Databricks Certified Machine Learning Professional Practice Test 33
Question 320
Consider the following fictional Databricks production case: 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 implementation is most appropriate for the stated incident?
- Repair the profile computation and alert on stale metric output separately from the model's predictions.
- Add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.
- Join labels to their originating predictions using stable IDs and a defined maturity window.
- Improve authorized logging and table design before configuring fine-grained inference time-series analysis.
Explanation
Correct Answer
B:This corrects the relevant failure boundary instead of merely hiding the symptom. Add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger. The issue matters because on-call teams are paged repeatedly for harmless short-lived noise.
Incorrect Answers
A:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A team has accurate inference logs but its monitoring job never refreshes computed drift metrics after a table schema change. The relevant corrective action here is add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.
C:That corrective action 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 relevant corrective action here is add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.
D:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A research team captures only daily batch inference totals with no row-level timestamps but insists on per-minute live inference drift analysis. The relevant corrective action here is add an appropriately sustained alert condition or window aggregate instead of a one-minute peak trigger.
Question 321
Review this production model incident: A low-latency feature table uses external customer names as lookup keys; names change and are not unique. What evidence would best distinguish the actual fault from neighboring ML concerns?
- Test zero, null, negative and normal synthetic input combinations against the declared feature schema.
- Inspect online table key cardinality and mismatches before and after a synthetic rename.
- Check entity/timestamp duplicate counts around job retries and writer transaction history.
- Inspect checkpoint paths and duplicate feature writes after a simulated restart.
Explanation
Correct Answer
B:This is the most discriminating check for the observed symptoms. Inspect online table key cardinality and mismatches before and after a synthetic rename. The underlying technical issue is that online feature lookup keys must be stable and uniquely identify the intended entity.
Incorrect Answers
A:That diagnostic evidence 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. For the stated incident, the decisive observation is inspect online table key cardinality and mismatches before and after a synthetic rename.
C:That diagnostic evidence is directed at a different advanced feature store concepts situation: A feature refresh writes duplicate records for each entity and logical timestamp after retries, breaking unique lookups. For the stated incident, the decisive observation is inspect online table key cardinality and mismatches before and after a synthetic rename.
D:That diagnostic evidence is directed at a different advanced feature store concepts situation: An online behavioral feature counter consumes a streaming source but does not retain checkpoint state after worker restarts. For the stated incident, the decisive observation is inspect online table key cardinality and mismatches before and after a synthetic rename.
Question 322
A production ML review contains this situation: A retail profile is created on a static materialized daily inventory table; a team expects its initial snapshot to show each prior hour as a separate trend. What is the most material immediate risk in this situation?
- The monitoring pipeline either floods the SOC or ignores new material shifts.
- Missing historical windows are treated as evidence of zero incidents or zero data change.
- Selection bias looks like model input deterioration even though each subgroup is stable.
- Users cannot receive predictions because of infrastructure failures that retraining cannot fix.
Explanation
Correct Answer
B:This consequence follows from the specific unresolved condition. Missing historical windows are treated as evidence of zero incidents or zero data change. The causal constraint is: Snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
Incorrect Answers
A:That direct consequence 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. The more immediate risk here is missing historical windows are treated as evidence of zero incidents or zero data change.
C:That direct consequence 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. The more immediate risk here is missing historical windows are treated as evidence of zero incidents or zero data change.
D:That direct consequence 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 more immediate risk here is missing historical windows are treated as evidence of zero incidents or zero data change.
Question 323
Consider the following fictional Databricks production case: A production deployment scales to 30 teams but all jobs share one compute pool with no capacity limits or cost attribution. Which result is the strongest acceptance criterion after remediation?
- The release pipeline validates all references and executes an authorized staging test job.
- A dry-run plan for dev resolves only development assets; prod changes require separate approval.
- A clean test workspace deployment from the same pinned bundle matches the approved configuration.
- Load tests show production SLAs hold while research workloads respect assigned capacity limits.
Explanation
Correct Answer
D:This verification measures the required result rather than just process liveness. Load tests show production SLAs hold while research workloads respect assigned capacity limits. It confirms whether the targeted action succeeded: apply approved workload isolation, compute policies, quotas and cost attribution for training versus serving.
Incorrect Answers
A:That acceptance test is directed at a different environment architectures situation: A continuous-delivery job deploys a notebook change through a bundle but skips validation of referenced experiment paths and job permissions. Here the actual success criterion is load tests show production SLAs hold while research workloads respect assigned capacity limits.
B:That acceptance test 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. Here the actual success criterion is load tests show production SLAs hold while research workloads respect assigned capacity limits.
