Arema Forecasting

The main idea came from that we wanted to simulate how oran ens and l t ms takes the imput to the gibart as well. So doing that, we ted to cut the data or the imput time series into windows and then predict a certain horizon. This is how we change the input of the g p art from being just naively feature engineered, or whatever in this case, to actually take the input as the oran and lt m. Will it be competitive to these models? And actually,. this what we found out. The most interesting part, is also the most interesting nding of our work, is how you re formulate such an forecasting task as it is