Oversee AI/ML Model Technique(s) (e.g., Algorithm, Selection) PMI-CPMAI Practice Questions (Page 6)

Questions 26–29

1. 26expert · hard

   A project manager is reviewing the algorithm selection documentation for a churn prediction model. The documentation states that a random forest was chosen because it had the highest AUC on the test set. However, the business team needs to explain the model's decisions to customers, and the documentation does not address interpretability. What is the most critical issue with the documentation?

   • AThe documentation does not include the exact hyperparameters used.
   • BThe documentation does not mention the size of the training dataset.
   • CThe documentation does not list the alternative algorithms considered.
   • DThe documentation lacks a discussion of interpretability trade-offs.
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   Which scenario is best suited for a supervised learning approach?

   • AReducing the dimensionality of a high-dimensional dataset for visualization
   • BEstimating the price of a house using historical sales data with known sale prices
   • CTraining an agent to play a game by maximizing cumulative rewards
   • DGrouping customers into segments based on purchasing behavior without predefined categories
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   A project manager is reviewing the documentation of algorithm selection criteria. What is the primary purpose of this review?

   • ATo verify that the algorithm is the most popular in the industry
   • BTo replace the data scientists' technical decisions with the project manager's preferences
   • CTo shorten the documentation to save time
   • DTo ensure the criteria are comprehensive, transparent, and aligned with stakeholder needs
   Reveal Solution Discussion Select an answer first
5. 29foundation · easy

   A project manager is overseeing the initial phase of an AI project. The team needs to systematically evaluate several candidate algorithms to determine which best fits the problem. Which approach best describes a systematic method for researching and evaluating candidate algorithms?

   • AAsking the most senior data scientist to pick an algorithm based on experience alone
   • BRunning a literature review, benchmarking on representative data, and comparing performance metrics against defined criteria
   • CSelecting the algorithm that is most popular in the industry without considering the specific data characteristics
   • DChoosing the algorithm that is easiest to implement regardless of performance
   Reveal Solution Discussion Select an answer first