BS Data Science · Machine Learning

Unit 6: Feature Engineering MCQs

50 solved multiple choice questions with answers and short explanations.

Q1. Which statement best describes Feature Engineering Concepts within Feature Engineering? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIts meaning can be determined without considering the data or question.
  2. BIt should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
  3. CIt is only a file-naming convention with no analytical effect.
  4. DIt guarantees correct conclusions regardless of data quality.
Show answer
Answer: B. It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.Machine Learning → Feature Engineering → Feature Engineering Concepts: It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Report a mistake

Q2. What is the primary purpose of Feature Engineering Concepts in a data science workflow? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIts purpose is always to maximize dataset size.
  2. BIt eliminates the need to define a business or research question.
  3. CIt guarantees causal conclusions from any dataset.
  4. DIt supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Show answer
Answer: D. It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.Machine Learning → Feature Engineering → Feature Engineering Concepts: It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Report a mistake

Q3. A data science team is working with Feature Engineering Concepts. Which approach is most appropriate? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore assumptions because software libraries automatically ensure validity.
  2. BDefine the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
  3. CChoose the most complex model before examining the data.
  4. DUse all data for fitting and skip independent validation.
Show answer
Answer: B. Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.Machine Learning → Feature Engineering → Feature Engineering Concepts: Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Report a mistake

Q4. Which data-related consideration is most important when applying Feature Engineering Concepts? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AUse any available data even when it leaks the target.
  2. BAssume large datasets are automatically representative.
  3. CIgnore missingness and provenance when a model produces high accuracy.
  4. DCheck data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Show answer
Answer: D. Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.Machine Learning → Feature Engineering → Feature Engineering Concepts: Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Report a mistake

Q5. Which principle best supports sound methodological use of Feature Engineering Concepts? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore statistical or algorithmic assumptions whenever computation succeeds.
  2. BChoose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
  3. CChoose a method solely because it is popular.
  4. DTreat training performance as proof of validity.
Show answer
Answer: B. Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.Machine Learning → Feature Engineering → Feature Engineering Concepts: Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Report a mistake

Q6. How should work involving Feature Engineering Concepts be evaluated? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AEvaluate only on data used to fit the model.
  2. BChoose a metric unrelated to the decision or task.
  3. CTreat an attractive visualization as sufficient validation.
  4. DUse metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Show answer
Answer: D. Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.Machine Learning → Feature Engineering → Feature Engineering Concepts: Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Report a mistake

Q7. Which risk should be considered carefully when using Feature Engineering Concepts? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AConfounding and leakage do not affect predictive conclusions.
  2. BConsider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
  3. CBias cannot occur when analysis uses mathematics.
  4. DPrivacy is irrelevant after data enters a database.
Show answer
Answer: B. Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.Machine Learning → Feature Engineering → Feature Engineering Concepts: Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Report a mistake

Q8. What is the best next step when results involving Feature Engineering Concepts are unsatisfactory? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChange many pipeline components simultaneously without diagnosis.
  2. BRemove difficult observations from evaluation.
  3. CIncrease complexity before checking data quality or error patterns.
  4. DDiagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Show answer
Answer: D. Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.Machine Learning → Feature Engineering → Feature Engineering Concepts: Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Report a mistake

Q9. Which trade-off should be assessed when choosing an approach to Feature Engineering Concepts? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AInterpretability is always irrelevant.
  2. BBalance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
  3. COnly predictive accuracy matters in every project.
  4. DCompute and operational cost never affect data science choices.
Show answer
Answer: B. Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.Machine Learning → Feature Engineering → Feature Engineering Concepts: Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Report a mistake

Q10. Which practice best represents responsible use of Feature Engineering Concepts? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. APublish sensitive raw data whenever it improves reproducibility.
  2. BHide limitations to make findings appear stronger.
  3. CReport point estimates without uncertainty when uncertainty is material.
  4. DDocument provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Show answer
Answer: D. Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.Machine Learning → Feature Engineering → Feature Engineering Concepts: Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Report a mistake

