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 17

Question 160

Review this production model incident: An image model needs 90 GB accelerator memory for one training replica while each available GPU has 24 GB and engineers propose only adding more identical nodes. Which failure risk follows most closely from the unresolved behavior?

  1. Unnecessary model partitioning adds network latency with little memory benefit.
  2. The rewritten job runs only on one worker or fails during model serialization.
  3. Every replica hits device OOM even though aggregate cluster memory appears ample.
  4. Multiple trials contend for the same accelerator and fail with memory exhaustion.

Explanation

Correct Answer

C:

This consequence follows from the specific unresolved condition. Every replica hits device OOM even though aggregate cluster memory appears ample. The causal constraint is: Horizontal replica count alone does not combine GPU memory for one model without a supported parallelization strategy.

Incorrect Answers

A:

That direct consequence is directed at a different scaling and tuning situation: 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. The more immediate risk here is every replica hits device OOM even though aggregate cluster memory appears ample.

B:

That direct consequence is directed at a different scaling and tuning situation: A team switches a SparkML training stage to Ray without verifying that its chosen estimator supports Ray's distributed fit contracts. The more immediate risk here is every replica hits device OOM even though aggregate cluster memory appears ample.

D:

That direct consequence is directed at a different scaling and tuning situation: A Ray Tune training function requests zero GPU resources but initializes a large GPU model inside every trial. The more immediate risk here is every replica hits device OOM even though aggregate cluster memory appears ample.

Question 161

Review this production model incident: A credit model's true-label feedback arrives 45 days late, yet the retraining scheduler assumes yesterday's unlabeled predictions provide yesterday's observed accuracy. Which mitigation resolves the defect rather than concealing it?

  1. Evaluate all eligible candidates against one controlled frozen benchmark and rank only comparable results.
  2. Rank only eligible candidates that pass accuracy and operational SLA gates before promotion.
  3. Compute performance only on matured labeled cohorts and use separate permitted unlabeled drift signals where relevant.
  4. Require a meaningful quality improvement supported by adequate evidence before changing the champion.

Explanation

Correct Answer

C:

This corrects the relevant failure boundary instead of merely hiding the symptom. Compute performance only on matured labeled cohorts and use separate permitted unlabeled drift signals where relevant. The issue matters because a scheduler misinterprets missing labels as performance collapse and promotes an untested model.

Incorrect Answers

A:

That corrective action is directed at a different automated retraining situation: A retraining pipeline ranks candidate models using metrics calculated on different dataset versions, then selects the numerically highest result. The relevant corrective action here is compute performance only on matured labeled cohorts and use separate permitted unlabeled drift signals where relevant.

B:

That corrective action is directed at a different automated retraining situation: An AutoML training job always chooses the model with the lowest training loss, even when the deployed service has a strict memory and P95 latency budget. The relevant corrective action here is compute performance only on matured labeled cohorts and use separate permitted unlabeled drift signals where relevant.

D:

That corrective action is directed at a different automated retraining 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. The relevant corrective action here is compute performance only on matured labeled cohorts and use separate permitted unlabeled drift signals where relevant.

Question 162

Review this production model incident: A monitor is pointed at a Unity Catalog table but the automation identity lacks the privileges needed to generate profile outputs. Which mitigation resolves the defect rather than concealing it?

  1. Grant only the documented required monitoring permissions and configure approved metric destinations.
  2. Retain a stable reference comparison in addition to short-term change detection.
  3. Add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.
  4. Route availability errors to serving reliability remediation, not to automatic training.

Explanation

Correct Answer

A:

This corrects the relevant failure boundary instead of merely hiding the symptom. Grant only the documented required monitoring permissions and configure approved metric destinations. The issue matters because the monitor exists in configuration but stops producing metrics for operational review.

Incorrect Answers

B:

That corrective action 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 relevant corrective action here is grant only the documented required monitoring permissions and configure approved metric destinations.

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 grant only the documented required monitoring permissions and configure approved metric destinations.

D:

That corrective action is directed at a different drift detection and lakehouse monitoring situation: An endpoint's error rate spikes while data profiles look normal, and responders attempt to retrain the model instead of checking serving infrastructure. The relevant corrective action here is grant only the documented required monitoring permissions and configure approved metric destinations.

Question 163

The engineering team encounters the following issue: A feature refresh writes duplicate records for each entity and logical timestamp after retries, breaking unique lookups. What investigation would most effectively isolate this behavior?

  1. Compare offline source update timestamps, online publication timestamps and stated staleness limits.
  2. Check entity/timestamp duplicate counts around job retries and writer transaction history.
  3. Map feature consumers, source commit IDs and before/after results for a fixed regression fixture.
  4. Compare event timestamps, ingestion timestamps and window assignment for late synthetic readings.

Explanation

Correct Answer

B:

This is the most discriminating check for the observed symptoms. Check entity/timestamp duplicate counts around job retries and writer transaction history. The underlying technical issue is that idempotent feature pipeline writes need stable entity/time keys and safe retry semantics.

