I was looking at this post which got me wondering: if an xreg for a fable model is hierarchical, does this influence model selection of the dependent variable? In other words, would using the hierarchical xreg vs an aggregate of the xreg values (I.e., the sums with no hierarchical structure) result in different models, assuming the dependent variable is an identical time series in both scenarios?
How to implement regressors in a Hierarchical Series in R, with the Fable package?
Haven’t tried this so I have no code.
