Hi all,
I would like to predict prediction interval for a SAR model (function errorsarlm in R - package spdep). While the function predict.lm allows to set interval='prediction' parameter to predict the upper and lower boundary of predictions, the function predict.sarlm does not. Is it because it is incorrect to estimate the prediction interval of a SAR or simply because it hasn't implemented? Is it possible to calculate those prediction intervals manually? Below a replicable example:
library(spdep)
data(oldcol)
lw
Hi,
Using the model errors to predict function and quantile of the t-student
PR1
Hi Sergio, thank you a lot. Just a clarification, does your code predict the prediction interval or the confidence interval?
Routine is for confidence intervals can be constructed prediction intervals using the diagonal of the hat matrix to evaluate the influence of observations, is a linear approximation.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package