I have been looking for this kind of stuff on the internet for a while and I cannot find any answer.
In the classical linear regression (without weights), one can compute the standard deviation and derive a prediction interval (+/- 1.96*sd) but I cannot figure out how to deal with it when used with weights. If I choose very low weights I will have a very tight interval which is not logical. However I want to take into account the fact that the model is improving because I put bigger weights on last values.
We have : σ^2 = 1/(n-p) Sum(w[i] * R[i]^2) (taken from R documentation).
I told myself that I should replace "n" by the sum of weights to make something more meaningful but what about "p"?
I am a bit confused about this stuff,
Thanks a lot if you know something about it..
Loïc