Certified Data Engineer Associate Practice Test 25
Question 240
At an education technology firm, a large fact DataFrame is joined to a very small reference dimension but shuffle volume dominates runtime. The dataset `daily_receipts` is in scope, and raw records must remain available for replay. Before release, which acceptance test would actually demonstrate that the root problem is controlled?
- Compare join plans and row counts with and without broadcasting the small dimension
- Run only a normal, unchanged `daily_receipts` sample and review reference table size.
- Review a platform-level availability report for `daily_receipts` without testing whether shuffling both sides is unnecessary when the small reference fits within safe broadcast memory.
- Check that the intended configuration exists for `daily_receipts` and that a healthy run succeeded, without inducing the earlier failure.
Explanation
Correct Answer
A:Compare join plans and row counts with and without broadcasting the small dimension. This reproduces the scenario-dependent condition, verifies reference table size, Spark physical plan, shuffle bytes, and joined row cardinality, and directly tests whether shuffling both sides is unnecessary when the small reference fits within safe broadcast memory is resolved. Observed results, not intent, determine approval.
Incorrect Answers
B:Run only a normal, unchanged `daily_receipts` sample and review reference table size. A normal sample does not challenge shuffling both sides is unnecessary when the small reference fits within safe broadcast memory.
C:Review a platform-level availability report for `daily_receipts` without testing whether shuffling both sides is unnecessary when the small reference fits within safe broadcast memory. Aggregate availability does not establish the behavior of affected `daily_receipts` records. It cannot replace the direct exercise: compare join plans and row counts with and without broadcasting the small dimension.
D:Check that the intended configuration exists for `daily_receipts` and that a healthy run succeeded, without inducing the earlier failure. Installed configuration and a healthy run are insufficient to prove the failure mode is gone. The original issue was shuffling both sides is unnecessary when the small reference fits within safe broadcast memory.
Question 241
At a charitable aid network, a nested commerce DataFrame contains arrays of product line items that need one row per product for inventory analysis. The dataset `tower_incidents` is in scope, and a finance audit requires reproducible aggregated figures. What evidence should the engineer collect to establish the cause rather than a coincidental symptom?
- Compare the current configuration and the signed change approval, without source-row evidence
- Correlate array item counts, order IDs across affected and unaffected `R034-CASE-049` runs.
- Review completed run status and aggregate cluster health without record-level correlation
- Examine only a normal successful run of `tower_incidents` and trend array item counts without recreating the fault.
Explanation
Correct Answer
B:Correlate array item counts, order IDs across affected and unaffected `R034-CASE-049` runs.. By joining array item counts, order IDs, output row counts, and nested-field types across the affected and unaffected conditions, the investigation directly tests the recorded mechanism: selecting the array as one column leaves all child products inside the same row. It also sets a measurable baseline for test orders with one, zero and three items and validate item-level row counts and parent associations.
Incorrect Answers
A:Compare the current configuration and the signed change approval, without source-row evidence. A desired configuration and approval ticket are not runtime proof. In this case they omit array item counts, order IDs, output row counts, and nested-field types, the direct means of distinguishing selecting the array as one column leaves all child products inside the same row.
C:Review completed run status and aggregate cluster health without record-level correlation. A successful aggregate platform state can coexist with selecting the array as one column leaves all child products inside the same row. This check never observes the affected `tower_incidents` records or their causal path.
D:Examine only a normal successful run of `tower_incidents` and trend array item counts without recreating the fault. A healthy-run trend can be useful, but it does not expose selecting the array as one column leaves all child products inside the same row. The failing-path array item counts, order IDs, output row counts, and nested-field types has to be correlated before attributing cause.
Question 242
At a charitable aid network, a nested commerce DataFrame contains arrays of product line items that need one row per product for inventory analysis. The dataset `tower_incidents` is in scope, and a finance audit requires reproducible aggregated figures. Which platform configuration or data-engineering change is most defensible?
- Use explode on the product array, retaining the order identifier alongside each child product record
- Split raw JSON on commas to form separate products; use that approach for `tower_incidents`.
- Apply distinct to the array column and expect one item per row; use that approach for `tower_incidents`.
- Rename the array field to product_id without changing the values; use that approach for `tower_incidents`.
Explanation
Correct Answer
A:Use explode on the product array, retaining the order identifier alongside each child product record. This directly addresses selecting the array as one column leaves all child products inside the same row. The design should be accepted only after test orders with one, zero and three items and validate item-level row counts and parent associations; the related operational boundary is to expand nested entities only in the curated model and preserve original bronze event context.
