The paper relies quite a bit on the quality of IGCS by looking at ABC / AUC. Would help to include the code for that as well here [would also help readers of paper if that is coded up for comparison; that part tbh was not entirely clear how it's calculated -- at least when quickly skimming paper].
Thank you for the comments.
I added the example that calculate feature contribution of a linear regression model and compare it to those of cohort shapley and kernel SHAP in terms of insertion/deletion ABC.
The paper relies quite a bit on the quality of IGCS by looking at ABC / AUC. Would help to include the code for that as well here [would also help readers of paper if that is coded up for comparison; that part tbh was not entirely clear how it's calculated -- at least when quickly skimming paper].