Plot a 3D linear surface from 2D functions in R

Question

I have the following data:

x <- c(1000,2000,3000,4000,5000,6000,8000,12000)

y_80 <- c(33276,33276,5913,2921,1052,411,219,146)
y_60 <- c(14724,14724,3755,1958,852,372,211,140)
y_40 <- c(9632,9632,2315, 1250,690,332,196,127)
y_20 <- c(4672,4672,1051,562,387,213,129,81)
y_5  <- c(825,825,210,118,88,44,27,17)

From this data I create 5 spline functions:

f_80 <- splinefun(x, y_80, method=c("monoH.FC"))
f_60 <- splinefun(x, y_60, method=c("monoH.FC"))
f_40 <- splinefun(x, y_40, method=c("monoH.FC"))
f_20 <- splinefun(x, y_20, method=c("monoH.FC"))
f_5 <- splinefun(x, y_5, method=c("monoH.FC"))

The Z axle is the number after the f_ i.e. for function f_80 the Z value is 80 and so on.

What I needed to do is to plot a 3D surface from those functions, with linear interpolation between the lines. Is that possible in R? I have already SO similar questions but I can't find an answer. Thanks

Answer 1

There are a variety of R pseudo-3d plotting functions. base-graphics has persp, and wireframe is in Lattice, and the coolest of all, surface3d is in rgl. They differ in how they handle the axis arguments but they all generally accept a matrix argument for the z-values. `persp is simplest:

 z_80 <- f_80(x)
 z_60 <- f_60(x)
 z_40 <- f_40(x)
 z_20 <- f_20(x)
 z_5  <- f_5(x) 

 png(); persp(matrix(c(z_80, z_60, z_40, z_20, z_5 ), nrow=length(x), 
              dimnames= list(X=x, Y=c(80,60,40,20,5) ) ) ) 
 dev.off()

You might find that using ticktype="detailed" is more informative:

 png(); persp(matrix(c(z_80,z_60,z_40,z_20,z_5 ), nrow=length(x), 
                      dimnames=list(X=x, Y=c(80,60,40,20,5) ) ) , 
                      ticktype="detailed", xlab="X axis", ylab="Y axis", 
                      zlab=".                Z= f(y)", theta=45) ; dev.off()

I generally need to play with viewing angle theta to get something I like. Getting the z-axis label to move away from the long z-tick-labels was a hack.