Predicting Hourly Trends

i'm curious about what you're optimizing for. Do you want to minimize on the weight time, or the chances that a bike is unavailable? What's the exact objective function you're chasing? In our case, we focused on one thing, to minimize the probability of a station to go out of stock. This is why, although the decision is at a daily level, having hourly predictions is very important. And so we have these two families of models, or the more traditional ones and the deep learning base ones. So where we use a collection of ornends as well as which is the model that we propose is a deep generative model.