SHAP: Exact Shapley Values
Advanced
The problem, from first principles
SHAP (Lundberg & Lee, 2017) explains a prediction by splitting it among the input features. The Shapley value of a feature is its average marginal contribution over every order in which features could be revealed, starting from a baseline.
From theory to code
Implement shapley_exact(f, x, baseline), which averages each feature's marginal contribution over all permutations.
Constraints
- This is exact enumeration, so it is practical only for a few features.
Hints
Hint 1
For each permutation, reveal features one at a time from the baseline and record how much each one changes the output. Average over permutations.
Keywords
research-papersclassical-mlinterpretabilityshap
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.