Intermediate R for reproducible scientific analysis

Apply functions

Learning objectives

  • To learn how to use apply to automate tasks efficiently
  • To know the difference between apply, lapply, sapply, tapply and mapply.

Vectorized task automation

Previously we introduced you to for loops: a basic programming construct that is common across many programming languages. R has more optimised way of automating tasks that are not only faster than for loops, but also take away the pain of having to pre-define your results object.

The most common function you will encounter is lapply, and the closely related sapply.

Lets take a look at the following for loop:

for (cc in gap[,unique(continent)]) {
  popsum <- gap[year == 2007 & continent == cc, sum(pop)]
  print(paste(cc, ":", popsum))
}

It calculates the total population on each continent, then prints out the result. If instead we want to save these results, we can either make a vector in advance and save the results, or use one of the apply to take care of this detail for us:

results <- lapply(gap[,unique(continent)], function(cc) {
  popsum <- gap[year == 2007 & continent == cc, sum(pop)]
  popsum
})
names(results) <- gap[,unique(continent)]
results
$Asia
[1] 3811953827

$Europe
[1] 586098529

$Africa
[1] 929539692

$Americas
[1] 898871184

$Oceania
[1] 24549947

lapply takes a vector (or list) as its first argument (in this case a vector of the continent names), then a function as its second argument. This function is then executed on every element in the first argument. This is very similar to a for loop: first, cc stores the first continent name, “Asia”, then runs the code in the function body, then cc stores the second continent name, and runs the function body, and so on. The code in the function body can be thought of in exactly the same way as the body of the for loop. The result of the last line is then returned to lapply, which combines the results into a list.

sapply is identical to lapply, except that it tries to simplify the results object. If we run the same code with sapply instead of lapply the results will be returned as a vector:

results <- sapply(gap[,unique(continent)], function(cc) {
  popsum <- gap[year == 2007 & continent == cc, sum(pop)]
  popsum
})
names(results) <- gap[,unique(continent)]
results
      Asia     Europe     Africa   Americas    Oceania 
3811953827  586098529  929539692  898871184   24549947 

apply

The apply function is useful for matrix data: it allows you loop over either the rows or columns of a matrix.

# create some dummy data
r <- matrix(rnorm(10*4), nrow=10)
colnames(r) <- letters[1:4]
rownames(r) <- LETTERS[1:10]
# Get the maximum value in each row:
apply(r, 1, max)
        A         B         C         D         E         F         G 
0.9621735 1.1248378 0.9209498 1.4553950 0.8625534 0.9564831 0.4842614 
        H         I         J 
0.9176132 1.0213789 0.8276343 
# and for each column:
apply(r, 2, max)
        a         b         c         d 
1.0213789 1.4553950 1.1248378 0.9209498 

mapply

The mapply function can be used to run a function with different combinations of arguments. Let’s take a look at an example:

a <- 1:4
b <- 4:1
mapply(rep, a, b)
[[1]]
[1] 1 1 1 1

[[2]]
[1] 2 2 2

[[3]]
[1] 3 3

[[4]]
[1] 4

This is the same as running the following code:

rep(a[1], b[1])
[1] 1 1 1 1
rep(a[2], b[2])
[1] 2 2 2
rep(a[3], b[3])
[1] 3 3
rep(a[4], b[4])
[1] 4

or the following lapply statement:

lapply(1:4, function(ii) {
  rep(a[ii], b[ii])
})
[[1]]
[1] 1 1 1 1

[[2]]
[1] 2 2 2

[[3]]
[1] 3 3

[[4]]
[1] 4

tapply

The tapply function allows you to run a function on different groups within a vector. Going back to our first example of the lesson, we can use tapply to calculate the total population for each continent in 2007:

gap2007 <- gap[year == 2007] # first filter by the year
tapply(
  gap2007[,pop], # The column of population counts from the data frame
  gap2007[,continent], # The column of continent labels for each entry
  sum # The function to apply to the population vector within each continent
)
    Africa   Americas       Asia     Europe    Oceania 
 929539692  898871184 3811953827  586098529   24549947