Ok, so just for anyone's interest, I managed to create the calibration plot for
the glmnet object using the val.prob() function from the rms package.
Now, my question moves slightly, how can I superimpose calibration curves from
two models, so that they can be graphically compared?
This is what I have tried. I start with two models, based on same predictors:
rms_fit and enet_fit
>validate(rms_fit,B=100)
>cal<-calibrate(rms_fit,B=100)
> mypred<-predict(enet_fit,type="response",s=lambda.min,newx=myDesignMatrix[1:703,])
> #get probabilities for training set
> val.prob(mypred, as.numeric(out.v), m=20, cex=.5) #out.v is a vector of
> outcomes for each sample
>par(new=TRUE)
>plot(cal)
would superimpose both graphs completely, including labels and axis.
Also tried doing:
> val.prob(mypred, as.numeric(out.v), m=20, cex=.5) #out.v is a vector of
> outcomes for each sample
>lines(plot(cal))
but this creates a new graph window for cal object.
Also, how can I colour them differently..
Thanks in advance,
Dave
[[alternative HTML version deleted]]
______________________________________________
R-help@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.
