Deep Generative Model Approach

Generative models give you, if in a way, forr free, a fully probablistic can of framework to your model. And so this is saying, we don't really forecast just an exact number of bikes for to morrow, but we give you a full confident interval of where we think our demand for tomorrow will fall. It seems logical that maybe the system could learn to express the hidden variable to describe shared things across these different stations. Ye, it seems intuitive to me that that could be useful in this problem.