Hands-On Exercise: Model Explainability CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   A SHAP summary plot shows that the feature 'credit_score' has a wide spread of SHAP values, with high values pushing predictions toward approval. What does this indicate?

   • AThe feature has a strong and consistent influence on predictions.
   • BThe feature is irrelevant to the model's decisions.
   • CThe model is overfitting to the feature.
   • DThe feature is highly correlated with another feature.
   Reveal Solution Discussion Select an answer first
2. 2application · medium

   A team is using SHAP to explain a model that predicts customer churn. They notice that the SHAP values for a specific feature are consistently high across many instances. What is the most appropriate conclusion about this feature?

   • AThe feature is only important for a few specific instances.
   • BThe feature is causing the model to make incorrect predictions.
   • CThe feature is likely an important driver of churn predictions across the dataset.
   • DThe feature has no impact on the model's predictions.
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A data scientist runs SHAP on a credit-scoring model and receives the following output for a single applicant: 'credit_history' = +0.35, 'income' = +0.20, 'debt_ratio' = -0.10, 'age' = +0.02. The model's base value is 0.5 and the prediction is 0.97. Which interpretation of this output is correct?

   • AThe positive SHAP values for credit_history and income pushed the prediction above the base value, while debt_ratio slightly reduced it.
   • BThe model's prediction is 0.97 because the base value is 0.5 and all features have positive contributions.
   • CThe age feature has no influence on the model's prediction because its SHAP value is close to zero.
   • DThe negative SHAP value for debt_ratio means the model predicts the applicant will default.
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   A tester uses LIME to explain a prediction from a black-box model. What does LIME require as input?

   • AThe model's source code
   • BThe training data and the model's prediction function
   • CThe internal weights of the model
   • DA separate interpretable model trained on the same data
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A data scientist is interpreting a SHAP force plot for a single prediction from a house-price model. The base value is $300,000, and the prediction is $350,000. The force plot shows that 'location' pushes the prediction up by $40,000, while 'square_footage' pushes it down by $10,000. What is the correct interpretation of this force plot?

   • AThe location feature increased the prediction, while square_footage decreased it, resulting in a net increase of $30,000 over the base value.
   • BThe base value is the minimum possible prediction of the model.
   • CThe square_footage feature is the most important feature in the model.
   • DThe model is overpricing the house because the prediction is higher than the base value.
   Reveal Solution Discussion Select an answer first