecting every row.

Question 893

In an enterprise ML workflow, the reported issue is: A production endpoint scales CPU instances up when traffic rises, but request queues and P95 response time keep growing despite low CPU usage. How should the engineer validate the correction?

  1. Weekend fixtures match appropriate seasonality while atypical weekend changes still alert.
  2. Queue length and tail latency fall under controlled peak traffic without unnecessary idle CPU provisioning.
  3. A known benign seasonal fixture stays below the revised threshold and a material anomaly exceeds it.
  4. Two fixtures with equal risk proportions but different volumes produce the same custom rate.

Explanation

Correct Answer

B:

This verification measures the required result rather than just process liveness. Queue length and tail latency fall under controlled peak traffic without unnecessary idle CPU provisioning. It confirms whether the targeted action succeeded: tune the proven bottleneck and concurrency/capacity settings instead of scaling CPU blindly.

Incorrect Answers

A:

That acceptance test 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. Here the actual success criterion is queue length and tail latency fall under controlled peak traffic without unnecessary idle CPU provisioning.

C:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A production alert threshold is hard-coded against last year's baseline, but seasonal traffic has shifted and the owner never reviewed the threshold. Here the actual success criterion is queue length and tail latency fall under controlled peak traffic without unnecessary idle CPU provisioning.

D:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A monitoring team wants a business metric for the share of transactions exceeding a risk threshold but records only the raw count of high-risk transactions. Here the actual success criterion is queue length and tail latency fall under controlled peak traffic without unnecessary idle CPU provisioning.

Question 894

A production ML review contains this situation: A new model beats the champion by 0.1 percentage points on a tiny evaluation set, and the team automatically promotes it despite large uncertainty. Which check shows that this particular problem has been resolved?

  1. A source-outage fixture triggers pipeline remediation rather than automatic model replacement.
  2. Repeated evaluation or confidence testing supports the chosen promotion decision at the required confidence.
  3. A controlled seasonal fixture is ignored while a material sustained shift triggers the planned workflow.
  4. An intentionally bad candidate is rejected while an approved candidate proceeds through controlled release.

Explanation

Correct Answer

B:

This verification measures the required result rather than just process liveness. Repeated evaluation or confidence testing supports the chosen promotion decision at the required confidence. It confirms whether the targeted action succeeded: require a meaningful quality improvement supported by adequate evidence before changing the champion.

Incorrect Answers

A:

That acceptance test is directed at a different automated retraining situation: A feature stream loses half its source events and a retraining policy treats the resulting input shift as proof that the learned concept changed. Here the actual success criterion is repeated evaluation or confidence testing supports the chosen promotion decision at the required confidence.

C:

That acceptance test is directed at a different automated retraining situation: An automated retraining job triggers whenever a numerical feature mean shifts by 0.1, regardless of normal seasonal changes or sample size. Here the actual success criterion is repeated evaluation or confidence testing supports the chosen promotion decision at the required confidence.

D:

That acceptance test is directed at a different automated retraining situation: An alert from a feature monitor automatically replaces the live model with a newly trained artifact without testing or approval. Here the actual success criterion is repeated evaluation or confidence testing supports the chosen promotion decision at the required confidence.

Question 895

Review this production model incident: A demographic fairness review uses a tiny sensitive subgroup with three labeled examples and reports its observed error as definitive. Which failure risk follows most closely from the unresolved behavior?

  1. Downstream risk thresholds overtrust predicted probabilities and misprice decisions.
  2. Material drift remains unaddressed because nobody receives the notification.
  3. A material disparity is hidden inside the overall average model performance.
  4. Small-cell disclosure and unstable rates cause privacy risk and misleading fairness decisions.

Explanation

Correct Answer

D:

This consequence follows from the specific unresolved condition. Small-cell disclosure and unstable rates cause privacy risk and misleading fairness decisions. The causal constraint is: Subgroup metrics should enforce privacy and minimum-sample limits before strong comparative claims.