Incorrect Answers
B:Split raw JSON on commas to form separate products; use that approach for `tower_incidents`. Delimiter parsing ignores nested objects and escaped text. For `tower_incidents`, this cannot remove the causal limitation: selecting the array as one column leaves all child products inside the same row.
C:Apply distinct to the array column and expect one item per row; use that approach for `tower_incidents`. Distinct deduplicates rows but does not flatten an array. For `tower_incidents`, this cannot remove the causal limitation: selecting the array as one column leaves all child products inside the same row.
D:Rename the array field to product_id without changing the values; use that approach for `tower_incidents`. Renaming changes only the label, not the nested cardinality. For `tower_incidents`, this cannot remove the causal limitation: selecting the array as one column leaves all child products inside the same row.
Question 243
At a charitable aid network, a nested commerce DataFrame contains arrays of product line items that need one row per product for inventory analysis. The dataset `tower_incidents` is in scope, and a finance audit requires reproducible aggregated figures. A teammate proposes: "Rename the array field to product_id without changing the values." Which technical limitation is the decisive reason not to approve that specific proposal?
- Renaming changes only the label, not the nested cardinality. The proposal does not repair `tower_incidents`.
- Delimiter parsing ignores nested objects and escaped text. This would apply to an alternative change for `tower_incidents`.
- The dashboard cached the prior result; no DataFrame operation or underlying data requires correction.
- Distinct deduplicates rows but does not flatten an array. This would apply to another option for `tower_incidents`.
Explanation
Correct Answer
A:Renaming changes only the label, not the nested cardinality. The proposal does not repair `tower_incidents`. This identifies the known limitation of the exact proposed action. Renaming changes only the label, not the nested cardinality. The underlying problem remains selecting the array as one column leaves all child products inside the same row; Use explode on the product array, retaining the order identifier alongside each child product record instead targets that failure.
Incorrect Answers
B:Delimiter parsing ignores nested objects and escaped text. This would apply to an alternative change for `tower_incidents`. This failure applies to the separate proposal "Split raw JSON on commas to form separate products".
C:The dashboard cached the prior result; no DataFrame operation or underlying data requires correction. The case provides a concrete failing data or control path; a mere alternate theory does not address selecting the array as one column leaves all child products inside the same row.
D:Distinct deduplicates rows but does not flatten an array. This would apply to another option for `tower_incidents`. This limitation belongs to the different alternative "Apply distinct to the array column and expect one item per row". The stem calls for the specific failure of "Rename the array field to product_id without changing the values".
Question 244
At a charitable aid network, a nested commerce DataFrame contains arrays of product line items that need one row per product for inventory analysis. The dataset `tower_incidents` is in scope, and a finance audit requires reproducible aggregated figures. What risk-and-cost compromise should the owner accept instead of the proposed shortcuts?
- Rename the array field to product_id without changing the values
- Apply distinct to the array column and expect one item per row
- Expand nested entities only in the curated model and preserve original bronze event context
- Split raw JSON on commas to form separate products; monitor results only after launch of `tower_incidents`.
Explanation
Correct Answer
C:Expand nested entities only in the curated model and preserve original bronze event context. This explicitly bounds the compromise while still addressing selecting the array as one column leaves all child products inside the same row. Acceptance relies on test orders with one, zero and three items and validate item-level row counts and parent associations rather than simply declaring the change complete.
Incorrect Answers
A:Rename the array field to product_id without changing the values. Renaming changes only the label, not the nested cardinality. Monitoring after deployment does not make this approach a safe bounded tradeoff for `tower_incidents`.
B:Apply distinct to the array column and expect one item per row. Distinct deduplicates rows but does not flatten an array. Monitoring after deployment does not make this approach a safe bounded tradeoff for `tower_incidents`.
D:Split raw JSON on commas to form separate products; monitor results only after launch of `tower_incidents`. Delimiter parsing ignores nested objects and escaped text. Monitoring after deployment does not make this approach a safe bounded tradeoff for `tower_incidents`.
Question 245
At a charitable aid network, a nested commerce DataFrame contains arrays of product line items that need one row per product for inventory analysis. The dataset `tower_incidents` is in scope, and a finance audit requires reproducible aggregated figures. Which direct test could falsify the proposed fix instead of only confirming deployment?
- Review a platform-level availability report for `tower_incidents` without testing whether selecting the array as one column leaves all child products inside the same row.
- Test orders with one, zero and three items and validate item-level row counts and parent associations
- Run only a normal, unchanged `tower_incidents` sample and review array item counts.
- Check that the intended configuration exists for `tower_incidents` and that a healthy run succeeded, without inducing the earlier failure.
