Re: [R-sig-Geo] spgwr and selecting aic for bandwidth configuration

Roger Bivand Thu, 06 May 2010 10:59:45 -0700

On Sat, 10 Apr 2010, Christopher Cardinal wrote:

I am trying to use gwr.sel to apply the AIC method for bandwidth selection
in SPGWR. R version is 2.10.1 (used for compatibility with Rpy--however, the
following problem occurs directly from within R). SPGWR version is 0.6-4.
First I run global regression, then gwr()with default CV method for
bandwidth selection with no problems. Then I run "aic" bandwidth selection
method, which outputs bandwidth results. Then I run gwr(), which produces no
output. Perhaps I'm using the wrong command? I can't find many examples of
gwr() implemented with "aic". Please share your ideas if you have any.

Sorry not to have replied earlier. There is no summary() method for "gwr"objects, only a print() method. The output of gwr.sel() is a singlenumber. It is impossible to say what is in data_bw_aic - it should besimilar to the last number shown in the verbose output - but theoptimisation in gwr.sel() does not seem to have converged - I don't thinkthat you have paid attention to error messages, or at least not shown themin your posting. Without a reproducible example, it is impossible to help.Try to reproduce your problem with the columbus or georgia data setsshipped with the package.


Hope this helps,

Roger

Thanks.

library(spgwr)

Loading required package: sp
Loading required package: maptools
Loading required package: foreign
Loading required package: lattice
NOTE: default kernel and CV criteria changed
see help pages for details

library(foreign)
data <-

read.dbf("C:/GIS/Hunter/Thesis/GISData/Parcel/Albany/AlbCnty/Polygon2004/test2.dbf")

data <- subset(data, GOOGDIST > 0)
data <- subset(data, ACRES > .5)
data_lm <- lm(SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH, data =

data)

data_lm_sum <- summary(data_lm)
tmp<-print(data_lm_sum)


Call:
lm(formula = SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH,
   data = data)

Residuals:
   Min      1Q  Median      3Q     Max
-890883  -92357  -66457   76176 1550181

Coefficients:
             Estimate Std. Error t value Pr(>|t|)
(Intercept)  8.997e+04  2.566e+04   3.507 0.000483 ***
ACRES       -1.999e+03  3.915e+02  -5.106 4.26e-07 ***
CUR_TOT_A    1.993e+00  9.443e-02  21.104  < 2e-16 ***
HHINC       -7.636e-02  5.363e-01  -0.142 0.886810
PERGTBACH    9.287e+03  5.020e+04   0.185 0.853272
---
Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1

Residual standard error: 186700 on 689 degrees of freedom
Multiple R-squared: 0.4061,     Adjusted R-squared: 0.4027
F-statistic: 117.8 on 4 and 689 DF,  p-value: < 2.2e-16

data_bw <- gwr.sel(SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH ,

data = data, coords = cbind(data$X, data$Y))
Bandwidth: 79378.23 CV score: 2.508302e+13
Bandwidth: 128308.4 CV score: 2.549703e+13
Bandwidth: 49137.69 CV score: 2.444211e+13
Bandwidth: 30448.02 CV score: 2.369592e+13
Bandwidth: 18897.16 CV score: 2.35517e+13
Bandwidth: 17792.42 CV score: 2.365362e+13
Bandwidth: 23918.27 CV score: 2.340727e+13
Bandwidth: 26412.41 CV score: 2.348778e+13
Bandwidth: 23178.31 CV score: 2.339931e+13
Bandwidth: 22740.56 CV score: 2.339894e+13
Bandwidth: 22908.52 CV score: 2.339868e+13
Bandwidth: 22910.54 CV score: 2.339868e+13
Bandwidth: 22909.77 CV score: 2.339868e+13
Bandwidth: 22909.78 CV score: 2.339868e+13
Bandwidth: 22909.78 CV score: 2.339868e+13
Bandwidth: 22909.78 CV score: 2.339868e+13
Bandwidth: 22909.78 CV score: 2.339868e+13

data_gauss <- gwr(SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH, data

= data, coords = cbind(data$X, data$Y), bandwidth = data_bw, hatmatrix =
TRUE)

data_gauss_sum <- summary(data_gauss)
print (data_gauss_sum)

         Length Class                  Mode
SDF            1 SpatialPointsDataFrame S4
lhat      481636 -none-                 numeric
lm            11 -none-                 list
results       14 -none-                 list
bandwidth      1 -none-                 numeric
adapt          0 -none-                 NULL
hatmatrix      1 -none-                 logical
gweight        1 -none-                 character
this.call      6 -none-                 call

print (data_gauss)

Call:
gwr(formula = SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH,
   data = data, coords = cbind(data$X, data$Y), bandwidth = data_bw,
   hatmatrix = TRUE)
Kernel function: gwr.Gauss
Fixed bandwidth: 22909.78
Summary of GWR coefficient estimates:
                  Min.    1st Qu.     Median    3rd Qu.       Max.
Global
X.Intercept. -2.883e+05 -4.166e+04  1.135e+05  1.655e+05  2.834e+05
89969.4089
ACRES        -3.259e+03 -2.678e+03 -1.572e+03  9.080e+02  8.611e+03
-1998.9124
CUR_TOT_A     2.233e-02  1.261e+00  1.628e+00  2.049e+00  2.380e+00
1.9928
HHINC        -5.563e+00 -1.219e+00  3.129e-01  2.753e+00  8.023e+00
-0.0764
PERGTBACH    -3.023e+05 -9.237e+04 -5.303e+04 -7.264e+03  1.014e+05
9287.3433
Number of data points: 694
Effective number of parameters (residual: 2traceS - traceS'S): 29.87522
Effective degrees of freedom (residual: 2traceS - traceS'S): 664.1248
Sigma (residual: 2traceS - traceS'S): 170145.2
Effective number of parameters (model: traceS): 21.99972
Effective degrees of freedom (model: traceS): 672.0003
Sigma (model: traceS): 169145.3
Sigma (ML): 166442.7
AICc (GWR p. 61, eq 2.33; p. 96, eq. 4.21): 18704.23
AIC (GWR p. 96, eq. 4.22): 18678.59
Residual sum of squares: 1.922601e+13

