Plot 3D plane (true regression surface)

Question

I\'m trying to simulate some data (x1 and x2 - my explanatory variables), calculate y using a specified function + random noise and plot the resulting observations AND the t

Answer 1

IRTFM's somewhat imperfect answers above let me to a thread on the CRAN help pages.
https://stat.ethz.ch/pipermail/r-help/2013-December/364037.html
I extracted the relevant bits of code and turned them into a function like so:

require(rgl)
pred.surf.3d <- function(df, x.nm,y.nm,z.nm, ...){
  x <- df[,x.nm]; y <- df[,y.nm]; z<-df[,z.nm]
  fit <- lm(z ~ x + y + x*y + x^2 + y^2)
  xnew <- seq(range(x)[1],range(x)[2],len=20)
  ynew <- seq(range(y)[1],range(y)[2],len=20)
  df <- expand.grid(x=xnew, y=ynew)
  df$z <- predict(fit, newdata=df)
  with(df, surface3d(xnew, ynew, z=df$z))
}

I may end up bundling this into my CRAN utility package at some point.
In the mean time, I hope you find it useful! (Run it on IRTFM's first code chunk like so:)

pred.surf.3d(data.frame(x1,x2,y),'x1','x2','y')