MLA-C01 Question 142
Select 3A data science team is building a predictive model for loan approval decisions at a financial institution. Regulatory requirements mandate that the model's predictions must be interpretable for auditors and explainable to customers. Which algorithm(s) should the team consider to comply with these requirements while maintaining reasonable predictive performance?
- A
Linear Regression
- B
Gradient Boosted Trees (e.g., XGBoost)
- C
Decision Trees
- D
Deep Neural Networks
- E
Logistic Regression
Show answer and explanation
Correct answers: A, C, E
Explanation
When regulatory or customer-facing requirements demand interpretability, models like Linear Regression, Decision Trees, and Logistic Regression are preferred due to their inherent transparency and explainability. While Gradient Boosted Trees and Deep Neural Networks may offer better predictive performance, they are less interpretable. Gradient Boosted Trees can be explained using external tools, but this is not sufficient to meet strict interpretability requirements.
- A. Correct.
Linear Regression is inherently interpretable as it provides coefficients for each feature, indicating their relationship with the target variable. This makes it suitable for scenarios requiring interpretability.
- B. Incorrect.
Gradient Boosted Trees, while powerful in terms of predictive performance, are less interpretable compared to simpler models. However, tools like SHAP or LIME can help explain their predictions to some extent, though the interpretability is not intrinsic.
- C. Correct.
Decision Trees are inherently interpretable because they provide a clear path of decisions based on feature values, making them easy to explain to non-technical stakeholders.
- D. Incorrect.
Deep Neural Networks are typically considered black-box models due to their complex architectures, making them unsuitable for cases where interpretability is a requirement.
- E. Correct.
Logistic Regression is interpretable because it provides coefficients for each feature, similar to Linear Regression, making it suitable for scenarios where interpretability is important.