Oversee AI/ML Model Technique(s) (e.g., Algorithm, Selection) PMI-CPMAI Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A healthcare startup is developing a model to predict patient readmission risk. They have a small labeled dataset with many features. The team is considering several algorithms. What is the most systematic first step in researching candidate algorithms?

   • AAsk the business stakeholders to decide the algorithm.
   • BChoose a deep learning model because it is the most advanced.
   • CRun all candidate algorithms on the full dataset and pick the one with the highest training accuracy.
   • DReview literature and prior studies on similar prediction tasks to identify promising algorithms.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   What is a key trade-off when choosing between a simple linear model and a complex ensemble model?

   • AComplex models require less data to train
   • BSimple models always have lower accuracy than complex models
   • CComplex models are always more interpretable
   • DSimple models are easier to interpret but may underfit complex patterns
   Reveal Solution Discussion Select an answer first
4. 8application · medium

   A bank is building a credit risk model. They have a large labeled dataset and need high accuracy, but regulators require the model to be auditable and decisions explainable to customers. The project manager must balance model complexity with interpretability. Which approach best meets these requirements?

   • AUse a black-box deep learning model without explanation.
   • BUse a highly complex ensemble model and rely on post-hoc explanations.
   • CUse an unsupervised anomaly detection model.
   • DUse a simple logistic regression model.
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A marketing team has a large dataset of customer transactions but no predefined customer segments. They want to discover natural groupings to tailor campaigns. The project manager needs to select an approach that is exploratory and does not require labeled outcomes. Which technique should be recommended?

   • ARandom forest classification
   • BK-means clustering
   • CLinear regression
   • DQ-learning
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   A team wants to build a model to predict whether a customer will churn based on historical data that includes past churn outcomes. Which learning approach is most appropriate?

   • AReinforcement learning
   • BUnsupervised learning
   • CSupervised learning
   • DSemi-supervised learning
   Reveal Solution Discussion Select an answer first