Data Skeptic Shapley Values
Mar 6, 2020
Linda Tran, a conversational co-presenter with practical instincts, joins a hands-on chat about using Shapley values for home renovation choices. They frame renovation items as coalition players. Short demos show how order and averaging change credit allocation. They also cover computational costs and how to apply Shapley ideas to model interpretability and real decisions.
AI Snips
Chapters
Transcript
Episode notes
Using Shapley To Prioritize Renovation Choices
- Linda and Kyle frame Shapley values around a real decision: which renovation items to prioritize within a budget.
- They weigh trade-offs like tile vs. sink to illustrate why fair contribution attribution matters.
Shapley Values Equalize Order Effects
- Shapley values assign each contributor an average marginal contribution across all possible joining orders.
- This controls for order-dependence and reveals true relative value when contributions interact or overlap.
Game Theory Roots Explain Fairness
- Shapley values originate in cooperative game theory for splitting group rewards fairly.
- They transfer naturally to ML interpretability by attributing model decisions to input features.