[R] trying to fit a thin plate spline to a triangular plot

[David Winsemius]

On Jun 16, 2013, at 7:39 AM, Rapkin, James wrote:

> Hi
> 
> Has anyone tried to fit a thin plate spine to a triangular plot before? I'm having trouble working out a piece of code to make it possible with three axes.

Only two axes in a triangular plot are independent.

> I have done it before with two axes. 
> 
> I've found a package I prefer that makes triangular plots (plotrix) using the triax.plot() function. An example of some data and the plot follows:
> 
> graph.data<-data.frame(a=c(9,6.2, 5.9,8.1,8.9,56.9,50,39.5,62.1,49.9,4.2,3,6,4.1),
>                       b=c(1.8,1.2,1.2,1.6,1.8,11.4,10,7.9,12.4,10,4.2,3,6,4.1),
>                       c=c(1.8,1.2,1.2,1.6,1.8,11.4,10,7.9,12.4,10,4.2,3,6,4.1),
>                       d=c(10,15,16,14,13,18,19,17,20,19,21,22,22,23))
> 
> triax.plot(graph.data, main="Title", no.add=F, show.grid=T,tick.labels=list(l=seq(10,90,by=10),r=seq(10,90,by=10),b=seq(10,90,by=10)),
>           pch=1,cc.axes=TRUE)
> 
> I want to fit the 'd' data as a spline to the plot. 

Looking at the data I am unable to imagine how that could meaningfully support a 2d spline fit.

> 
> Any help is greatly appreciated. 

This was cross-posted to StackOverflow and R-help. SO does not have a policy against cross-posting, but Rhelp does.

-- 
David Winsemius
Alameda, CA, USA