Objective 4.4 Address Business Risks, Ethical Concerns, and Related Concepts in Operationalizing the Model AIP-210 Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A manufacturing company uses an ML model to predict equipment failures. The model occasionally misses failures, leading to costly downtime. The company wants to reduce the financial impact of false negatives. What mitigation strategy should the company implement?

   • ALower the decision threshold so the model flags more equipment for maintenance.
   • BIncrease the decision threshold so the model only flags equipment with very high failure probability.
   • CRetrain the model with more historical failure data to improve its accuracy.
   • DImplement a cost-sensitive learning approach that penalizes false negatives more heavily during training.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A fintech startup is launching a credit-scoring model. The model uses alternative data such as social media activity. The startup's legal advisor warns that using certain data types may violate fair lending laws. The startup wants to proceed with the model but minimize legal risk. What is the most appropriate action?

   • AConduct a legal review of the data sources and model features to ensure compliance with fair lending regulations.
   • BUse the model only for pre-qualification, not final credit decisions.
   • CRemove all social media data from the model to avoid any legal issues.
   • DProceed with the model but add a disclaimer that the model is not the sole basis for credit decisions.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   An e-commerce company deploys a recommendation model that influences product placements. After three months, the marketing team notices that the model's recommendations have become less diverse, showing the same few products to most users. The data science team suspects the model is exploiting a feedback loop. What should the team implement to detect and address this issue?

   • ASet up a dashboard that tracks recommendation diversity and user engagement metrics over time.
   • BRetrain the model monthly with the latest user interaction data.
   • CAdd a randomization component to the recommendations to force diversity.
   • DImplement a monitoring process that tracks model performance and data drift, and triggers a review when diversity metrics fall below a threshold.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which of the following is a primary purpose of data protection laws like GDPR in the context of ML operations?

   • ATo require all models to be deployed on-premises.
   • BTo standardize the algorithms used in machine learning models.
   • CTo ensure models are trained on the most recent data available.
   • DTo protect individuals' personal data and privacy rights.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A government agency is deploying an ML model to identify fraudulent benefit claims. The model has a high false-positive rate, which could lead to legitimate claimants being investigated. The agency's leadership is concerned about public trust. The data science team needs to communicate the model's limitations to the public. What is the most effective communication strategy?

   • AHold a public forum where the agency explains the model, its limitations, and the appeals process, and answers questions from the public.
   • BDelay the deployment until the model has a zero false-positive rate.
   • CPublish a detailed technical report on the model's accuracy and false-positive rate.
   • DIssue a press release that explains the model's purpose and the safeguards in place to protect claimants.
   Reveal Solution Discussion Select an answer first