C:That acceptance test 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. Here the actual success criterion is load tests show production SLAs hold while research workloads respect assigned capacity limits.
Question 324
Consider the following fictional Databricks production case: A retailer trains a demand classifier on 180 million Delta rows already partitioned across a Spark cluster, but an analyst attempts to convert the entire training set to a local pandas object. Which result is the strongest acceptance criterion after remediation?
- Two sample sentences produce the expected token-array column from a direct transform.
- Every validation segment follows its training cutoff with no post-cutoff feature inputs.
- A representative training job completes through executor tasks without materializing all training rows on the driver.
- Holdout-only categories no longer influence the training-fitted indexer and metrics are recomputed on untouched holdout records.
Explanation
Correct Answer
C:This verification measures the required result rather than just process liveness. A representative training job completes through executor tasks without materializing all training rows on the driver. It confirms whether the targeted action succeeded: keep the eligible features as distributed Spark DataFrames and fit a suitable SparkML pipeline on the cluster.
Incorrect Answers
A:That acceptance test 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. Here the actual success criterion is a representative training job completes through executor tasks without materializing all training rows on the driver.
B:That acceptance test 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. Here the actual success criterion is a representative training job completes through executor tasks without materializing all training rows on the driver.
D:That acceptance test 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. Here the actual success criterion is a representative training job completes through executor tasks without materializing all training rows on the driver.
Question 325
Consider the following fictional Databricks production case: An account-risk model's fraud recall is stable overall but unexpectedly falls for new customers in their first 30 days. Which implementation is most appropriate for the stated incident?
- Aggregate or minimize the custom metric and apply authorized monitoring-table permissions.
- Add onboarding-age slices with controlled thresholds to the inference monitor.
- Tune the proven bottleneck and concurrency/capacity settings instead of scaling CPU blindly.
- Include governed dataset/model identifiers, metric values and comparison periods in the alert payload.
Explanation
Correct Answer
B:This corrects the relevant failure boundary instead of merely hiding the symptom. Add onboarding-age slices with controlled thresholds to the inference monitor. The issue matters because early-account fraud receives inadequate detection despite acceptable combined recall.
Incorrect Answers
A:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A monitor owner writes a custom metric that logs unaggregated account-level health data in a broadly shared results table. The relevant corrective action here is add onboarding-age slices with controlled thresholds to the inference monitor.
C:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A production endpoint scales CPU instances up when traffic rises, but request queues and P95 response time keep growing despite low CPU usage. The relevant corrective action here is add onboarding-age slices with controlled thresholds to the inference monitor.
D:That corrective action is directed at a different drift detection and lakehouse monitoring situation: A drift alert fires but its message contains only the word anomaly without dataset, model version or baseline window. The relevant corrective action here is add onboarding-age slices with controlled thresholds to the inference monitor.
Question 326
In an enterprise ML workflow, the reported issue is: A retail profile is created on a static materialized daily inventory table; a team expects its initial snapshot to show each prior hour as a separate trend. What technical distinction is essential to this investigation?
- Snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
- Distributional drift cannot be ruled out by one stable mean; numeric quantiles and supported statistical comparisons can expose shape changes.
- A snapshot profile suits a general current table when no prediction or event-time model-monitoring contract exists.
- Alert windows and persistence conditions must align with the service objective to avoid transient false positives.
Explanation
Correct Answer
A:The technical constraint is specific to the reported failure. Snapshot profiles describe current table state rather than reconstruct missing hourly time-series history. To test it, inspect the stored snapshot rows and timestamps to determine whether historical hourly observations are retained.
Incorrect Answers
B:That underlying principle is directed at a different drift detection and lakehouse monitoring situation: 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. The current incident instead turns on snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
C:That underlying principle 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. The current incident instead turns on snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
D:That underlying principle 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 current incident instead turns on snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
Question 327
Consider the following fictional Databricks production case: A risk service measures feature drift but neglects observed-label performance and continues serving an accurate-looking yet ineffective model. What underlying principle determines the most appropriate interpretation?
- Input distribution and predictive outcome quality are complementary monitoring signals with different evidence needs.
- Snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
- Statistical significance and practical effect size differ; large samples can make negligible shifts detectable.
- Reference populations should match relevant seasonality and business context when interpreting drift.
Explanation
Correct Answer
A:The technical constraint is specific to the reported failure. Input distribution and predictive outcome quality are complementary monitoring signals with different evidence needs. To test it, compare feature drift trends with label-matured error rates and calibration by model version.
Incorrect Answers
B:That underlying principle is directed at a different drift detection and lakehouse monitoring situation: A retail profile is created on a static materialized daily inventory table; a team expects its initial snapshot to show each prior hour as a separate trend. The current incident instead turns on input distribution and predictive outcome quality are complementary monitoring signals with different evidence needs.