Incorrect Answers

A:

That direct consequence is directed at a different drift detection and lakehouse monitoring situation: A binary model's average predicted probability remains 0.4, but outcomes show that 0.8 predictions are now correct only half the time. The more immediate risk here is small-cell disclosure and unstable rates cause privacy risk and misleading fairness decisions.

B:

That direct consequence 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 more immediate risk here is small-cell disclosure and unstable rates cause privacy risk and misleading fairness decisions.

C:

That direct consequence is directed at a different drift detection and lakehouse monitoring situation: A fairness reviewer needs the difference in false-negative rates between two approved cohorts, but the monitor computes an aggregate without cohort labels. The more immediate risk here is small-cell disclosure and unstable rates cause privacy risk and misleading fairness decisions.

Question 896

In an enterprise ML workflow, the reported issue is: A tabular classifier fits comfortably on one GPU and 100 million independent examples are available, but a team proposes splitting the small model's layers across hosts. Which change addresses the real cause without undermining other requirements?

  1. Use an appropriate Ray actor-based framework while reserving Spark for eligible data preprocessing.
  2. Configure supported Ray Tune trials with declared GPU and CPU requirements and controlled concurrency.
  3. Increase the training host's memory within budget or change to a supported distributed estimator.
  4. Use supported data parallelism with appropriately batched examples and convergence checks.

Explanation

Correct Answer

D:

This corrects the relevant failure boundary instead of merely hiding the symptom. Use supported data parallelism with appropriately batched examples and convergence checks. The issue matters because unnecessary model partitioning adds network latency with little memory benefit.

Incorrect Answers

A:

That corrective action is directed at a different scaling and tuning situation: A reinforcement-learning simulator requires stateful actors exchanging messages and irregular task graphs, but the team plans one Spark SQL groupBy per simulation step. The relevant corrective action here is use supported data parallelism with appropriately batched examples and convergence checks.

B:

That corrective action 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 relevant corrective action here is use supported data parallelism with appropriately batched examples and convergence checks.

C:

That corrective action is directed at a different scaling and tuning situation: A single-node model training run spends almost all time swapping a 40 GB feature matrix on a 16 GB VM; no distributed version exists. The relevant corrective action here is use supported data parallelism with appropriately batched examples and convergence checks.

Question 897

The engineering team encounters the following issue: A new merchant group contributes only 12 observations, yet a monitor reports a large categorical divergence from a baseline of 50,000 and the team assumes the result is definitive. Which experiment most directly demonstrates a successful fix?

  1. Low-count groups are flagged insufficient while adequately powered shifted groups are evaluated normally.
  2. A synthetic region-only failure triggers the scoped alert without misclassifying healthy regions.
  3. All accepted test timestamps land in the correct expected monitoring interval.
  4. A fixture with two model versions produces identifiable versioned prediction metrics when labels are available.

Explanation

Correct Answer

A:

This verification measures the required result rather than just process liveness. Low-count groups are flagged insufficient while adequately powered shifted groups are evaluated normally. It confirms whether the targeted action succeeded: apply documented sample-sufficiency checks and collect more representative data before strong conclusions.

Incorrect Answers

B:

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 low-count groups are flagged insufficient while adequately powered shifted groups are evaluated normally.

C:

That acceptance test 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. Here the actual success criterion is low-count groups are flagged insufficient while adequately powered shifted groups are evaluated normally.

D:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A churn model logs requests, predictions, model versions and later ground-truth labels, but monitoring is configured as a generic feature snapshot. Here the actual success criterion is low-count groups are flagged insufficient while adequately powered shifted groups are evaluated normally.

Question 898

Consider the following fictional Databricks production case: A churn model logs requests, predictions, model versions and later ground-truth labels, but monitoring is configured as a generic feature snapshot. Which result is the strongest acceptance criterion after remediation?

