[R-sig-ME] comparing GLS residuals with lme + correlation=corExp(form=~1, nugget=T) residuals
Ben Bolker
bbo|ker @end|ng |rom gm@||@com
Tue Oct 12 01:28:22 CEST 2021
Yes, that should be correct. type="normalized" should get you the
residuals corrected for autocorrelation and heteroscedasticity (if it's
in the model). A lot of people (including me!) routinely get confused
because the default residuals are "response" (i.e. accounting for
neither heterosced. nor correlation).
On 10/11/21 7:18 PM, Javier Moreira wrote:
> hi,
> i'm trying to make a comparison between a series of models, and I'm trying
> to understand how to find the autocorrelation that the lme + correlation
> model removed.
> Is the solution to just use type="normalized"? if I do this I'm getting the
> final residuals ( with the already subtract spatial ) or the residuals
> without considering spatial autocorrelation?
>
> if anybody has struggled with the same, I could use some help.
> thanks a lot!
> great forum!
>
> greatings
>
More information about the R-sig-mixed-models
mailing list