Incorrect Answers

A:

That diagnostic evidence is directed at a different advanced feature store concepts situation: A team publishes an online feature table but neglects its refresh schedule while the predictive model assumes values no older than 15 minutes. For the stated incident, the decisive observation is check entity/timestamp duplicate counts around job retries and writer transaction history.

C:

That diagnostic evidence 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. For the stated incident, the decisive observation is check entity/timestamp duplicate counts around job retries and writer transaction history.

D:

That diagnostic evidence 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. For the stated incident, the decisive observation is check entity/timestamp duplicate counts around job retries and writer transaction history.

Question 164

Review this production model incident: A batch scoring service loads a classifier version from last month while using today's new feature indexer, silently changing categorical positions. What is the central engineering constraint behind this behavior?

  1. Scoring should use preprocessing components trained and versioned with the exact estimator artifact.
  2. Any fitted preprocessing stage can learn from holdout records and must be fitted within the training boundary.
  3. VectorAssembler is a Transformer that creates a vector column, not a trainable predictor.
  4. Temporal evaluation must respect what data would have been available at the forecast origin, not simply use random folds.

Explanation

Correct Answer

A:

The technical constraint is specific to the reported failure. Scoring should use preprocessing components trained and versioned with the exact estimator artifact. To test it, compare saved preprocessing stage metadata and feature vector order across the two deployed artifacts.

Incorrect Answers

B:

That underlying principle is directed at a different using spark ml situation: A credit-risk pipeline fits StringIndexer on the entire dataset before splitting train and holdout, and rare holdout labels influence its learned category mapping. The current incident instead turns on scoring should use preprocessing components trained and versioned with the exact estimator artifact.

C:

That underlying principle 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 current incident instead turns on scoring should use preprocessing components trained and versioned with the exact estimator artifact.

D:

That underlying principle is directed at a different using spark ml situation: A time-ordered customer-retention dataset includes events from the future, yet a developer randomly cross-validates across time and then claims a forward-looking forecast is unbiased. The current incident instead turns on scoring should use preprocessing components trained and versioned with the exact estimator artifact.

Question 165

Review this production model incident: An alert from a feature monitor automatically replaces the live model with a newly trained artifact without testing or approval. What evidence would best distinguish the actual fault from neighboring ML concerns?

  1. Compare champion and candidate per-segment false-negative rates on equivalent untouched evaluation cohorts.
  2. Compare candidate predictive metrics with model footprint, serving latency and expected concurrency.
  3. Inspect evaluation sample size, confidence interval or bootstrap variation around the candidate-champion delta.
  4. Trace the workflow after a drift alert, including training, evaluation, registry and endpoint changes.

Explanation

Correct Answer

D:

This is the most discriminating check for the observed symptoms. Trace the workflow after a drift alert, including training, evaluation, registry and endpoint changes. The underlying technical issue is that retraining triggers and production promotion are different stages with separate evaluation and rollout gates.

Incorrect Answers

A:

That diagnostic evidence is directed at a different automated retraining situation: A new fraud classifier improves overall F1 but doubles false negatives for high-value payment accounts, and an automated policy chooses it on F1 alone. For the stated incident, the decisive observation is trace the workflow after a drift alert, including training, evaluation, registry and endpoint changes.

B:

That diagnostic evidence is directed at a different automated retraining situation: An AutoML training job always chooses the model with the lowest training loss, even when the deployed service has a strict memory and P95 latency budget. For the stated incident, the decisive observation is trace the workflow after a drift alert, including training, evaluation, registry and endpoint changes.

C:

That diagnostic evidence is directed at a different automated retraining 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. For the stated incident, the decisive observation is trace the workflow after a drift alert, including training, evaluation, registry and endpoint changes.

Question 166

The engineering team encounters the following issue: A production endpoint scales CPU instances up when traffic rises, but request queues and P95 response time keep growing despite low CPU usage. What direct consequence should the owner prioritize if no fix is made?

  1. A material disparity is hidden inside the overall average model performance.
  2. The deployment wastes compute while inference delays and timeouts continue increasing.
  3. Errors are assigned to the wrong predictions and reported performance becomes meaningless.
  4. Routine weekend traffic launches unnecessary retraining and distracts engineers.

Explanation

Correct Answer

B:

This consequence follows from the specific unresolved condition. The deployment wastes compute while inference delays and timeouts continue increasing. The causal constraint is: Endpoint throughput can be constrained by queueing, request concurrency, external features or rate limits beyond CPU utilization.

Incorrect Answers

A:

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 the deployment wastes compute while inference delays and timeouts continue increasing.

C:

That direct consequence 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. The more immediate risk here is the deployment wastes compute while inference delays and timeouts continue increasing.

D:

That direct consequence 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 more immediate risk here is the deployment wastes compute while inference delays and timeouts continue increasing.

Question 167

Consider the following fictional Databricks production case: A SparkML grid-search experiment selects the model with the largest error metric even though the evaluator reports RMSE to minimize. What underlying principle determines the most appropriate interpretation?

