Objective 2.5 Address Business Risks, Ethical Concerns, and Related Concepts in Data Exploration/feature Engineering AIP-210 Practice Questions (Page 2)

Questions 6–10

1. 6expert · hard

   A university is building an AI system to predict student dropout risk. The team engineers a feature called 'parental education level' and finds it significantly improves model accuracy. However, an ethics review board is concerned that this feature may disadvantage students from first-generation college families. The team is considering several options. Which approach best balances the ethical concern with the model's predictive value?

   • AUse the feature only for students who opt in to provide this information.
   • BRemove the feature entirely and accept the decrease in model accuracy.
   • CKeep the feature because it is a strong predictor and the model's overall accuracy is high.
   • DReplace the 'parental education level' feature with a composite feature that combines it with other socioeconomic indicators and monitor the model's fairness metrics.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   What is the primary purpose of encrypting data during the feature engineering process?

   • ATo improve the interpretability of the model
   • BTo handle missing values in the dataset
   • CTo ensure data is protected from unauthorized access
   • DTo reduce the dimensionality of the feature space
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   To mitigate bias in feature engineering, a data scientist should:

   • AIncrease the number of features to capture more patterns
   • BRemove all categorical features to avoid bias
   • CEnsure the data used for feature engineering is representative of the target population
   • DUse only features that are directly available in the raw data
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A data science team is evaluating a model that predicts whether a customer will default on a loan. They compute the model's overall accuracy and find it is 95%. However, they are concerned about fairness. Which additional metric would be most useful to assess potential bias?

   • ACalculate the false positive rate separately for different demographic groups.
   • BCalculate the model's training time.
   • CCalculate the total number of predictions made by the model.
   • DCalculate the average confidence score of the model's predictions.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A telecommunications company is exploring a dataset of customer call records to engineer features for a network-quality model. The dataset includes precise GPS locations of cell towers. The legal team warns that this data could be used to identify individual customers' movements. What is the most appropriate action during data exploration?

   • ARemove all location data from the dataset to eliminate any privacy risk.
   • BGeneralize the GPS coordinates to a larger geographic region, such as city or district level.
   • CDocument the GPS coordinates in the data lineage report and proceed without changes.
   • DEncrypt the GPS coordinates and store the key in the same data lake.
   Reveal Solution Discussion Select an answer first