Q11. Which statement best describes Feature Engineering Methods within Feature Engineering? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIt is only a file-naming convention with no analytical effect.
  2. BIt guarantees correct conclusions regardless of data quality.
  3. CIts meaning can be determined without considering the data or question.
  4. DIt should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Show answer
Answer: D. It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.Machine Learning → Feature Engineering → Feature Engineering Methods: It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Report a mistake

Q12. What is the primary purpose of Feature Engineering Methods in a data science workflow? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIt guarantees causal conclusions from any dataset.
  2. BIt supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
  3. CIts purpose is always to maximize dataset size.
  4. DIt eliminates the need to define a business or research question.
Show answer
Answer: B. It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.Machine Learning → Feature Engineering → Feature Engineering Methods: It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Report a mistake

Q13. A data science team is working with Feature Engineering Methods. Which approach is most appropriate? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChoose the most complex model before examining the data.
  2. BUse all data for fitting and skip independent validation.
  3. CIgnore assumptions because software libraries automatically ensure validity.
  4. DDefine the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Show answer
Answer: D. Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.Machine Learning → Feature Engineering → Feature Engineering Methods: Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Report a mistake

Q14. Which data-related consideration is most important when applying Feature Engineering Methods? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore missingness and provenance when a model produces high accuracy.
  2. BCheck data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
  3. CUse any available data even when it leaks the target.
  4. DAssume large datasets are automatically representative.
Show answer
Answer: B. Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.Machine Learning → Feature Engineering → Feature Engineering Methods: Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Report a mistake

Q15. Which principle best supports sound methodological use of Feature Engineering Methods? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChoose a method solely because it is popular.
  2. BTreat training performance as proof of validity.
  3. CIgnore statistical or algorithmic assumptions whenever computation succeeds.
  4. DChoose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Show answer
Answer: D. Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.Machine Learning → Feature Engineering → Feature Engineering Methods: Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Report a mistake

Q16. How should work involving Feature Engineering Methods be evaluated? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. ATreat an attractive visualization as sufficient validation.
  2. BUse metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
  3. CEvaluate only on data used to fit the model.
  4. DChoose a metric unrelated to the decision or task.
Show answer
Answer: B. Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.Machine Learning → Feature Engineering → Feature Engineering Methods: Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Report a mistake

Q17. Which risk should be considered carefully when using Feature Engineering Methods? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. ABias cannot occur when analysis uses mathematics.
  2. BPrivacy is irrelevant after data enters a database.
  3. CConfounding and leakage do not affect predictive conclusions.
  4. DConsider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Show answer
Answer: D. Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.Machine Learning → Feature Engineering → Feature Engineering Methods: Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Report a mistake

Q18. What is the best next step when results involving Feature Engineering Methods are unsatisfactory? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIncrease complexity before checking data quality or error patterns.
  2. BDiagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
  3. CChange many pipeline components simultaneously without diagnosis.
  4. DRemove difficult observations from evaluation.
Show answer
Answer: B. Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.Machine Learning → Feature Engineering → Feature Engineering Methods: Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Report a mistake

Q19. Which trade-off should be assessed when choosing an approach to Feature Engineering Methods? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AOnly predictive accuracy matters in every project.
  2. BCompute and operational cost never affect data science choices.
  3. CInterpretability is always irrelevant.
  4. DBalance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Show answer
Answer: D. Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.Machine Learning → Feature Engineering → Feature Engineering Methods: Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Report a mistake

Q20. Which practice best represents responsible use of Feature Engineering Methods? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AReport point estimates without uncertainty when uncertainty is material.
  2. BDocument provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
  3. CPublish sensitive raw data whenever it improves reproducibility.
  4. DHide limitations to make findings appear stronger.
Show answer
Answer: B. Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.Machine Learning → Feature Engineering → Feature Engineering Methods: Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Report a mistake

