Thanks guys. I think GP models are great to do dimensionality reduction but I am not sure of their generalization capability in nonstationnary...
Is anyone using Gaussian Process models to build indicators. Something I'd like to investigate to model cycles.
Separate names with a comma.