[R] R2 for SAR and validation
saniye keser
Hi everyone,
How can i obtain R2 for SAR model? and how can i validate the results, can i use the coefficients directly in a simple formula like y=b0+b1*x1+... or do i have to use the complicated formula for SAR (the one with the weight matrix and rho and...)?
Thanks for any help!
[[alternative HTML version deleted]]
Roger Bivand
Did you read the posting guide? Your posting is not comprehensible to most of
the readers of this list, and you have not explained whether this has
anything to do with R, or any of its contributed packages; there are no
clues in included code or code snippets illustrating the problem. Your TLA
is capable of many interpretations. Finally, if you are asking about spatial
autoregression, you might have posted to the R-sig-geo list instead.
If you are asking why lagsarlm() in the spdep package does not provide an R2
value in its summary, try using its log likelihood value, and comparing that
with logLik() of the equivalent lm() model - this is reported in the
summary() of the fitted object. You can calculate a pseudo-R2 by correlating
the observed and fitted values, but the likelihood ratio (or comparisons of
the AIC) are perhaps better founded. The same measures are found in other
implementations (GeoDa, Matlab spatial econometrics toolbox). It is not
obvious what you mean by "validate" in this context.
Roger Bivand
> ______________________________________________
> R-h...@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
>
--
View this message in context: http://www.nabble.com/R2-for-SAR-and-validation-tp25225433p25239901.html
Sent from the R help mailing list archive at Nabble.com.
______________________________________________
R-h...@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.