Dear jianghua and all R professor:
I use the suggestion that jianghua gives. But the problem is still there.
Use "summary(m1)" or "summary.gam(m1)" is the same result and it doesn't show
the significance results of parametric terms.
Should I choose another smoother to get significance results of parametric
terms ?
I know another "gam way" in the package "mgcv" and they can show me the
significance results of parametric terms in package mgcv.
But the smoother is different between in package "gam" and "mgcv"
Is it possible to find significance results of parametric terms in package
"gam" ?
or maybe I have to try another smoother in package "mgcv" !!
Is it possible that in package "mgcv" do some assunptions which can make the
result is the same in package "gam" ?
Ashely, Yang
in package gam
¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ
> library(gam)
Loading required package: splines
>
>
> m1=gam(y~ost+wst+park10+sch50+comm+build+suite+y95+y96+y97+y98+y99+s(builarea)+s(age)+s(fl)+s(totfl)+s(cbd)+s(redl))
>
> summary.gam(m1)
Call: gam(formula = y ~ ost + wst + park10 + sch50 + comm + build +
suite + y95 + y96 + y97 + y98 + y99 + s(builarea) + s(age) +
s(fl) + s(totfl) + s(cbd) + s(redl))
Deviance Residuals:
Min 1Q Median 3Q Max
-753.51 -118.98 -15.27 99.16 1253.28
(Dispersion Parameter for gaussian family taken to be 41952.11)
Null Deviance: 564111793 on 4714 degrees of freedom
Residual Deviance: 196251969 on 4678 degrees of freedom
AIC: 63607.24
Number of Local Scoring Iterations: 2
DF for Terms and F-values for Nonparametric Effects
Df Npar Df Npar F Pr(F)
(Intercept) 1
ost 1
wst 1
park10 1
sch501
comm 1
build1
suite1
y95 1
y96 1
y97 1
y98 1
y99 1
s(builarea) 1 3 13.852 5.440e-09 ***
s(age) 1 3 13.410 1.033e-08 ***
s(fl)1 3 41.732 < 2.2e-16 ***
s(totfl) 1 3 19.146 2.454e-12 ***
s(cbd) 1 3 15.464 5.222e-10 ***
s(redl) 1 3 4.839 0.002300 **
---
Signif. codes: 0 ¡¥***¡¦ 0.001 ¡¥**¡¦ 0.01 ¡¥*¡¦ 0.05 ¡¥.¡¦ 0.1 ¡¥ ¡¦ 1
>
> summary(m1)
Call: gam(formula = y ~ ost + wst + park10 + sch50 + comm + build +
suite + y95 + y96 + y97 + y98 + y99 + s(builarea) + s(age) +
s(fl) + s(totfl) + s(cbd) + s(redl))
Deviance Residuals:
Min 1Q Median 3Q Max
-753.51 -118.98 -15.27 99.16 1253.28
(Dispersion Parameter for gaussian family taken to be 41952.11)
Null Deviance: 564111793 on 4714 degrees of freedom
Residual Deviance: 196251969 on 4678 degrees of freedom
AIC: 63607.24
Number of Local Scoring Iterations: 2
DF for Terms and F-values for Nonparametric Effects
Df Npar Df Npar F Pr(F)
(Intercept) 1
ost 1
wst 1
park10 1
sch501
comm 1
build1
suite1
y95 1
y96 1
y97 1
y98 1
y99 1
s(builarea) 1 3 13.852 5.440e-09 ***
s(age) 1 3 13.410 1.033e-08 ***
s(fl)1 3 41.732 < 2.2e-16 ***
s(totfl) 1 3 19.146 2.454e-12 ***
s(cbd) 1 3 15.464 5.222e-10 ***
s(redl) 1 3 4.839 0.002300 **
---
Signif. codes: 0 ¡¥***¡¦ 0.001 ¡¥**¡¦ 0.01 ¡¥*¡¦ 0.05 ¡¥.¡¦ 0.1 ¡¥ ¡¦ 1
>
in package mgcv
¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ¡õ
> library(mgcv)
This is mgcv 1.3-29
Attaching package: 'mgcv'
The following object(s) are masked from package:gam :
anova.gam,
gam,
gam.control,
gam.fit,
plot.gam,
predict.gam,
print.gam,
print.summary.gam,
s,
summary.gam
>
> m1=gam(y~ost+wst+park10+sch50+comm+build+suite+y95+y96+y97+y98+y99+s(builarea)+s(age)+s(fl)+s(totfl)+s(cbd)+s(redl))
>
> summary(m1)
Family: gaussian
Link function: identity
Formula:
y ~ ost + wst + park10 + sch50 + comm + build + suite + y95 +
y96 + y97 + y98 + y99 + s(builarea) + s(age) + s(fl) + s(totfl) +
s(cbd) + s(redl)
Parametric coefficients:
Estimate S