  1. A fixture with two model versions produces identifiable versioned prediction metrics when labels are available.
  2. A synthetic request can be traced to a versioned, privacy-safe inference record and matching endpoint response.
  3. A controlled new-category fixture changes the categorical drift signal without relying on event-count changes.
  4. An injected shift remains visible against the preserved baseline after repeated monitor runs.

Explanation

Correct Answer

A:

This verification measures the required result rather than just process liveness. A fixture with two model versions produces identifiable versioned prediction metrics when labels are available. It confirms whether the targeted action succeeded: choose a compatible inference profile and map the correct prediction, version and label columns.

Incorrect Answers

B:

That acceptance test 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. Here the actual success criterion is a fixture with two model versions produces identifiable versioned prediction metrics when labels are available.

C:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A customer product-category feature acquires many new category values and analysts try to detect the change using only the overall row count. Here the actual success criterion is a fixture with two model versions produces identifiable versioned prediction metrics when labels are available.

D:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A baseline table is overwritten every night with the current day's data before a drift calculation, making the comparison always look unchanged. Here the actual success criterion is a fixture with two model versions produces identifiable versioned prediction metrics when labels are available.

Question 899

Consider the following fictional Databricks production case: An Optuna optimization reuses the same MLflow run ID for concurrent trials, causing overlapping parameter values and artifacts. Which result is the strongest acceptance criterion after remediation?

  1. End-to-end throughput improves after removing the true serialized bottleneck.
  2. Measured model updates follow the supported synchronization contract across workers.
  3. A small actor simulation preserves state and message order across expected task dependencies.
  4. Each parallel trial resolves to distinct reproducible metrics and artifact storage.

Explanation

Correct Answer

D:

This verification measures the required result rather than just process liveness. Each parallel trial resolves to distinct reproducible metrics and artifact storage. It confirms whether the targeted action succeeded: assign each trial a unique tracked run with a shared parent experiment and stable trial metadata.

Incorrect Answers

A:

That acceptance test is directed at a different scaling and tuning situation: An ML project has a single sequential external feature-service dependency and expects data-parallel workers to increase throughput 20-fold. Here the actual success criterion is each parallel trial resolves to distinct reproducible metrics and artifact storage.

B:

That acceptance test is directed at a different scaling and tuning situation: An ML engineer tries to parallelize a stateful neural-training loop by executing independent mapPartitions calls that share one mutable global model in the driver. Here the actual success criterion is each parallel trial resolves to distinct reproducible metrics and artifact storage.

C:

That acceptance test is directed at a different scaling and tuning situation: A reinforcement-learning simulator requires stateful actors exchanging messages and irregular task graphs, but the team plans one Spark SQL groupBy per simulation step. Here the actual success criterion is each parallel trial resolves to distinct reproducible metrics and artifact storage.

Question 900

A production ML review contains this situation: An engineer uses only production replay data to validate a code change and exposes customer events in an unrestricted test workspace. What is the most material immediate risk in this situation?

  1. A hidden offset causes training-serving skew and potential future-data leakage.
  2. Invalid standardized values contaminate model input vectors or crash scoring.
  3. Small logic errors require slow integration jobs for every test iteration.
  4. Sensitive events leak into lower-trust test assets while making debugging harder.

Explanation

Correct Answer

D:

This consequence follows from the specific unresolved condition. Sensitive events leak into lower-trust test assets while making debugging harder. The causal constraint is: Test stages need appropriate data classification and isolation, not just increasingly realistic inputs.

Incorrect Answers

A:

That direct consequence 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 more immediate risk here is sensitive events leak into lower-trust test assets while making debugging harder.

B:

That direct consequence is directed at a different validation testing situation: A preprocessing function divides by a standard deviation and returns infinity for a constant numeric column. The more immediate risk here is sensitive events leak into lower-trust test assets while making debugging harder.

C:

That direct consequence is directed at a different validation testing situation: A team hides all preprocessing logic inside one long notebook, making it hard to test independently without starting a Spark cluster. The more immediate risk here is sensitive events leak into lower-trust test assets while making debugging harder.

Certified Machine Learning Professional Practice Tests