How to Make a Distributed Forecasting Effect?

We just consider that single on time serious and commersionally, we do a rema model. For each smaller trunk of time, for could be within a few hours or within a few days, you can make u arema model like rima, three, one, five. And in the end, for the long time span, you can put those small rema modes together. We capture the very long, but slowly changing time train. That is advantage. So if you look at our application of our paper, you will notice that commensorate to arimemodo, iw end up with a really stationary forecast.