Getting Started with R

Once you've installed R and and a shortcut on your Windows Start menu (see the previous page on this subject), click that shortcut to start the software. You should see an initial screen looking like the following:

You should take this opportunity to look around, kick the tires by looking through the menus at the top of the window and resting the mouse pointer over the control buttons to see the pop-up explanations.

Note also that the manuals are available from within the workspace by choosing the Manuals option under the Help menu (for the PDF versions --- use either GSview in the Ghostview/Ghostscript section of this web site or the free Adobe Acrobat reader available on the web).

This is also a good time to try out the demos included with the distribution. For example, type demo() for a list of what's available:

Graphics are usually pretty cool (notice how many "screenshots" pages there are for free software out there), let's try the "graphic" demo first.

For example, the R website has an interesting screenshot. To see it , click here.

Type "demos(graphics)" at the prompt in the console window:

and then press Return to see a blank graphics window open and to see the following text appear in the console:

You'll need to hit Return again to see the plot, but make sure you've given the focus back to the console window by physically clicking on it. Otherwise R has given the focus to the graphics window and won't know what you're trying to do. The result should look like the following:

Keep hitting Return to see more examples of the graphical capabilities of R. You'll have to keep placing the focus back on the text console window with your house, however. Now let's try to exercise the software a bit. Suppose we want to generate a set x of 5000 random numbers which have a Gaussian distribution with mean m = 0 and standard deviation s = 1. The command for doing this is

x <- rnorm(5000, mean=0, sd=0)

We can get more information on the rnorm() command by typing help(rnorm) at the prompt in the console window. The result is a new "information" window which looks like this:

So let's go ahead and do this:

The "<-" operator is the assignment operator. We are taking the result of the rnorm(5000, mean=0, sd=1) expression on the right-hand side and assigning that particular realization to variable x. If you want to see the contents (all 5000 numbers), simply type x at the prompt and hit return. We can plot the set of numbers by typing plot(x) at the prompt. The result is a new graphics window looking like this:

We can see by inspection that the mean m of this realization is probably zero, although it's a little tougher to see that the standard deviation s is unity. However, we can check this using R commands mean() and sd():

Now suppose we scale the set of data points by multiplying each by, say, 5. We can do so, and then check that the mean of the realization remains close to zero, but that the standard deviation increases from 1 to 5:

Another quick glance at the plot, using plot(y), reveals

Notice the scale on the y-axis now: the data has spread considerably. Now let's try biasing the data. Let's add the value, say 10, to each of the points in the realization:

This web site copyright © 1999-2000 by Robert A. Luke.  All rights reserved.