Q21. Which statement best describes Feature Engineering Analysis within Feature Engineering? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIts meaning can be determined without considering the data or question.
  2. BIt should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
  3. CIt is only a file-naming convention with no analytical effect.
  4. DIt guarantees correct conclusions regardless of data quality.
Show answer
Answer: B. It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.Machine Learning → Feature Engineering → Feature Engineering Analysis: It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Report a mistake

Q22. What is the primary purpose of Feature Engineering Analysis in a data science workflow? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIts purpose is always to maximize dataset size.
  2. BIt eliminates the need to define a business or research question.
  3. CIt guarantees causal conclusions from any dataset.
  4. DIt supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Show answer
Answer: D. It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.Machine Learning → Feature Engineering → Feature Engineering Analysis: It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Report a mistake

Q23. A data science team is working with Feature Engineering Analysis. Which approach is most appropriate? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore assumptions because software libraries automatically ensure validity.
  2. BDefine the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
  3. CChoose the most complex model before examining the data.
  4. DUse all data for fitting and skip independent validation.
Show answer
Answer: B. Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.Machine Learning → Feature Engineering → Feature Engineering Analysis: Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Report a mistake

Q24. Which data-related consideration is most important when applying Feature Engineering Analysis? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AUse any available data even when it leaks the target.
  2. BAssume large datasets are automatically representative.
  3. CIgnore missingness and provenance when a model produces high accuracy.
  4. DCheck data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Show answer
Answer: D. Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.Machine Learning → Feature Engineering → Feature Engineering Analysis: Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Report a mistake

Q25. Which principle best supports sound methodological use of Feature Engineering Analysis? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore statistical or algorithmic assumptions whenever computation succeeds.
  2. BChoose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
  3. CChoose a method solely because it is popular.
  4. DTreat training performance as proof of validity.
Show answer
Answer: B. Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.Machine Learning → Feature Engineering → Feature Engineering Analysis: Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Report a mistake

Q26. How should work involving Feature Engineering Analysis be evaluated? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AEvaluate only on data used to fit the model.
  2. BChoose a metric unrelated to the decision or task.
  3. CTreat an attractive visualization as sufficient validation.
  4. DUse metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Show answer
Answer: D. Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.Machine Learning → Feature Engineering → Feature Engineering Analysis: Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Report a mistake

Q27. Which risk should be considered carefully when using Feature Engineering Analysis? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AConfounding and leakage do not affect predictive conclusions.
  2. BConsider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
  3. CBias cannot occur when analysis uses mathematics.
  4. DPrivacy is irrelevant after data enters a database.
Show answer
Answer: B. Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.Machine Learning → Feature Engineering → Feature Engineering Analysis: Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Report a mistake

Q28. What is the best next step when results involving Feature Engineering Analysis are unsatisfactory? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChange many pipeline components simultaneously without diagnosis.
  2. BRemove difficult observations from evaluation.
  3. CIncrease complexity before checking data quality or error patterns.
  4. DDiagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Show answer
Answer: D. Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.Machine Learning → Feature Engineering → Feature Engineering Analysis: Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Report a mistake

Q29. Which trade-off should be assessed when choosing an approach to Feature Engineering Analysis? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AInterpretability is always irrelevant.
  2. BBalance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
  3. COnly predictive accuracy matters in every project.
  4. DCompute and operational cost never affect data science choices.
Show answer
Answer: B. Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.Machine Learning → Feature Engineering → Feature Engineering Analysis: Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Report a mistake

Q30. Which practice best represents responsible use of Feature Engineering Analysis? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. APublish sensitive raw data whenever it improves reproducibility.
  2. BHide limitations to make findings appear stronger.
  3. CReport point estimates without uncertainty when uncertainty is material.
  4. DDocument provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Show answer
Answer: D. Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.Machine Learning → Feature Engineering → Feature Engineering Analysis: Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Report a mistake

Q31. Which statement best describes Feature Engineering Evaluation within Feature Engineering? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIt is only a file-naming convention with no analytical effect.
  2. BIt guarantees correct conclusions regardless of data quality.
  3. CIts meaning can be determined without considering the data or question.
  4. DIt should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Show answer
Answer: D. It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.Machine Learning → Feature Engineering → Feature Engineering Evaluation: It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Report a mistake

