Closed famuvie closed 9 years ago
Thanks VS for reporting. This happens because the last random effect in the AR process is not associated with any observation. The last column(s) of the incidence matrix are (wrongly) discarded as they are zeros. Later multiplication with the predicted spatial process failed because of non-conformable dimensions. This is used for computing the fitted values, which are used to compute the residuals, which are used to compute the variogram.
VS reports this error when doing
where selmod is a spatial autoregressive model.