> All of them are flawed, either in their mathematics or assumptions (that will most likely never be true, like assumptions of linearity and your data set being

> All of them are flawed, either in their mathematics or assumptions (that will most likely never be true, like assumptions of linearity and your data set being on a manifold). Although I can see global linearity being unlikely in most cases; why is local linearity unlikely?