Databricks Certified Machine Learning Professional Practice Test 58
Question 570
In an enterprise ML workflow, the reported issue is: A research team captures only daily batch inference totals with no row-level timestamps but insists on per-minute live inference drift analysis. Before modifying the pipeline, which evidence should be collected first?
- Stratify historical samples by day type and examine whether the observed difference is expected seasonality.
- Inspect effective owner permissions, catalog/schema access and creation error logs for the target table.
- Inspect whether row-level event timestamps and prediction records actually exist for the desired minute windows.
- Compare regional event counts and failure proportions against the aggregate alert trigger.
Explanation
Correct Answer
C:This is the most discriminating check for the observed symptoms. Inspect whether row-level event timestamps and prediction records actually exist for the desired minute windows. The underlying technical issue is that monitoring granularity cannot exceed the information preserved in source rows and ingestion cadence.
Incorrect Answers
A:That diagnostic evidence 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. For the stated incident, the decisive observation is inspect whether row-level event timestamps and prediction records actually exist for the desired minute windows.
B:That diagnostic evidence is directed at a different drift detection and lakehouse monitoring situation: A monitor is pointed at a Unity Catalog table but the automation identity lacks the privileges needed to generate profile outputs. For the stated incident, the decisive observation is inspect whether row-level event timestamps and prediction records actually exist for the desired minute windows.
D:That diagnostic evidence 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. For the stated incident, the decisive observation is inspect whether row-level event timestamps and prediction records actually exist for the desired minute windows.
Question 571
The engineering team encounters the following issue: 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 technical principle most directly explains the root of the issue?
- Traffic-routing configuration is an explicit version allocation and must match the authorized canary percentage.
- A canary rollout can expose a small controlled share of live traffic while retaining a proven model for most requests.
- Random traffic splitting alone does not guarantee policy-based tenant or regional routing.
- Rare-outcome evaluation requires sufficient exposure, observation time and mature labels before concluding parity.
Explanation
Correct Answer
A:The technical constraint is specific to the reported failure. Traffic-routing configuration is an explicit version allocation and must match the authorized canary percentage. To test it, read the serving endpoint's active model versions and actual traffic percentages after update.
Incorrect Answers
B:That underlying principle 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. The current incident instead turns on traffic-routing configuration is an explicit version allocation and must match the authorized canary percentage.
C:That underlying principle 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. The current incident instead turns on traffic-routing configuration is an explicit version allocation and must match the authorized canary percentage.
D:That underlying principle is directed at a different deployment strategies situation: A candidate fraud model has near-identical average error to the champion, but an organization wants to compare rare fraudulent transactions during a brief canary. The current incident instead turns on traffic-routing configuration is an explicit version allocation and must match the authorized canary percentage.
Question 572
A production ML review contains this situation: A team switches a SparkML training stage to Ray without verifying that its chosen estimator supports Ray's distributed fit contracts. Which check shows that this particular problem has been resolved?
- Minimal groups with empty, rare and common categories all produce schema-valid results.
- Tracing shows the intended supported distributed training path with correct outputs and resource use.
- Ray reports parallel trial schedules that respect assigned GPUs and complete the planned search.
- A single training step completes within per-device memory limits under the chosen model placement.
Explanation
Correct Answer
B:This verification measures the required result rather than just process liveness. Tracing shows the intended supported distributed training path with correct outputs and resource use. It confirms whether the targeted action succeeded: retain SparkML for its supported distributed estimator or migrate to a validated Ray-compatible training algorithm.
Incorrect Answers
A:That acceptance test is directed at a different scaling and tuning situation: A per-city pandas UDF returns different prediction schemas for cities with absent categories, causing a distributed aggregation failure. Here the actual success criterion is tracing shows the intended supported distributed training path with correct outputs and resource use.
C:That acceptance test 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. Here the actual success criterion is tracing shows the intended supported distributed training path with correct outputs and resource use.
D:That acceptance test is directed at a different scaling and tuning situation: 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. Here the actual success criterion is tracing shows the intended supported distributed training path with correct outputs and resource use.
Question 573
The engineering team encounters the following issue: During a controlled release, the candidate model's P95 latency triples and engineers want to change the registered model alias without adjusting endpoint routing. What direct consequence should the owner prioritize if no fix is made?
- A premature full rollout promotes a model before enough rare fraud outcomes can be observed.
- Regulated traffic reaches an unapproved model version contrary to residency controls.
- High-latency candidate traffic continues despite a registry metadata change.
- The new version overwhelms replicas and creates customer-facing timeouts despite good offline accuracy.
Explanation
Correct Answer
C:This consequence follows from the specific unresolved condition. High-latency candidate traffic continues despite a registry metadata change. The causal constraint is: Rollout recovery must address the actual endpoint traffic configuration, not assume an alias automatically changes every active route.