Explanation
Correct Answer
B:Test orders with one, zero and three items and validate item-level row counts and parent associations. This reproduces the scenario-dependent condition, verifies array item counts, order IDs, output row counts, and nested-field types, and directly tests whether selecting the array as one column leaves all child products inside the same row is resolved. Observed results, not intent, determine approval.
Incorrect Answers
A:Review a platform-level availability report for `tower_incidents` without testing whether selecting the array as one column leaves all child products inside the same row. Aggregate availability does not establish the behavior of affected `tower_incidents` records. It cannot replace the direct exercise: test orders with one, zero and three items and validate item-level row counts and parent associations.
C:Run only a normal, unchanged `tower_incidents` sample and review array item counts. A normal sample does not challenge selecting the array as one column leaves all child products inside the same row.
D:Check that the intended configuration exists for `tower_incidents` and that a healthy run succeeded, without inducing the earlier failure. Installed configuration and a healthy run are insufficient to prove the failure mode is gone. The original issue was selecting the array as one column leaves all child products inside the same row.
Question 246
At a gaming analytics company, a DataFrame cleanup drops records when a numeric threshold is missing, but policy requires retaining those rows for review. The dataset `content_clicks` is in scope, and the release must preserve the original source evidence. Which correlated observation is the strongest basis for the change decision?
- Correlate null input count, filter predicate across affected and unaffected `R034-CASE-050` runs.
- Compare the current configuration and the signed change approval, without source-row evidence
- Inspect normal-period success metrics for the stated dataset without reproducing the source error
- Review completed run status and aggregate cluster health without record-level correlation
Explanation
Correct Answer
A:Correlate null input count, filter predicate across affected and unaffected `R034-CASE-050` runs.. By joining null input count, filter predicate, kept and rejected row counts, and downstream reporting totals across the affected and unaffected conditions, the investigation directly tests the recorded mechanism: ordinary numeric comparisons against null produce unknown predicates that filters do not retain. It also sets a measurable baseline for inject null, zero and positive amounts and assert each enters the intended output or review route.
Incorrect Answers
B:Compare the current configuration and the signed change approval, without source-row evidence. A desired configuration and approval ticket are not runtime proof. In this case they omit null input count, filter predicate, kept and rejected row counts, and downstream reporting totals, the direct means of distinguishing ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
C:Inspect normal-period success metrics for the stated dataset without reproducing the source error. A healthy-run trend can be useful, but it does not expose ordinary numeric comparisons against null produce unknown predicates that filters do not retain. The failing-path null input count, filter predicate, kept and rejected row counts, and downstream reporting totals has to be correlated before attributing cause.
D:Review completed run status and aggregate cluster health without record-level correlation. A successful aggregate platform state can coexist with ordinary numeric comparisons against null produce unknown predicates that filters do not retain. This check never observes the affected `content_clicks` records or their causal path.
Question 247
At a gaming analytics company, a DataFrame cleanup drops records when a numeric threshold is missing, but policy requires retaining those rows for review. The dataset `content_clicks` is in scope, and the release must preserve the original source evidence. What implementation should the engineer select for the failing workflow?
- Use a null-aware filter or explicit error category and keep the unmatched rows in a separate exception dataset
- Write filter("amount > 0") and assume nulls remain; use that approach for `content_clicks`.
- Replace every null amount with an invented high value; use that approach for `content_clicks`.
- Use dropna on every DataFrame column before saving; use that approach for `content_clicks`.
Explanation
Correct Answer
A:Use a null-aware filter or explicit error category and keep the unmatched rows in a separate exception dataset. This directly addresses ordinary numeric comparisons against null produce unknown predicates that filters do not retain. The design should be accepted only after inject null, zero and positive amounts and assert each enters the intended output or review route; the related operational boundary is to retain uncertain records for remediation without treating them as valid positive transactions.
Incorrect Answers
B:Write filter("amount > 0") and assume nulls remain; use that approach for `content_clicks`. A null comparison is not true and rows are removed by the filter. For `content_clicks`, this cannot remove the causal limitation: ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
C:Replace every null amount with an invented high value; use that approach for `content_clicks`. Fabricated positive amounts can incorrectly satisfy business rules. For `content_clicks`, this cannot remove the causal limitation: ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
D:Use dropna on every DataFrame column before saving; use that approach for `content_clicks`. Dropping all null-bearing rows increases data loss and ignores field semantics. For `content_clicks`, this cannot remove the causal limitation: ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
Question 248
At a gaming analytics company, a DataFrame cleanup drops records when a numeric threshold is missing, but policy requires retaining those rows for review. The dataset `content_clicks` is in scope, and the release must preserve the original source evidence. A teammate proposes: "Write filter("amount > 0") and assume nulls remain." Which technical limitation is the decisive reason not to approve that specific proposal?
