I'm trying to implement a multivariate nearest neighbour method in R to use in time-series forecasting (reference: Reick and Page, 2000). I managed to validate the model (with cross-validation), and now I'd like to calculate prediction intervals for the forecast values. My question is, how to calculate them, and what is the best way to do it?

I found two related questions here and here that discuss a similar issue for univariate data, and from the answers I'm not sure if bootstrapping would be suitable for this situation...

Any thoughts and suggestions would be highly appreciated, I've never worked with this type of forecasting method before.

EDIT: I just found this reference that seems useful, I'm studying the article now... still any suggestions on this matter would be appreciated.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.