Incorrect Answers
A:That direct consequence is directed at a different deployment strategies situation: A candidate fraud model has near-identical average error to the champion, but an organization wants to compare rare fraudulent transactions during a brief canary. The more immediate risk here is high-latency candidate traffic continues despite a registry metadata change.
B:That direct consequence 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. The more immediate risk here is high-latency candidate traffic continues despite a registry metadata change.
D:That direct consequence is directed at a different deployment strategies situation: A new recommendation model consumes three times as much memory and the team plans a canary based solely on prediction accuracy, ignoring serving capacity. The more immediate risk here is high-latency candidate traffic continues despite a registry metadata change.
Question 574
Review this production model incident: An ML owner adds a custom metric computed from the average of a string-typed currency column without validating conversion. What evidence would best distinguish the actual fault from neighboring ML concerns?
- Inspect metric reference table identity and original model training-set distribution for the deployed version.
- Inspect source column schema, conversion failures and metric SQL behavior for malformed strings.
- Compare historical baselines, current seasonal patterns and change approvals for the existing threshold.
- Correlate 5xx status bursts with request rate, resource use and deployment events.
Explanation
Correct Answer
B:This is the most discriminating check for the observed symptoms. Inspect source column schema, conversion failures and metric SQL behavior for malformed strings. The underlying technical issue is that custom monitor metrics need defined input types and missing/invalid-value semantics.
Incorrect Answers
A:That diagnostic evidence 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. For the stated incident, the decisive observation is inspect source column schema, conversion failures and metric SQL behavior for malformed strings.
C:That diagnostic evidence 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. For the stated incident, the decisive observation is inspect source column schema, conversion failures and metric SQL behavior for malformed strings.
D:That diagnostic evidence is directed at a different drift detection and lakehouse monitoring situation: A prediction endpoint returns HTTP 503 for intermittent bursts while the model's offline accuracy remains unchanged. For the stated incident, the decisive observation is inspect source column schema, conversion failures and metric SQL behavior for malformed strings.
Question 575
Consider the following fictional Databricks production case: 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. What consequence most specifically follows from leaving the issue unchanged?
- A scheduler misinterprets missing labels as performance collapse and promotes an untested model.
- Random sampling noise causes frequent unnecessary production model swaps.
- The winning model cannot serve requests on time or within allocated capacity.
- A statistically better average model causes larger losses for the protected operational segment.
Explanation
Correct Answer
A:This consequence follows from the specific unresolved condition. A scheduler misinterprets missing labels as performance collapse and promotes an untested model. The causal constraint is: Ground-truth delay constrains performance-triggered retraining; fresh unlabeled predictions cannot establish predictive accuracy.
Incorrect Answers
B:That direct consequence 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 more immediate risk here is a scheduler misinterprets missing labels as performance collapse and promotes an untested model.
C:That direct consequence 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 more immediate risk here is a scheduler misinterprets missing labels as performance collapse and promotes an untested model.
D:That direct consequence 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. The more immediate risk here is a scheduler misinterprets missing labels as performance collapse and promotes an untested model.
Question 576
The engineering team encounters the following issue: 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. Which technical principle most directly explains the root of the issue?
- Model Serving deployments must validate backward compatibility of request schema across existing clients.
- Rollout recovery must address the actual endpoint traffic configuration, not assume an alias automatically changes every active route.
- A canary rollout can expose a small controlled share of live traffic while retaining a proven model for most requests.
- Traffic-routing configuration is an explicit version allocation and must match the authorized canary percentage.
Explanation
Correct Answer
C:The technical constraint is specific to the reported failure. A canary rollout can expose a small controlled share of live traffic while retaining a proven model for most requests. To test it, compare serving-error, latency and outcome metrics separately for champion and canary model versions at the proposed traffic share.
Incorrect Answers
A:That underlying principle is directed at a different deployment strategies situation: A rollout updates the serving model but a renamed input column breaks requests from an older mobile client. The current incident instead turns on a canary rollout can expose a small controlled share of live traffic while retaining a proven model for most requests.
B:That underlying principle is directed at a different deployment strategies situation: During a controlled release, the candidate model's P95 latency triples and engineers want to change the registered model alias without adjusting endpoint routing. The current incident instead turns on a canary rollout can expose a small controlled share of live traffic while retaining a proven model for most requests.
D:That underlying principle is directed at a different deployment strategies 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%. The current incident instead turns on a canary rollout can expose a small controlled share of live traffic while retaining a proven model for most requests.
Question 577
In an enterprise ML workflow, the reported issue is: During a controlled release, the candidate model's P95 latency triples and engineers want to change the registered model alias without adjusting endpoint routing. How should the engineer validate the correction?
- All synthetic protected-region requests remain on their authorized version during canary traffic.
- A controlled failback restores the previous version without rebuilding model artifacts.
- The candidate remains within memory and latency headroom at each authorized traffic increment.
