MGCV: overlay fitted (marginal) curves over a plot of the original data

Dear R users,

I?ve stumbled over a problem that can be easily seen from the R code below:

- When I use plot.gam() on a fitted model object, I get a nice and well-looking smooth curve for all
terms in the model.

- However, when I use predict(model) for a given predictor, with values of all other predictors set
to their means, the resulting curve doesn?t fit well at all.

Is there a way to "overlay" the curve produced by plot.gam() over a plot of the original data?

Here?s some reproducible code with mgcv version 1.7-22 on R3.0.1 (Windows 7):

##

library(mgcv)
set.seed(2)
dat <- gamSim(1,n=400,dist="normal",scale=2)
b <- gam(y~s(x0)+s(x1)+s(x2)+s(x3),data=dat)
plot(b,select=1)

plot(y~x0,dat)
mydata=data.frame(x0=0:1,x1=mean(dat$x1),x2=mean(dat$x2),x3=mean(dat$x3))
lines(0:1,predict(b,mydata,type="response"))

##

Best wishes,
Christoph

Probably you didn't want to set x0=0:1? Here is some code, to do what you want.
(The CI shape is not identical to the plot(b) version as the uncertainty includes
the uncertainty in the other smooths and the intercept now.)

library(mgcv)
set.seed(2)
dat <- gamSim(1,n=400,dist="normal",scale=2)
b <- gam(y~s(x0)+s(x1)+s(x2)+s(x3),data=dat)
plot(b,select=1)

plot(y~x0,dat)
mydata=data.frame(x0=0:200/200,x1=mean(dat$x1),x2=mean(dat$x2),x3=mean(dat$x3))
pv <- predict(b,mydata,type="response",se=TRUE)
lines(mydata$x0,pv$fit)
lines(mydata$x0,pv$fit+2*pv$se.fit,lty=2)
lines(mydata$x0,pv$fit-2*pv$se.fit,lty=2)


  

On 16/07/13 09:52, Christoph Scherber wrote:

Dear R users,

I?ve stumbled over a problem that can be easily seen from the R code below:

- When I use plot.gam() on a fitted model object, I get a nice and well-looking smooth curve for all
terms in the model.

- However, when I use predict(model) for a given predictor, with values of all other predictors set
to their means, the resulting curve doesn?t fit well at all.

Is there a way to "overlay" the curve produced by plot.gam() over a plot of the original data?

Here?s some reproducible code with mgcv version 1.7-22 on R3.0.1 (Windows 7):

##

library(mgcv)
set.seed(2)
dat <- gamSim(1,n=400,dist="normal",scale=2)
b <- gam(y~s(x0)+s(x1)+s(x2)+s(x3),data=dat)
plot(b,select=1)

plot(y~x0,dat)
mydata=data.frame(x0=0:1,x1=mean(dat$x1),x2=mean(dat$x2),x3=mean(dat$x3))
lines(0:1,predict(b,mydata,type="response"))

##

Best wishes,
Christoph

Probably you didn't want to set x0=0:1? Here is some code, to do what you want.
(The CI shape is not identical to the plot(b) version as the uncertainty includes
the uncertainty in the other smooths and the intercept now.)

library(mgcv)
set.seed(2)
dat <- gamSim(1,n=400,dist="normal",scale=2)
b <- gam(y~s(x0)+s(x1)+s(x2)+s(x3),data=dat)
plot(b,select=1)

plot(y~x0,dat)
mydata=data.frame(x0=0:200/200,x1=mean(dat$x1),x2=mean(dat$x2),x3=mean(dat$x3))
pv <- predict(b,mydata,type="response",se=TRUE)
lines(mydata$x0,pv$fit)
lines(mydata$x0,pv$fit+2*pv$se.fit,lty=2)
lines(mydata$x0,pv$fit-2*pv$se.fit,lty=2)


  

On 16/07/13 09:52, Christoph Scherber wrote:

Dear R users,

I?ve stumbled over a problem that can be easily seen from the R code below:

- When I use plot.gam() on a fitted model object, I get a nice and well-looking smooth curve for all
terms in the model.

- However, when I use predict(model) for a given predictor, with values of all other predictors set
to their means, the resulting curve doesn?t fit well at all.

Is there a way to "overlay" the curve produced by plot.gam() over a plot of the original data?

Here?s some reproducible code with mgcv version 1.7-22 on R3.0.1 (Windows 7):

##

library(mgcv)
set.seed(2)
dat <- gamSim(1,n=400,dist="normal",scale=2)
b <- gam(y~s(x0)+s(x1)+s(x2)+s(x3),data=dat)
plot(b,select=1)

plot(y~x0,dat)
mydata=data.frame(x0=0:1,x1=mean(dat$x1),x2=mean(dat$x2),x3=mean(dat$x3))
lines(0:1,predict(b,mydata,type="response"))

##

Best wishes,
Christoph