C:That underlying principle is directed at a different drift detection and lakehouse monitoring situation: A monitoring dashboard reports a statistically significant tiny difference after one million observations, and the team immediately treats it as operationally material drift. The current incident instead turns on input distribution and predictive outcome quality are complementary monitoring signals with different evidence needs.
D:That underlying principle 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 current incident instead turns on input distribution and predictive outcome quality are complementary monitoring signals with different evidence needs.
Question 328
Review this production model incident: A SparkML grid-search experiment selects the model with the largest error metric even though the evaluator reports RMSE to minimize. What end-to-end verification is most persuasive here?
- Holdout-only categories no longer influence the training-fitted indexer and metrics are recomputed on untouched holdout records.
- Model accuracy and coverage are reported on untouched realistic evaluation cohorts.
- The selected candidate has the minimum validated RMSE among comparable parameter configurations.
- Two sample sentences produce the expected token-array column from a direct transform.
Explanation
Correct Answer
C:This verification measures the required result rather than just process liveness. The selected candidate has the minimum validated RMSE among comparable parameter configurations. It confirms whether the targeted action succeeded: configure selection against the intended minimization metric and rerun model comparison.
Incorrect Answers
A:That acceptance test 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. Here the actual success criterion is the selected candidate has the minimum validated RMSE among comparable parameter configurations.
B:That acceptance test is directed at a different using spark ml situation: A recommender experiment computes all evaluation metrics on training interactions and reports the result as generalization to new customers. Here the actual success criterion is the selected candidate has the minimum validated RMSE among comparable parameter configurations.
D:That acceptance test 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. Here the actual success criterion is the selected candidate has the minimum validated RMSE among comparable parameter configurations.
Question 329
Consider the following fictional Databricks production case: A fleet sensor table records measurement timestamps and analysts need daily drift trends; a snapshot-only profile erases the time dimension. What underlying principle determines the most appropriate interpretation?
- Observability starts with traceable, appropriately governed prediction logging before statistical analysis.
- Snapshot profiles describe current table state rather than reconstruct missing hourly time-series history.
- An alert is operationally useful only with explicit routing, threshold, urgency and escalation ownership.
- Time-series data profiles use a timestamp and granularity to compare measurements across periods.
Explanation
Correct Answer
D:The technical constraint is specific to the reported failure. Time-series data profiles use a timestamp and granularity to compare measurements across periods. To test it, inspect timestamp type, event cadence and intended daily windows before choosing the monitor.
Incorrect Answers
A:That underlying principle 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. The current incident instead turns on time-series data profiles use a timestamp and granularity to compare measurements across periods.
B:That underlying principle is directed at a different drift detection and lakehouse monitoring situation: A retail profile is created on a static materialized daily inventory table; a team expects its initial snapshot to show each prior hour as a separate trend. The current incident instead turns on time-series data profiles use a timestamp and granularity to compare measurements across periods.
C:That underlying principle 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 current incident instead turns on time-series data profiles use a timestamp and granularity to compare measurements across periods.
Question 330
The engineering team encounters the following issue: A retailer trains a demand classifier on 180 million Delta rows already partitioned across a Spark cluster, but an analyst attempts to convert the entire training set to a local pandas object. What investigation would most effectively isolate this behavior?
- Compare absolute and squared residual distributions and inspect the high-error tail on holdout data.
- Compare training-set byte size with driver memory and inspect Spark task distribution versus the pandas conversion plan.
- Inspect its declared output column and confirm the downstream estimator consumes a vector rather than predicted labels.
- Inspect fitted category mappings, handleInvalid behavior and a sample newly introduced category.
Explanation
Correct Answer
B:This is the most discriminating check for the observed symptoms. Compare training-set byte size with driver memory and inspect Spark task distribution versus the pandas conversion plan. The underlying technical issue is that distributed feature transformations and a SparkML estimator avoid mandatory collection of the full training set on one driver.
Incorrect Answers
A:That diagnostic evidence is directed at a different using spark ml situation: A forecasting model has a few catastrophic numeric errors hidden by a low median absolute error and operations care about large misses. For the stated incident, the decisive observation is compare training-set byte size with driver memory and inspect Spark task distribution versus the pandas conversion plan.
C:That diagnostic evidence 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. For the stated incident, the decisive observation is compare training-set byte size with driver memory and inspect Spark task distribution versus the pandas conversion plan.
D:That diagnostic evidence 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. For the stated incident, the decisive observation is compare training-set byte size with driver memory and inspect Spark task distribution versus the pandas conversion plan.