dplyr-timing-test.R

library(dplyr)
library(microbenchmark)

set.seed(42)
n <- 10000
df <- data.frame(a = rep(seq(n), 2), 
                 x = rnorm(2 * n))

group_summarize_mutate <- function(df)
{
    df %>%
        group_by(a) %>%
        summarize(x = mean(x)) %>%
        mutate(x = x / 100)
}

group_summarize <- function(df)
{
    df %>%
        group_by(a) %>%
        summarize(x = mean(x) / 100)
}

mutate_group_summarize <- function(df)
{
    df %>%
        mutate(x = x / 100) %>%
        group_by(a) %>%
        summarize(x = mean(x))
}

microbenchmark(group_summarize_mutate(df), 
               group_summarize(df), 
               mutate_group_summarize(df), 
               times = 20)