Which method explains a decision by showing what would need to change for a different outcome?

Which method explains a decision by showing what would need to change for a different outcome?

• A. Counterfactual Explanations
• B. Surrogate Models
• C. Automation Bias
• D. Shapley Values

Answer: (A)

Counterfactual explanations explain a decision by showing what would need to change in the input to produce a different outcome. This approach gives you a concrete sense of which factors would flip the result and by how much, framing the model’s decision as something actionable rather than abstract. For example, it might indicate that increasing annual income by a certain amount or improving a credit score by a set number of points would change an approval to a denial or vice versa. This makes it easy to understand leverage points and what to adjust if you want a different result next time. Surrogate models, while useful for interpretable approximations, don’t inherently reveal the exact input changes needed to alter the outcome. Shapley values describe how much each feature contributed to the observed decision, but they don’t specify what changes would be sufficient to obtain a different outcome. Automation bias is a cognitive tendency to rely on automated systems too heavily, not a method for explaining decisions.

Counterfactual explanations explain a decision by showing what would need to change in the input to produce a different outcome. This approach gives you a concrete sense of which factors would flip the result and by how much, framing the model’s decision as something actionable rather than abstract. For example, it might indicate that increasing annual income by a certain amount or improving a credit score by a set number of points would change an approval to a denial or vice versa. This makes it easy to understand leverage points and what to adjust if you want a different result next time. Surrogate models, while useful for interpretable approximations, don’t inherently reveal the exact input changes needed to alter the outcome. Shapley values describe how much each feature contributed to the observed decision, but they don’t specify what changes would be sufficient to obtain a different outcome. Automation bias is a cognitive tendency to rely on automated systems too heavily, not a method for explaining decisions.