A deep dive into EnbPI, a Conformal Prediction method for time collection forecasting
Most of my latest articles revolved round realizing how positive a mannequin is about its predictions. If we all know the uncertainty of predictions, we are able to make well-informed selections. I confirmed you the way we are able to use Conformal Prediction to quantify a mannequin’s uncertainty. I wrote about Conformal Prediction approaches for classification and regression issues.
For these approaches, we assume that the order of remark doesn’t matter, i.e., that our knowledge is exchangeable. That is cheap for classification and regression issues. Nevertheless, the idea doesn’t maintain for time collection issues. Right here, the order of observations typically comprises vital info, comparable to traits…
