[R-sig-Geo] spatialreg::predict.sarlm

[Weldensie Embaye]

I am trying to run  out-of-sample prediction
using  spatialreg::predict.sarlm package. I tried to prepare my data set as
per the requirements.However, it did not work.

Below is the code.
 csv_train <- read.csv("train.csv") #upload your csv. R needs to know your
region ID
> hhid <- csv_train$hhid #Explicitly created a variable to hold the regions
> nb_train <- read.gwt2nb('train.gwt', region.id=hhid)
> Q_train<-nb2listw(nb_train)
> csv_test <- read.csv("test.csv") #upload your csv. R needs to know your
region ID
> hhid <- csv_test$hhid #Explicitly created a variable to hold the regions
> nb_test <- read.gwt2nb('test.gwt', region.id=hhid)
> Q_test<-nb2listw(nb_test)
> y_test <- dataTable$annualrent[test]
> ######## spatial lag process model
> GM5<-spatialreg::lagsarlm(formula, data=traindata, listw=Q_train)
> summary(GM5)

Call:spatialreg::lagsarlm(formula = formula, data = traindata, listw =
Q_train)

Residuals:
      Min        1Q    Median        3Q       Max
-54.38677  -7.32082  -0.78772   5.69047  90.56889

Type: lag
Coefficients: (asymptotic standard errors)
             Estimate Std. Error z value  Pr(>|z|)
(Intercept) -20.63671    4.08382 -5.0533 4.343e-07
coveredpri   -4.04663    4.71961 -0.8574  0.391219
coveredsha   -3.70054    2.57827 -1.4353  0.151207
vipprivate   -3.19640    4.57774 -0.6982  0.485022
vipshare     -2.26742    2.30487 -0.9838  0.325238
unoveredla   -1.29846    3.62671 -0.3580  0.720322
flushpriva   10.22320    3.31060  3.0880  0.002015
electricit   11.25641    2.07508  5.4246 5.810e-08
privatetap    6.48159    2.63420  2.4606  0.013872
publictap    -2.91930    3.46341 -0.8429  0.399285
watertanke   -5.52987    7.58483 -0.7291  0.465959
protectewe   -5.44122    5.17858 -1.0507  0.293389
river        -1.76395    3.57625 -0.4932  0.621843
numberofro   12.47265    0.94757 13.1628 < 2.2e-16
roof         -2.83454    3.62586 -0.7818  0.434357
externalwa   10.19840    1.92089  5.3092 1.101e-07
floor         3.81212    2.55901  1.4897  0.136307

Rho: 0.3742, LR test value: 4.3541, p-value: 0.03692
Asymptotic standard error: 0.16111
    z-value: 2.3227, p-value: 0.020197
Wald statistic: 5.3948, p-value: 0.020197

Log likelihood: -1842.303 for lag model
ML residual variance (sigma squared): 273.25, (sigma: 16.53)
Number of observations: 436
Number of parameters estimated: 19
AIC: 3722.6, (AIC for lm: 3725)
LM test for residual autocorrelation
test value: 1.3748, p-value: 0.24099

> GM5_predict <- spatialreg::predict.sarlm(GM5, newdata = testdata, listw =
Q_test)
Error in spatialreg::predict.sarlm(GM5, newdata = testdata, listw = Q_test)
:
  mismatch between newdata and spatial weights. newdata should have
region.id as row.names
>

Any idea, what is going on?

	[[alternative HTML version deleted]]