Code changes to detect an discord time series

[matbord]

Hi @datacubeR , I am changing the code in the repo in order to detect discord time series from a dataset. From your paper I understood that I have to change p_w (e.g p_w=10) to predict the next 10 values in the window. From your paper I also understood that the window is sliding by 1 index at time, hence the preds_losses variable contains the losses for each single index and not for each single "window of 10 indexes". How can I change the code in order to predict the next discord time series? How can I plot it as in your paper?