[R-sig-Geo] GWR SDF

[Roger Bivand]

On Mon, 13 Nov 2006, Raphael Saldanha wrote:

> Thanks Roger! I'm just making some tests with the method
> 
> Take a look below, this is what I have.
> 
> > summary(lm(V03 ~ V04 + V05, as.data.frame(x)))
> 
> Call:
> lm(formula = V03 ~ V04 + V05, data = as.data.frame(x))
> 
> Residuals:
>     Min      1Q  Median      3Q     Max
>  -5.909  -3.855   -3.380  -2.694 185.091
> 
> Coefficients:
>             Estimate Std. Error t value Pr(>|t|)
> (Intercept)   5.9086     1.9982   2.957  0.00324 **
> V04           0.8069     0.3342   2.415  0.01607 *
> V05           0.1833     0.3356   0.546  0.58524
> ---
> Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
> 
> Residual standard error: 16.84 on 558 degrees of freedom
> Multiple R-Squared: 0.965,      Adjusted R-squared: 0.9649
> F-statistic:  7701 on 2 and 558 DF,  p-value: < 0.00000000000000022
> 
> > res.adpt
> Call:
> gwr(formula = V03 ~ V04 + V05, data = x, adapt = x.adapt.gauss,
>     hatmatrix = TRUE)
> Kernel function: gwr.gauss
> Adaptive quantile: 0.1087359 (about 61 of 561)
> Summary of GWR coefficient estimates:
>                  Min.  1st Qu.   Median  3rd Qu.     Max. Global OLS
> X.Intercept. -16.0400   0.7464   2.7780   7.1090   42.5200     5.9086
> V04           -0.5183   0.6884   0.8273   1.4080   3.7480     0.8069
> V05           -2.6950  -0.4328   0.1553   0.2978   1.4020     0.1833
> Number of data points: 561
> Effective number of parameters: 37.25909
> Effective degrees of freedom: 523.7409
> Sigma squared (ML): 240.1169
> AICc (GWR p. 61, eq 2.33; p. 96, eq. 4.21): 4723.854
> AIC (GWR p. 96, eq. 4.22): 4692.986
> Residual sum of squares: 134705.6
> 
> > names(res.adpt$SDF)
>  [1] "sum.w"        "X.Intercept." "V04"          "V05"          "gwr.R2"
> "X1"           "X2"           "X3"
>  [9] "coord.x"      "coord.y"

In order: sum of weights, three local coefficient estimates, the local 
R-square, three local coefficient standard errors, and the data point 
coordinates. I was puzzled that you said you included V06, V07, V08, but 
got names V04, V05, but understand that that was just an example.

> 
> 
> On 11/13/06, Roger Bivand <Roger.Bivand at nhh.no> wrote:
> >
> > On Mon, 13 Nov 2006, Raphael Saldanha wrote:
> >
> > > Help!
> > >
> > > In the SDF results, what exactly means these fields:
> > >
> > > sum_w
> > > X.Intercept
> > > V04
> > > V05
> > > X1
> > > X2
> > > X3
> >
> > The spgwr package is not really intended to help people use GWR, which is
> > not generally accepted as a technique of analysis (because it forces
> > coefficients to co-vary), though it can be used for exploration. It is
> > rather a toolbox for examining the method. For this reason, not much
> > effort has been put into things like names. You would have to say what all
> >
> > the names are in this case:
> >
> > names(res.adpt$SDF)
> >
> > The v04, V05 look odd, but without seeing all the names, it is difficult
> > to tell. The X1-X3 are probably three of the local standard errors. s_w is
> > the sum of weights at that point, and X.Intercept is the local intercept
> > estimate.
> >
> > What happens when you run a regular regression? Do any of the coefficients
> > disappear (any collinear variables)?
> >
> > Roger
> >
> > >
> > > I'm using this code:
> > >
> > > > x <- readShapePoly("domicilio_3136702.shp", IDvar = "ID_")
> > > > x.adapt.gauss <- gwr.sel(V03 ~ V06 + V07 + V08, data=x, adapt=TRUE)
> > > > res.adpt <- gwr(V03 ~ V06 + V07 + V08, data=x, adapt= x.adapt.gauss,
> > > hatmatrix = TRUE)
> > >
> > >
> >
> > --
> > Roger Bivand
> > Economic Geography Section, Department of Economics, Norwegian School of
> > Economics and Business Administration, Helleveien 30, N-5045 Bergen,
> > Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
> > e-mail: Roger.Bivand at nhh.no
> >
> >
> 
> 
> 

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no