Different AI Models and Development Processes AIE Practice Questions (Page 8)

Questions 36–38

1. 36application · medium

   A data science team is developing a spam detection model. After the first training iteration, the model achieves high precision but low recall. The business wants to ensure that important emails are not marked as spam. What should the team do?

   • ASwitch to an unsupervised model to avoid the precision-recall trade-off
   • BDeploy the model because high precision is the only important metric
   • CAdjust the classification threshold to increase recall, even if precision decreases slightly
   • DCollect more training data without changing the model
   Reveal Solution Discussion Select an answer first
2. 37expert · hard

   An e-commerce company wants to recommend products to users. They have a large dataset of user purchase history and product ratings. The goal is to maximize user engagement, but the system must also handle new users with no purchase history. Which model selection approach best addresses this constraint?

   • AUse a collaborative filtering model only
   • BUse a content-based filtering model only
   • CUse a hybrid model combining collaborative and content-based filtering
   • DUse a reinforcement learning model to explore user preferences
   Reveal Solution Discussion Select an answer first
3. 38application · medium

   A logistics company wants to build a model to predict delivery delays. The team has defined the problem and collected historical shipment data. They are about to start training the model. What should they do before training to ensure reliable evaluation?

   • ASplit the data into training, validation, and test sets
   • BSelect a model type without any data preparation
   • CDeploy the model to production immediately
   • DCollect more data from external sources
   Reveal Solution Discussion Select an answer first