Profiling the execution time

So far, we have seen how to measure the execution time of a whole R expression. What about a more complex expression with multiple parts such as calls to other functions? Is there a way to dig deeper and profile the execution time of each of the parts that make up the expression? R comes with the profiling tool Rprof() that allows us to do just that. Let's see how it works.

Profiling a function with Rprof()

In this example, we write the following sampvar() function to calculate the unbiased sample variance of a numeric vector. This is obviously not the best way to write this function (in fact R provides the var() function to do this), but it serves to illustrate how code profiling works:

# Compute sample variance of numeric vector x
sampvar <- function(x) {
    # Compute sum of vector x
    my.sum <- function(x) {
        sum <- 0
        for (i in x) {
            sum <- sum + i
        }
        sum
    }
    
    # Compute sum of squared variances...