- The dashboard cached the prior result; no DataFrame operation or underlying data requires correction.
- Dropping all null-bearing rows increases data loss and ignores field semantics. This would apply to another option for `content_clicks`.
- Fabricated positive amounts can incorrectly satisfy business rules. This would apply to an alternative change for `content_clicks`.
- A null comparison is not true and rows are removed by the filter. The proposal does not repair `content_clicks`.
Explanation
Correct Answer
D:A null comparison is not true and rows are removed by the filter. The proposal does not repair `content_clicks`. This identifies the known limitation of the exact proposed action. A null comparison is not true and rows are removed by the filter. The underlying problem remains ordinary numeric comparisons against null produce unknown predicates that filters do not retain; Use a null-aware filter or explicit error category and keep the unmatched rows in a separate exception dataset instead targets that failure.
Incorrect Answers
A:The dashboard cached the prior result; no DataFrame operation or underlying data requires correction. The case provides a concrete failing data or control path; a mere alternate theory does not address ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
B:Dropping all null-bearing rows increases data loss and ignores field semantics. This would apply to another option for `content_clicks`. This limitation belongs to the different alternative "Use dropna on every DataFrame column before saving". The stem calls for the specific failure of "Write filter("amount > 0") and assume nulls remain".
C:Fabricated positive amounts can incorrectly satisfy business rules. This would apply to an alternative change for `content_clicks`. This failure applies to the separate proposal "Replace every null amount with an invented high value".
Question 249
At a gaming analytics company, a DataFrame cleanup drops records when a numeric threshold is missing, but policy requires retaining those rows for review. The dataset `content_clicks` is in scope, and the release must preserve the original source evidence. Which decision preserves the stated data contract while keeping the change scope proportionate?
- Use dropna on every DataFrame column before saving
- Replace every null amount with an invented high value
- Write filter("amount > 0") and assume nulls remain
- Retain uncertain records for remediation without treating them as valid positive transactions
Explanation
Correct Answer
D:Retain uncertain records for remediation without treating them as valid positive transactions. This explicitly bounds the compromise while still addressing ordinary numeric comparisons against null produce unknown predicates that filters do not retain. Acceptance relies on inject null, zero and positive amounts and assert each enters the intended output or review route rather than simply declaring the change complete.
Incorrect Answers
A:Use dropna on every DataFrame column before saving. Dropping all null-bearing rows increases data loss and ignores field semantics. Monitoring after deployment does not make this approach a safe bounded tradeoff for `content_clicks`.
B:Replace every null amount with an invented high value. Fabricated positive amounts can incorrectly satisfy business rules. Monitoring after deployment does not make this approach a safe bounded tradeoff for `content_clicks`.
C:Write filter("amount > 0") and assume nulls remain. A null comparison is not true and rows are removed by the filter. Monitoring after deployment does not make this approach a safe bounded tradeoff for `content_clicks`.
Question 250
At a gaming analytics company, a DataFrame cleanup drops records when a numeric threshold is missing, but policy requires retaining those rows for review. The dataset `content_clicks` is in scope, and the release must preserve the original source evidence. What verification provides evidence of correct behavior under the original adverse condition?
- Review a platform-level availability report for `content_clicks` without testing whether ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
- Inject null, zero and positive amounts and assert each enters the intended output or review route
- Run only a normal, unchanged `content_clicks` sample and review null input count.
- Check that the intended configuration exists for `content_clicks` and that a healthy run succeeded, without inducing the earlier failure.
Explanation
Correct Answer
B:Inject null, zero and positive amounts and assert each enters the intended output or review route. This reproduces the scenario-dependent condition, verifies null input count, filter predicate, kept and rejected row counts, and downstream reporting totals, and directly tests whether ordinary numeric comparisons against null produce unknown predicates that filters do not retain is resolved. Observed results, not intent, determine approval.
Incorrect Answers
A:Review a platform-level availability report for `content_clicks` without testing whether ordinary numeric comparisons against null produce unknown predicates that filters do not retain. Aggregate availability does not establish the behavior of affected `content_clicks` records. It cannot replace the direct exercise: inject null, zero and positive amounts and assert each enters the intended output or review route.
C:Run only a normal, unchanged `content_clicks` sample and review null input count. A normal sample does not challenge ordinary numeric comparisons against null produce unknown predicates that filters do not retain.
D:Check that the intended configuration exists for `content_clicks` and that a healthy run succeeded, without inducing the earlier failure. Installed configuration and a healthy run are insufficient to prove the failure mode is gone. The original issue was ordinary numeric comparisons against null produce unknown predicates that filters do not retain.