In krlmlr/dplyr.benchmark: Run Time Benchmarks for the 'dplyr' Package

Historical dplyr benchmark results

r <- rprojroot::is_r_package$make_fix_file()

devtools::load_all()
library(dplyr)
library(ggplot2)
knitr::opts_chunk$set(echo = FALSE)
set.seed(20161031)

plot_data <- 
  get_plot_data() %>%
  compute_calibrated_time %>%
  filter_(~grepl("dplyr_df", name)) %>%
  detect_jumps

plot_data %>%
  tidyr::nest(-name) %>%
  purrrlyr::by_row(
    function(plot_data_slice) {
      data <- plot_data_slice$data[[1]]
      max_y <- min(
        tail(data$calibrated_time, 1) * 2,
        max(data$calibrated_time)
      )

      ggplot(data %>% filter(!is.na(kind)), aes(x = commit_id, y = calibrated_time)) +
        geom_line(data = data) +
        geom_point(aes(color = kind)) +
        ggrepel::geom_label_repel(aes(label = substr(sha, 1, 5), color = kind), alpha = 0.8) +
        scale_y_log10() +
        ggtitle(plot_data_slice$name)
    }) %>%
  .$.out

krlmlr/dplyr.benchmark documentation built on May 6, 2019, 6:06 p.m.