data_bw_aic <- gwr.sel(SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH

, data = data, coords = cbind(data$X, data$Y), method="aic")
Bandwidth: 79378.23 AIC: 18795.23
Bandwidth: 128308.4 AIC: 18811.03
Bandwidth: 49137.69 AIC: 18768.93
Bandwidth: 30448.02 AIC: 18733.32
Bandwidth: 18897.16 AIC: 18678.52
Bandwidth: 11758.34 AIC: 18593.90

data_bw_aic_sum <- summary(data_bw_aic)
data_bw_aic_sum
print(data_bw_aic)
data_gauss_aic <- gwr(SALE_PRICE ~ ACRES + CUR_TOT_A + HHINC + PERGTBACH,

data = data, coords = cbind(data$X, data$Y), bandwidth = data_bw_aic,
hatmatrix = TRUE)

print(data_gauss_aic)
summary(data_gauss_aic)



data looks like this:
       UID YEAR        X       Y      LAT      LONG       ACRES SLOPE
CUR_TOT_A     FLOODPER
1      6482 2004 698260.3 1446285 42.80053 -73.73233   0.7309970     1
40000 0.0000000000
5      8554 2004 695351.2 1438002 42.77801 -73.74342   0.8890091     1
7400 0.0000000000
7     15413 2004 696033.0 1431745 42.76088 -73.74108   0.5565335     1
50000 0.0000000000
10     8220 2004 680872.2 1435917 42.77268 -73.79744   0.8693349     1
27200 0.0000000000
11     8221 2004 681023.1 1436053 42.77303 -73.79688   0.8018674     1
25900 0.0000000000
12     8792 2004 691488.3 1439555 42.78245 -73.75765   0.8236566     1
42000 0.0000000000
13     8793 2004 691556.9 1439692 42.78277 -73.75746   0.8610513     1
42000 0.0000000000
14    14411 2004 682812.3 1426310 42.74625 -73.79048   1.9767706     1
150000 0.0000000000
15    14656 2004 688672.5 1432607 42.76348 -73.76833   3.5036509     1
65700 0.0000000000
16    15103 2004 686513.4 1428931 42.75343 -73.77653   4.5957896     1
352800 0.0000000000
19    89916 2004 702825.7 1439163 42.78107 -73.71557   5.4003465     1
23200 0.0000000000
22    93898 2004 704736.3 1432229 42.76197 -73.70865   0.9085830     1
8400 0.0000000000
25    10109 2004 657502.9 1427039 42.74891 -73.88458   3.1770620     1
53600 0.0000000000
26    10296 2004 657568.5 1427583 42.75031 -73.88438   0.9898759     1
39200 0.5419490988

                                                        GOOGADDR GOOGDIST
PROP_CLASS SALE_PRICE HHINC
1                       1251 New Loudon Rd, Cohoes, NY 12047, USA
21597        311      42500 25964
5                             22 Meadow St, Cohoes, NY 12047, USA
19195        311      31500 25964
7                            44 Graffin Dr, Latham, NY 12110, USA
18764        311     396500 25964
10                      4119-4195 River Rd, Latham, NY 12110, USA
18941        311      71454 16222
11                      4119-4195 River Rd, Latham, NY 12110, USA
18941        311      71454 16222
12                   112 Dunsbach Ferry Rd, Cohoes, NY 12047, USA
19164        311      65000 16222
13                   114 Dunsbach Ferry Rd, Cohoes, NY 12047, USA
19208        311      68000 16222
14                             155 Wade Rd, Latham, NY 12110, USA
13289        330     100000 16222
15                      100 Sparrowbush Rd, Latham, NY 12110, USA
16458        311     275000 16222
16                           4 Stanley Cir, Latham, NY 12110, USA
15605        330     307000 16222
19                            28 Edward St, Cohoes, NY 12047, USA
21813        311     114685 51214
22                            8 Oxford Cir, Cohoes, NY 12047, USA
22499        311     342551 51214
25                           72 Morris Rd, Colonie, NY 12205, USA
14924        330     190000 19865
26                       79 Morris Rd, Schenectady, NY 12304, USA
14929        330      18000 19865

     PERGTBACH PERLTBACH PERCAPINC
1    0.11521926 0.8847807     14998
5    0.11521926 0.8847807     14998
7    0.11521926 0.8847807     14998
10   0.12627425 0.8737257     12147
11   0.12627425 0.8737257     12147
12   0.12627425 0.8737257     12147
13   0.12627425 0.8737257     12147
14   0.12627425 0.8737257     12147
15   0.12627425 0.8737257     12147
16   0.12627425 0.8737257     12147
19   0.37195452 0.6280455     25915
22   0.37195452 0.6280455     25915
25   0.16253444 0.8374656     10992
26   0.16253444 0.8374656     10992

        [[alternative HTML version deleted]]


--
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.biv...@nhh.no

_______________________________________________
R-sig-Geo mailing list
R-sig-Geo@stat.math.ethz.ch
https://stat.ethz.ch/mailman/listinfo/r-sig-geo

Re: [R-sig-Geo] spgwr and selecting aic for bandwidth configuration

Reply via email to