Benchmark of Different Mean Functions

Function in Rcpp

#include <Rcpp.h>
//[[Rcpp::export]]
double muRcpp(Rcpp::NumericVector x){
  int n = x.size(); // Size of vector
  double sum = 0; // Sum value
  // For loop, note cpp index shift to 0
  for(int i = 0; i < n; i++){
    // Shorthand for sum = sum + x[i]
    sum += x[i];
  }
  return sum/n; // Obtain and return the Mean
}

Function in R Vectorized

muVec <- function(x){
  x <- tibble::enframe(x)
  a <- x %>% mutate(Aux = value/nrow(x)) %>% summarise(Mu = sum(Aux))
  a$Mu
}

Base Function of R

# Example
x <- 0:10
mean(x)

## [1] 5

Conclusion

x <- 0:10

base <- microbenchmark(Rcpp = muRcpp(x),
                       R_Vec = muVec(x),
                       R_base = mean(x),
                       times = 100)

base %>% ggplot() + 
  geom_boxplot(aes(x = reorder(expr, time, median), y = time, color = factor(expr)))

Rcpp really is faster than base R, however it always seem to have that initial run that it is slower than R base. I think that this is the time difference it takes to call the package and/or the function. Vectorized it is indeed slower, for it deals with much more memory, however we get information that we wouldn’t otherwise, as to what is happening inside the function.