SHAP (SHapley Additive exPlanations) values are numeric measures of feature credit; they indicate how much each feature pushes a specific prediction higher or lower. Where features are correlated, credit will be distributed among them rather than assigned to a single feature. SHAP profiles show the relationship between feature credit and the feature's value (z-score). Below is an example SHAP profile for the paralog pair AKT1/AKT2.