Q32. What is the primary purpose of Feature Engineering Evaluation in a data science workflow? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIt guarantees causal conclusions from any dataset.
  2. BIt supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
  3. CIts purpose is always to maximize dataset size.
  4. DIt eliminates the need to define a business or research question.
Show answer
Answer: B. It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.Machine Learning → Feature Engineering → Feature Engineering Evaluation: It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Report a mistake

Q33. A data science team is working with Feature Engineering Evaluation. Which approach is most appropriate? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChoose the most complex model before examining the data.
  2. BUse all data for fitting and skip independent validation.
  3. CIgnore assumptions because software libraries automatically ensure validity.
  4. DDefine the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Show answer
Answer: D. Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Report a mistake

Q34. Which data-related consideration is most important when applying Feature Engineering Evaluation? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore missingness and provenance when a model produces high accuracy.
  2. BCheck data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
  3. CUse any available data even when it leaks the target.
  4. DAssume large datasets are automatically representative.
Show answer
Answer: B. Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Report a mistake

Q35. Which principle best supports sound methodological use of Feature Engineering Evaluation? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChoose a method solely because it is popular.
  2. BTreat training performance as proof of validity.
  3. CIgnore statistical or algorithmic assumptions whenever computation succeeds.
  4. DChoose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Show answer
Answer: D. Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Report a mistake

Q36. How should work involving Feature Engineering Evaluation be evaluated? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. ATreat an attractive visualization as sufficient validation.
  2. BUse metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
  3. CEvaluate only on data used to fit the model.
  4. DChoose a metric unrelated to the decision or task.
Show answer
Answer: B. Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Report a mistake

Q37. Which risk should be considered carefully when using Feature Engineering Evaluation? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. ABias cannot occur when analysis uses mathematics.
  2. BPrivacy is irrelevant after data enters a database.
  3. CConfounding and leakage do not affect predictive conclusions.
  4. DConsider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Show answer
Answer: D. Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Report a mistake

Q38. What is the best next step when results involving Feature Engineering Evaluation are unsatisfactory? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIncrease complexity before checking data quality or error patterns.
  2. BDiagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
  3. CChange many pipeline components simultaneously without diagnosis.
  4. DRemove difficult observations from evaluation.
Show answer
Answer: B. Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Report a mistake

Q39. Which trade-off should be assessed when choosing an approach to Feature Engineering Evaluation? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AOnly predictive accuracy matters in every project.
  2. BCompute and operational cost never affect data science choices.
  3. CInterpretability is always irrelevant.
  4. DBalance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Show answer
Answer: D. Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Report a mistake

Q40. Which practice best represents responsible use of Feature Engineering Evaluation? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AReport point estimates without uncertainty when uncertainty is material.
  2. BDocument provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
  3. CPublish sensitive raw data whenever it improves reproducibility.
  4. DHide limitations to make findings appear stronger.
Show answer
Answer: B. Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.Machine Learning → Feature Engineering → Feature Engineering Evaluation: Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Report a mistake

Q41. Which statement best describes Feature Engineering Applications within Feature Engineering? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIts meaning can be determined without considering the data or question.
  2. BIt should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
  3. CIt is only a file-naming convention with no analytical effect.
  4. DIt guarantees correct conclusions regardless of data quality.
Show answer
Answer: B. It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.Machine Learning → Feature Engineering → Feature Engineering Applications: It should be understood through its assumptions, data requirements, method, outputs, and role in the analytical workflow.
Report a mistake

Q42. What is the primary purpose of Feature Engineering Applications in a data science workflow? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIts purpose is always to maximize dataset size.
  2. BIt eliminates the need to define a business or research question.
  3. CIt guarantees causal conclusions from any dataset.
  4. DIt supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Show answer
Answer: D. It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.Machine Learning → Feature Engineering → Feature Engineering Applications: It supports a defined analytical objective by turning appropriate data into evidence, estimates, predictions, or decisions.
Report a mistake

