Evaluate Test Results CERTIFIED-DATA-SCIENCE-PRACTITIONER Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A model for predicting customer satisfaction (binary: satisfied/not satisfied) has a test-set accuracy of 90%, but the recall for the 'not satisfied' class is 0.15. The business wants to identify dissatisfied customers to offer them incentives. Which metric should be the primary focus?

   • ARecall for the 'not satisfied' class
   • BF1-score for the 'satisfied' class
   • CPrecision for the 'not satisfied' class
   • DOverall accuracy
   Reveal Solution Discussion Select an answer first
2. 2application · medium

   A data science team has trained a binary classifier to predict customer churn. On the held-out test set, the model achieves 92% accuracy, but the precision for the minority class (churners) is only 0.35. The business has stated that false positives (incorrectly flagging a customer as a churn risk) are costly because they trigger retention offers. Which evaluation metric should the team prioritize when deciding whether to deploy the model?

   • APrecision for the churn class
   • BOverall accuracy
   • CF1-score for the churn class
   • DRecall for the churn class
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   After evaluating a regression model on a test set, a data scientist observes that the mean squared error (MSE) is low, but the mean absolute error (MAE) is relatively high. What does this combination of results most likely indicate?

   • AThe model is underfitting the data, so the test set is not representative.
   • BThe model is overfitting the training data, so the test set should be reduced.
   • CThe model has many small errors, which increase the MAE more than the MSE.
   • DThe model has a few large errors, which inflate the MSE more than the MAE.
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A team evaluated a multiclass classifier on a test set. The confusion matrix shows that the model confuses classes 'dog' and 'wolf' frequently, but rarely confuses 'cat' with other classes. Which action is most likely to improve overall performance?

   • ASwitch to a regression model
   • BCollect more training data for the 'dog' and 'wolf' classes
   • CIncrease the test set size
   • DRemove the 'wolf' class from the dataset
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A data science team is evaluating a model for a churn prediction task. The model's test-set accuracy is 85%, but the recall for churners is 0.40. The business wants to identify as many churners as possible, even if it means more false positives. Which action would be most appropriate?

   • ARaise the classification threshold
   • BLower the classification threshold
   • CCollect more test data
   • DUse a different evaluation metric
   Reveal Solution Discussion Select an answer first