Dear all,
In the spdep package, there is an eigenvector spatial filtering function (SpatialFiltering).
In the package document, it can be used with Linear regression (lm) to remove
spatial autocorrelation by adding it to the lm formula as a covariate.
The example does not provide though, how to apply this Spatial filtering regression (sfr) on
unseen (hold-out) sample data. I'd like to predict on 'test' data using the fitted sfr model.
Any insight is welcome.
chris