Q43. A data science team is working with Feature Engineering Applications. Which approach is most appropriate? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore assumptions because software libraries automatically ensure validity.
  2. BDefine the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
  3. CChoose the most complex model before examining the data.
  4. DUse all data for fitting and skip independent validation.
Show answer
Answer: B. Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.Machine Learning → Feature Engineering → Feature Engineering Applications: Define the question and success metric, inspect the data and assumptions, apply an appropriate method, and validate the result.
Report a mistake

Q44. Which data-related consideration is most important when applying Feature Engineering Applications? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AUse any available data even when it leaks the target.
  2. BAssume large datasets are automatically representative.
  3. CIgnore missingness and provenance when a model produces high accuracy.
  4. DCheck data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Show answer
Answer: D. Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.Machine Learning → Feature Engineering → Feature Engineering Applications: Check data quality, representativeness, leakage, missingness, measurement, provenance, and relevance to the target question.
Report a mistake

Q45. Which principle best supports sound methodological use of Feature Engineering Applications? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AIgnore statistical or algorithmic assumptions whenever computation succeeds.
  2. BChoose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
  3. CChoose a method solely because it is popular.
  4. DTreat training performance as proof of validity.
Show answer
Answer: B. Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.Machine Learning → Feature Engineering → Feature Engineering Applications: Choose a method whose assumptions fit the data and objective, then validate it with appropriate evidence rather than convenience.
Report a mistake

Q46. How should work involving Feature Engineering Applications be evaluated? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AEvaluate only on data used to fit the model.
  2. BChoose a metric unrelated to the decision or task.
  3. CTreat an attractive visualization as sufficient validation.
  4. DUse metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Show answer
Answer: D. Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.Machine Learning → Feature Engineering → Feature Engineering Applications: Use metrics, diagnostics, uncertainty, and held-out or otherwise valid evidence that reflect the actual analytical objective.
Report a mistake

Q47. Which risk should be considered carefully when using Feature Engineering Applications? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AConfounding and leakage do not affect predictive conclusions.
  2. BConsider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
  3. CBias cannot occur when analysis uses mathematics.
  4. DPrivacy is irrelevant after data enters a database.
Show answer
Answer: B. Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.Machine Learning → Feature Engineering → Feature Engineering Applications: Consider bias, leakage, confounding, privacy, poor generalization, and misleading interpretation relevant to the analysis.
Report a mistake

Q48. What is the best next step when results involving Feature Engineering Applications are unsatisfactory? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AChange many pipeline components simultaneously without diagnosis.
  2. BRemove difficult observations from evaluation.
  3. CIncrease complexity before checking data quality or error patterns.
  4. DDiagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Show answer
Answer: D. Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.Machine Learning → Feature Engineering → Feature Engineering Applications: Diagnose errors and data limitations first, then change preprocessing, features, method, or model based on evidence.
Report a mistake

Q49. Which trade-off should be assessed when choosing an approach to Feature Engineering Applications? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. AInterpretability is always irrelevant.
  2. BBalance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
  3. COnly predictive accuracy matters in every project.
  4. DCompute and operational cost never affect data science choices.
Show answer
Answer: B. Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.Machine Learning → Feature Engineering → Feature Engineering Applications: Balance accuracy or analytical value with interpretability, compute, latency, data needs, robustness, maintainability, and cost.
Report a mistake

Q50. Which practice best represents responsible use of Feature Engineering Applications? Context: BS Data Science / Machine Learning / Feature Engineering.

  1. APublish sensitive raw data whenever it improves reproducibility.
  2. BHide limitations to make findings appear stronger.
  3. CReport point estimates without uncertainty when uncertainty is material.
  4. DDocument provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Show answer
Answer: D. Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.Machine Learning → Feature Engineering → Feature Engineering Applications: Document provenance and limitations, protect sensitive data, test for bias or failure modes, and communicate uncertainty.
Report a mistake