shapper
shapper copied to clipboard
implement SHAP feature importance
I implemented the calculation of SHAP feature importance values in the "shap_feature_importance()" function. Docs were updated accordingly.
How it works: SHAP values are calculated for each instance of the given data. The mean of the absolute SHAP values for each feature is then returned.
Additionally, to cope with large data sets, this implements the kmeans function of the SHAP python lib. (options "kmeans" and "k")
Hey, Your function looks cool. I have not accepted PR because of failing Travis checks. Are you going to fix it? :)