[R-sig-Geo] gwr, calculate hatmatrix when fit points are given

[Maximilian Sproß]

Ok, thank you for clarifying Roger!

best wishes,

Max

On 08/09/2012 02:15 PM, Roger Bivand wrote:
> On Thu, 9 Aug 2012, Maximilian Sproß wrote:
>
>> Dear list,
>>
>>
>> when running gwr() with given fit.points, it is not possible to 
>> calculate the hatmatrix. But it would be very helpful to get the 
>> fitting characteristics such as AIC or RMSE with the gwr output. In 
>> my case the fit points have to be set, as i run gwr on a cluster.
>
> According to information from the authors, it seems that GWR3 fits 
> without fit points first to generate a hat matrix for the data points. 
> This is then used in a further step for the fit points, for which no 
> covariates are observed. The assumption is that #data points << #fit 
> points, as when making a raster. This may be done I think by doing the 
> same, first running gwr() on the data points only and making the hat 
> matrix, next fitting on the fit points and including the output of the 
> first step in the fittedGWRobject= argument. However, the fitted 
> object will only help in giving local R2 (GWR4 version), and local 
> coefficient and prediction standard errors, not AIC.
>
> Roger
>
>>
>> I appreciate if anyone can help!
>>
>> all the best,
>>
>> Max
>>
>> _______________________________________________
>> R-sig-Geo mailing list
>> R-sig-Geo at r-project.org
>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>>
>


-- 
Mag. J. Maximilian Sproß
Institute of Geography, University of Innsbruck
Innrain 52
A-6020 INNSBRUCK

Tel. +43 (0)512 507 5413
web: http://www.uibk.ac.at/geographie/projects/lidar/