Data Management Specific to AI AAIA Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A bank is developing a fraud detection model. The model is trained on transaction data from the past 5 years. The data science team has noticed that the model's false positive rate has increased significantly in the last month. The bank's data quality team has confirmed that the data pipeline is running correctly and no new features have been added. Which data quality dimension is most likely causing the increase in false positives?

   • ACompleteness — the training data has missing values for some transaction attributes.
   • BTimeliness — the model is being trained on data that is not current enough to reflect recent fraud patterns.
   • CAccuracy — the data contains a higher proportion of incorrect transaction amounts.
   • DConsistency — the data from different branches uses different currency codes.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   Which technique is commonly used to detect bias in training data?

   • AStatistical analysis of group distributions
   • BData encryption
   • CNetwork monitoring
   • DData backup
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   Which governance control is most critical during the data retention stage of the AI data lifecycle?

   • AObtaining consent from data subjects
   • BDefining and enforcing retention periods based on legal and business requirements
   • CValidating data accuracy through profiling
   • DImplementing encryption for data at rest
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which regulation specifically addresses the protection of personal data of individuals in the European Union?

   • ACCPA
   • BGDPR
   • CPCI DSS
   • DHIPAA
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   An AI auditor is investigating a model that produces different outputs for the same input over time, even though the model code has not changed. The auditor suspects the issue is related to the data pipeline. The organization has a data governance framework but no automated lineage tool. Which investigation approach is most likely to identify the root cause?

   • ATrace the data lineage manually by reviewing the data pipeline code and the logs of the data ingestion and transformation steps.
   • BCheck if the model's hyperparameters have been changed by the data science team.
   • CCompare the model's output distribution to the training data distribution to see if they match.
   • DReview the model's performance metrics over time to see if they have degraded.
   Reveal Solution Discussion Select an answer first