Selecting a Form of ML CT-AI Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A healthcare startup has a dataset of patient records with labels for a rare disease. The dataset is highly imbalanced: only 1% of patients have the disease. The team wants to build a model to screen patients, and they need to minimize false negatives (missing the disease) while keeping false positives manageable. Which approach is most appropriate?

   • AUse supervised classification with a focus on recall for the positive class
   • BUse supervised regression to predict the probability of the disease
   • CUse unsupervised clustering to identify disease subgroups
   • DUse reinforcement learning to learn a screening policy
   Reveal Solution Discussion Select an answer first
2. 12expert · hard

   A marketing agency has a dataset of customer interactions but no labels. They want to segment customers into groups for targeted campaigns. The team is considering two approaches: (A) use clustering to find natural groups, or (B) use dimensionality reduction to visualize the data and then manually define segments. Which approach is more appropriate for the stated goal?

   • AUse clustering to automatically identify distinct customer segments
   • BUse reinforcement learning to optimize campaign targeting
   • CUse dimensionality reduction only, then manually define segments
   • DUse supervised classification with manually labeled segments
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   In reinforcement learning, what is the role of the 'policy'?

   • AIt determines which action the agent should take in a given state.
   • BIt represents the external system with which the agent interacts.
   • CIt defines the set of all possible actions the agent can take.
   • DIt provides the reward signal to the agent after each action.
   Reveal Solution Discussion Select an answer first
5. 14foundation · easy

   What is the primary goal of clustering in unsupervised learning?

   • ATo reduce the number of features by combining them into a smaller set of new features.
   • BTo predict a numeric outcome based on input features.
   • CTo group similar data points together based on their features without using predefined labels.
   • DTo assign a class label to each data point using a training set of labeled examples.
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   Which type of machine learning is characterized by training on data that includes both input features and the corresponding correct output labels?

   • AReinforcement learning
   • BSemi-supervised learning
   • CSupervised learning
   • DUnsupervised learning
   Reveal Solution Discussion Select an answer first