- Sampled live requests and endpoint configuration confirm candidate traffic is removed or bounded as approved.
Explanation
Correct Answer
D:This verification measures the required result rather than just process liveness. Sampled live requests and endpoint configuration confirm candidate traffic is removed or bounded as approved. It confirms whether the targeted action succeeded: reduce candidate routing or restore champion traffic using the supported endpoint configuration rollback.
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 sampled live requests and endpoint configuration confirm candidate traffic is removed or bounded as approved.
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 sampled live requests and endpoint configuration confirm candidate traffic is removed or bounded as approved.
C:That acceptance test is directed at a different deployment strategies situation: A new recommendation model consumes three times as much memory and the team plans a canary based solely on prediction accuracy, ignoring serving capacity. Here the actual success criterion is sampled live requests and endpoint configuration confirm candidate traffic is removed or bounded as approved.
Question 578
The engineering team encounters the following issue: 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. Which technical principle most directly explains the root of the issue?
- Data parallelism duplicates model parameters, while model parallelism or sharding distributes parameters when required.
- Concurrent study workers need a supported shared trial-storage design and correct coordination, not a fragile local-only backend.
- Vertical versus horizontal scaling decisions must consider dataset size, execution architecture and fixed coordination overhead.
- Separate Spark tasks do not safely share or synchronously mutate one Python model instance across workers.
Explanation
Correct Answer
D:The technical constraint is specific to the reported failure. Separate Spark tasks do not safely share or synchronously mutate one Python model instance across workers. To test it, trace parameter ownership and update communication in the proposed training algorithm.
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 separate Spark tasks do not safely share or synchronously mutate one Python model instance across workers.
B: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 separate Spark tasks do not safely share or synchronously mutate one Python model instance across workers.
C:That underlying principle is directed at a different scaling and tuning situation: A moderately sized dataset fits on one worker but a team provisions a huge distributed cluster and observes overhead higher than model fit. The current incident instead turns on separate Spark tasks do not safely share or synchronously mutate one Python model instance across workers.
Question 579
Consider the following fictional Databricks production case: A distributed retraining workflow uses ephemeral personal compute and fails when the original engineer's account is disabled. What underlying principle determines the most appropriate interpretation?
- Declarative automation or asset-bundle definitions can version environment-specific ML assets and reproduce deployments.
- Bundle target parameters must resolve to explicitly authorized environment-specific resources.
- Production ML jobs need owned service identities, durable environments and resource configuration independent of individuals.
- Product naming and resource schema changes should be checked against the active Databricks CLI and documentation for the deployed version.
Explanation
Correct Answer
C:The technical constraint is specific to the reported failure. Production ML jobs need owned service identities, durable environments and resource configuration independent of individuals. To test it, inspect job ownership, compute policy and identity access after a simulated employee offboarding.
Incorrect Answers
A:That underlying principle is directed at a different environment architectures situation: A model serving endpoint is created manually in the UI and every staging deployment differs subtly from production because no deployment definition is saved. The current incident instead turns on production ML jobs need owned service identities, durable environments and resource configuration independent of individuals.
B:That underlying principle is directed at a different environment architectures situation: A bundle references a model registry name without qualifying its environment, and a development deployment accidentally updates production registered-model aliases. The current incident instead turns on production ML jobs need owned service identities, durable environments and resource configuration independent of individuals.
D:That underlying principle is directed at a different environment architectures situation: An ML platform moves its configuration from older Databricks Asset Bundles terminology to newer Declarative Automation Bundles references and assumes every resource schema stayed identical. The current incident instead turns on production ML jobs need owned service identities, durable environments and resource configuration independent of individuals.
Question 580
Review this production model incident: 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 mitigation resolves the defect rather than concealing it?
- Send a small approved percentage of eligible live requests to the candidate and expand only after defined gates pass.
- Set a statistically and operationally adequate rollout observation period with strict safety abort thresholds.
- Set and verify the intended approved candidate-versus-champion traffic split before exposure.
- Prepare a complete tested green environment and switch traffic after approval while retaining blue for rollback.
Explanation
Correct Answer
C:This corrects the relevant failure boundary instead of merely hiding the symptom. Set and verify the intended approved candidate-versus-champion traffic split before exposure. The issue matters because all requests are exposed to the candidate during what operators believed was a small canary.
Incorrect Answers
A:That corrective action 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. The relevant corrective action here is set and verify the intended approved candidate-versus-champion traffic split before exposure.
B:That corrective action is directed at a different deployment strategies situation: A candidate fraud model has near-identical average error to the champion, but an organization wants to compare rare fraudulent transactions during a brief canary. The relevant corrective action here is set and verify the intended approved candidate-versus-champion traffic split before exposure.
D:That corrective action 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. The relevant corrective action here is set and verify the intended approved candidate-versus-champion traffic split before exposure.