Digital Geography

9. July 2012

introduction to R: learning by doing (part 1)

Geography is often about statistics as it is the basis for fast exchange of information: providing a mean and standard deviation to the audience is often much easier then showing raw data: Learning a script language for this purpose can be a hard-ass work. But I think it is more often a need of practice. And by practice I mean typing, reading and trying out. has featured a little list of tutorials for MatLab. Due to my feelings about MatLab I would confront you with my own R introduction for statistical programming inspired by this one:
Type the code below into your R console to get some insight into the R syntax. If you need to install R beforehand follow this instruction

a <- 7 #a will be treated as a scalar and will get the value of 7
b <- c(1,2,3) #b is now a &amp;quot;c&amp;quot;olumn vector with three entries
e <- c(1, a, 3) #whohoo look where the &amp;quot;a&amp;quot; is coming from
f <- r(2,3,4) #nope, there are no row-vectors in R as far as I know
f <- t(c(1, a, 3)) #this could be the problem solver but this is not the same as it is a matrix...
e #will show you all entries in e
e[2]#will give you the second element from e 
ls() #will show you all available variables
rm(b) # let us &amp;quot;r&amp;quot;e&amp;quot;m&amp;quot;ove the be from our workspace

As we have already created a very easy matrix lets go a little further with matrices:

A <- diag(3) #will show you the identity matrix with ones on the diagonal
A <- A * a #lets guess ...
A[3,2] <- pi # will set the third row and second column with the value of pi
X=matrix(1, nrow=3, ncol=10) # a mtrix with one as a the only value for 30 cells.
E <- diag(e) #lets create a diagonal matrix of e
#transposition of a matrix
E = t(E); #with the semi-colon we are suppressing the output of the operation; and the "=" seves as "<-"

Lets get back to vectors as they are important for statistical calculations: normally a vector represents a list observations of one parameter only:

#let us do some vector creation:
for (i in c(1:10)) {
v[i] &amp;lt;- i*2
} #will give you a  vetcor with ten numbers ranging from 2 to 20 in steps of two
v2 <- seq(2,20,2) #will do the same sequence 😉

Now: operating on matrices with R. keep in mind to think column wise as the splitting of the vector is done by columns first.

rm(list=ls()) #will remove everything from your workspace
A <- matrix(c(1,2,2,1,3,0), nrow=2, ncol=3)
B <- matrix(c(2,1,0,2,0,1), nrow=3, ncol=2)
C <- matrix(c(0,5,1,1,0,3), nrow=2, ncol=3)
dim(A) #will provide info about the numbers of rows and colums
dim_A <- dim(A) #is fine as well
A%*%B #multiplication
A%*%C #failure due to &amp;quot;misshape&amp;quot; matrices
A%*%t(C) #multiplication with the transpose
diag(A%*%t(C)) #will give you the values along the diagonal of the resulting matrix
D <- A+C #Matrix addition
c(2,1,3) -> x # "<-" the other way around... still "<-" is not the same as "=" as "=" is without a "direction"
g <- A*x

Try it out…