SHAP

• @lundbergUnifiedApproachInterpreting2017
• help users interpret the predictions of complex models
• unclear how these methods are related and when one method is preferable over another
• unified framework for interpreting predictions
• SHAP
• SHapley Additive exPlanations
• game theoretic approach to explain the output of any machine learning model
• connects optimal credit allocation with local explanations using the classic Shapley values from game theory and their related extensions
• assigns each feature an importance value for a particular prediction
• identification of a new class of additive feature importance measures
• theoretical results showing there is a unique solution in this class with a set of desirable properties
• notable because several recent methods in the class lack the proposed desirable properties
• present new methods that show improved computational performance and/or better consistency with human intuition than previous approaches