FANCY A LITTLE BIT OF 3D?

… Wow…. That’s beautiful…. Don’t you think? Go ahead, try to move it with your mouse (hint: use left click)… I’ll wait here…

Are you back? Yep, beautiful, isn’t it?

Look back at it, grab your mouse, try to move it again, while right-clicking!

Impressive, isn’t it?

I almost cried when I look at it the first time.

The “persp()” function

This function draws perspective plot of a surface. Let’s take the example that is already included in R, because, for once, it’s pretty cool (and also because I’m running out of ideas for fake datasets)! First we need to create a set of x-y coordinates:

x <- seq(-10, 10, length= 30)   # Creating a vector of equally spaced
                                # values between -10 and 10
y <- x                          # Same vector for the y-coordinates
# Creating a function to combine x and y in a response variable z
# (we'll see in the next class how to create functions)
f <- function(x,y)
       {r<-sqrt(x^2+y^2); 10*sin(r)/r}  

z <- outer(x, y, f)     # Creating an array x*y where each cell
                        # value is the result of the previous function
z[is.na(z)] <- 1        # Taking care of NAs

Now for the fun part!

We now just need to plug those coordinates in the “persp()” function:

persp(x,y,z)

A little raw, but it’s already nice. First, let’s put some color. What about a nice light blue?   “col=”

persp(x, y, z, col = "lightblue")

Ok, better. The angle could be improved. What about looking at it from slighty higher:    “phi=”

persp(x, y, z, phi = 30, col = "lightblue")

Let’s also rotate it a little bit:     “theta=”

persp(x, y, z, theta = 30, phi = 30, col = "lightblue")

And to make it look ultra-cool, shading:    “shade=”

persp(x, y, z,
      theta = 30, phi = 30,
      col = "lightblue", shade = 0.75)

Eww… Too dark! Let’s move the light source up:     “ltheta=”

persp(x, y, z, 
      theta = 30, phi = 30,
      col = "lightblue", ltheta = 120, shade = 0.75)       

Side note: as you might have realized by now, the jumps are just there to make things easier to read. As long as the current instruction/function is not closed, R will wait for it, even if a jump is done.

Better. Let’s shrink the z coordinates representation. The argument “expand” permits to expand the z-scale by the specified ratio. A ratio under 1 will therefore shrink it.     “expand=“

persp(x, y, z, theta = 30, phi = 30, col = "lightblue",
      ltheta = 120, shade = 0.75,
      expand = 0.5 )

Much better! And finally, because why the hell not, axes’ labels!!!

persp(x, y, z, theta = 30, phi = 30, col = "lightblue",
      ltheta = 120, shade = 0.75,
      expand = 0.5,
      xlab = "X", ylab = "Y", zlab = "Sinc( r )" )

Phew, that was not as easy as we would have all probably wished, but, hey, the result is worth it!

Now, you know how to make beautiful plots! While it might look like a lot at first, once mastered, the possibilities are fantastic, and definitely worth it.

Ok, thanks, and see you next week!

What? Oh, I forgot to let you know how to save your beautiful graphs? Oops! Let’s quickly go over it then