[R] plotting confidence intervals of regression line

[Eik Vettorazzi]

you can use the newdata statement in the predict function, as in

n<-50
x<-sample(40:70,n,rep=T)
y<-.7*x+rnorm(n,sd=5)
plot(x,y,xlim=c(20,90),ylim=c(0,80))
mylm<-lm(y~x)
abline(mylm,col="red")
newx<-seq(20,90)
prd<-predict(mylm,newdata=data.frame(x=newx),interval = c("confidence"), 
level = 0.90,type="response")
lines(newx,prd[,2],col="red",lty=2)
lines(newx,prd[,3],col="red",lty=2)

hth.


Tim Holland schrieb:
> Hello,
>
> I am trying to generate a confidence interval (90 or 95%) of a regression
> line.  This is primarily just for illustration on a scatter plot (i.e. I am
> trying to make this
> http://www.ast.cam.ac.uk/~rgm/scratch/statsbook/graphics/anima4.gif).
>
> I have been trying to use the predict.lm function, with interval set as
> "confidence", but this still seems to be giving me a prediction interval (as
> below).  It is giving the interval based on the x value of a given point, as
> opposed to the confidence interval of the regression line in general.  Is
> predict.lm the correct function to be using, or is there something else more
> appropriate?
>
>   
>> predict.lm(test, interval = c("confidence"), level = 0.90)
>>     
>         fit        lwr       upr
> 1  4.170807 -0.2171226  8.558738
> 2  5.807453  2.1283100  9.486597
> 3  6.625776  3.2397994 10.011753
> 4  8.262422  5.2795295 11.245315
> 5  9.080745  6.1821741 11.979317
> 6  9.899068  6.9954692 12.802667
> 7  7.444099  4.2947040 10.593495
> 8 13.172360  9.4577068 16.887014
> 9 20.537267 13.0143098 28.060224
>
> I have also tried using package 'ggplot' as follows:
>
>   
>> install.packages("ggplot")
>> library(ggplot)
>> qplot(wt, mpg, data=mtcars, type=c("point","smooth"), method=lm)
>>     
> (following Hadley Wickham on this help list)
>
> This generates a plot that looks roughly like what I want, but there is some
> customizing of the plot that I would prefer to do in the standard 'plot'
> function of the 'graphics' package. Also, ggplot does not give me a vector
> of the values of the CI themselves, which would be useful to have.
>
> Any help would be very appreciated.
>
> Thanks,
> Tim
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>   

-- 
Eik Vettorazzi
Institut für Medizinische Biometrie und Epidemiologie
Universitätsklinikum Hamburg-Eppendorf

Martinistr. 52
20246 Hamburg

T ++49/40/42803-8243
F ++49/40/42803-7790