Week 14: Interpretability Methods
Learning Objectives
By the end of this week, students will be able to:
- Distinguish intrinsic interpretability from post-hoc explainability
- Apply SHAP values to explain individual predictions and global feature importance
- Apply LIME to explain predictions from any black-box model
- Critically evaluate the limitations of explanation methods and connect them to accountability
Perspectival Reading
Reading: TBD — e.g., Molnar “Interpretable Machine Learning”
Reflection Questions
- Explanation methods produce simplified stories about complex models. When do those stories mislead rather than illuminate?
- Interpretability is often demanded for high-stakes decisions — but by whom, and for whom?
- Can an explanation be technically correct and still be ethically inadequate?
Slides
Notebook Demo
Open in Google Colab (link TBD)
Lab Assignment
Week 14 Lab — GitHub Classroom (link TBD)