I guess 4.1 is definitely worth reading and will be cited this time for distinguishing model complexity from model flexibility for me… thanks for sharing so many helpful resources under this thread! https://t.co/nmaXANWFcO
89 followers
89 followers
1,109 followers
Adding here so folks reading the thread don't miss this. We should think critically about prediction vs explanation. See important notes added by @bolkerb and @f2harrell in the thread below.
5,366 followers
@bhjuarez I'm torn about this. I agree that PCs shouldn't *unthinkingly* be discarded, but IMO it can make sense to drop PCs as part of a principled/*a priori* choice of model complexity (cf @f2harrell RMS https://t.co/RaFe7UUlU2 §§4.1, 11.4)