  1. Pipeline stages execute in declared sequence and each estimator expects columns already produced upstream.
  2. Parameter search must use the correct evaluator direction for the chosen loss or utility metric.
  3. Accuracy alone is misleading under extreme class imbalance; minority precision, recall and PR metrics carry critical information.
  4. Repeated driver collection defeats the distributed processing advantages relevant to large SparkML feature pipelines.

Explanation

Correct Answer

B:

The technical constraint is specific to the reported failure. Parameter search must use the correct evaluator direction for the chosen loss or utility metric. To test it, inspect evaluator.isLargerBetter or equivalent metric semantics alongside the ordered candidate RMSE values.

Incorrect Answers

A:

That underlying principle 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 current incident instead turns on parameter search must use the correct evaluator direction for the chosen loss or utility metric.

C:

That underlying principle is directed at a different using spark ml situation: A fraud detector predicts the majority no-fraud class for 99.4% of cases and advertises 99.4% accuracy while missing most positives. The current incident instead turns on parameter search must use the correct evaluator direction for the chosen loss or utility metric.

D:

That underlying principle is directed at a different using spark ml situation: An insurer already stores its risk features as a large Spark DataFrame but exports them into Python lists before every vector transformation and repeatedly hits out-of-memory errors. The current incident instead turns on parameter search must use the correct evaluator direction for the chosen loss or utility metric.

Question 168

In an enterprise ML workflow, the reported issue is: An ML project stores test files directly in each production notebook and execution of a deployed notebook inadvertently reruns expensive tests. Before modifying the pipeline, which evidence should be collected first?

  1. Compare expected and actual normalized timestamps for UTC, offset and daylight-transition examples.
  2. Inspect test dataset sensitivity, workspace access rights and synthetic alternative fixtures.
  3. Inspect notebook entrypoints, test invocation conditions and deployment job definitions.
  4. Trace a synthetic labeled record from feature source through registered model to endpoint request and output.

Explanation

Correct Answer

C:

This is the most discriminating check for the observed symptoms. Inspect notebook entrypoints, test invocation conditions and deployment job definitions. The underlying technical issue is that test organization should separate production orchestration from the test runner and control execution stages.

Incorrect Answers

A:

That diagnostic evidence is directed at a different validation testing situation: 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. For the stated incident, the decisive observation is inspect notebook entrypoints, test invocation conditions and deployment job definitions.

B:

That diagnostic evidence 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. For the stated incident, the decisive observation is inspect notebook entrypoints, test invocation conditions and deployment job definitions.

D:

That diagnostic evidence is directed at a different validation testing situation: A churn feature job succeeds and a classifier achieves acceptable F1, but the deployed endpoint returns HTTP 400 because request column names differ. For the stated incident, the decisive observation is inspect notebook entrypoints, test invocation conditions and deployment job definitions.

Question 169

The engineering team encounters the following issue: 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. Which experiment most directly demonstrates a successful fix?

  1. A two-minute spike does not page while a sustained ten-minute breach does.
  2. All accepted test timestamps land in the correct expected monitoring interval.
  3. A synthetic v8-only error increase appears in its own model-version metric.
  4. Matching-filter fixtures show no artificial drift while a real within-segment change remains detectable.

Explanation

Correct Answer

A:

This verification measures the required result rather than just process liveness. A two-minute spike does not page while a sustained ten-minute breach does. It confirms whether the targeted action succeeded: add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.

Incorrect Answers

B:

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 a two-minute spike does not page while a sustained ten-minute breach does.

C:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A scoring endpoint quietly moves from model v7 to v8 during a rollout but an inference trend aggregates both versions in one error curve. Here the actual success criterion is a two-minute spike does not page while a sustained ten-minute breach does.

D:

That acceptance test is directed at a different drift detection and lakehouse monitoring situation: A customer-risk monitoring job compares an unfiltered current high-risk segment to an all-customer historical baseline and interprets the difference as temporal drift. Here the actual success criterion is a two-minute spike does not page while a sustained ten-minute breach does.

Question 170

The engineering team encounters the following issue: A batch scoring pipeline trains one per-region model and then joins model predictions back using region alone, duplicating rows because each region contains multiple versioned models. What corrective action should the engineer prefer?

  1. Bind the selected model version to each group and join predictions on a unique provenance key.
  2. Increase the training host's memory within budget or change to a supported distributed estimator.
  3. Retain SparkML for its supported distributed estimator or migrate to a validated Ray-compatible training algorithm.
  4. Adopt validated model-parallel or sharded training to fit the model and its optimizer state.

Explanation

Correct Answer

A:

This corrects the relevant failure boundary instead of merely hiding the symptom. Bind the selected model version to each group and join predictions on a unique provenance key. The issue matters because scores multiply or silently mix old and new model versions.

Incorrect Answers

B:

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 bind the selected model version to each group and join predictions on a unique provenance key.

C:

That corrective action is directed at a different scaling and tuning situation: A team switches a SparkML training stage to Ray without verifying that its chosen estimator supports Ray's distributed fit contracts. The relevant corrective action here is bind the selected model version to each group and join predictions on a unique provenance key.

D:

That corrective action 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 relevant corrective action here is bind the selected model version to each group and join predictions on a unique provenance key.

Certified Machine Learning Professional Practice Tests