Hello everyone!
Iíve been pondering this question for some time. Iíve searched online and on this forum and I havenít found the answer to my question.
Throughout my searches online Iíve found that Shapley values are obtained by running all the possible combinations of predictors against the dependent variable and by averaging their contribution to the R2 afterwards. From my understanding, relaimpo package in R is ďdecomposingĒ the R2 and providing the relative importance of each predictor. It seems to me that both do pretty much the same thing, but I want to be sure that Iím not making a mistake.
Therefore my question would be: is it correct to use relaimpo package in R to estimate Shapley values for my predictors? If not, what would be the alternative?
Thanks a lot
Alexandra