# load libraries
    # devtools::install_github("tidyverse/ggplot2")
    library(ggplot2)
    library(ggchicklet)
    library(ggtext)
    library(dplyr)
    library(forcats)
    library(grid)

    # import data
    dat <- dat <- data.frame(
      Sport = c("NFL", "NFL", "NFL", "MLB", "MLB", "MLB", "NBA", "NBA",
                "NBA", "NHL", "NHL", "NHL", "EPL", "EPL", "EPL"),
      Type = c("Game Action", "Nonaction", "Commercials", "Game Action",
                "Nonaction", "Commercials", "Game Action", "Nonaction", "Commercials",
                "Game Action", "Nonaction", "Commercials", "Game Action", "Nonaction",
                "Commercials"),
      Time = c(18, 140.6, 49.9, 22.5, 150.9, 51.8, 49.6, 61.8,
               33.5, 63, 56.6, 37.4, 58.7, 47.8, 10.1),
      stringsAsFactors = FALSE)

    # refactor levels
    dat <- dat %>% group_by(Sport) %>% mutate(Percent = Time/sum(Time)) %>% ungroup() %>%
      mutate(Sport = fct_relevel(
        Sport,
        rev(c("NFL", "MLB", "NBA", "NHL", "EPL")))
      ) %>%
      mutate(Type = fct_relevel(
        Type,
        c("Commercials","Nonaction","Game Action"))
      )

    # keep trailing zeroes and add "min" to first bar value labels
    dat$Label <- as.character(sprintf("%.1f", dat$Time))
    dat$Label <- ifelse(dat$Type == "Game Action", paste0(dat$Label, " min"), dat$Label)

    # generate plot
    gg <- ggplot(dat, aes(Sport, Percent,  fill = Type, label = Label)) +
      geom_chicklet(width = 1,radius = unit(6,units = "pt"), position = ggplot2::position_stack(reverse = FALSE)) +
      geom_text(size = 4, fontface= "bold", position = position_stack(vjust = 0.5)) +
      scale_y_continuous(limits = c(0,1),expand = c(0, 0)) +  coord_flip() +
      theme_minimal() +
      theme(
            legend.position = "top",
            plot.title = element_markdown(hjust =0.5),
            plot.subtitle = element_markdown(hjust = 0.5),
            plot.caption = element_markdown(hjust = 0, size = 11, margin = unit(c(0, 0, 0, 0), "cm"), color = "#718c9e"),
            legend.text = element_markdown(size = 11),
            axis.text = element_text(face = "bold", size = 11),
            axis.text.x = element_blank(),
            axis.title.y = element_markdown(hjust = 0, size = 20, margin = unit(c(0, 0, 0, 0), "cm"), color = "#718c9e"),
            panel.grid = element_blank(),
            axis.title.x = element_markdown(
              halign = 0,
              margin = margin(2, 0, 15, 0),
              fill = "transparent"
            )

      ) +
      scale_fill_manual(
        name = NULL,
        values = c(`Game Action` = "#FA759F", Nonaction = "#B5BEC9", Commercials = "#72D4DB"),
        labels = c(
          # `Game Action` = "<strong style='color:#FA759F'>GAME ACTION</strong> (BALL OR PUCK IN PLAY)",
          # Nonaction = "<strong style='color:#B5BEC9'>NONACTION</strong> (GAME STOPPAGE, COMMENTARY, ETC.)",
          # Commercials = "<strong style='color:#72D4DB'>COMMERCIALS</strong>")
          `Game Action` = "<strong>GAME ACTION</strong> (BALL OR PUCK IN PLAY)",
          Nonaction = "<strong>NONACTION</strong> (GAME STOPPAGE, COMMENTARY, ETC.)",
          Commercials = "<strong>COMMERCIALS</strong>"),
        guide = guide_legend(reverse = TRUE)
      ) +
      labs(
        y = "<span style='font-size:13pt;'>The average share of broadcast time showing <strong style='color:#FA759F'>GAME ACTION</strong> is highest in<br>the English Premier League - but there is more total action in an average<br>National Hockey League game, which lasts longer.</span>",    x = NULL, fill = NULL,
        title = "<b>NFL and MLB games are long, slow affairs</b>",
        subtitle = "Minutes by broadcast by what is shown on screen across five major men's sports leagues",
        caption = "Games that we included: 10 NFL regular-season games between Nov. 7 amd Nov. 18, 2019. 17 MLB postseason games, including all the games in the 2019
        ACLS, NLCS, and World<br>Series; 10 NBA regular-season games between Nov. 6 and Nov. 15, 2019; 10 NHL regular-season games between Nov. 5 and Nov. 19, 2019, including three overtime games;
        and<br>seven English Premier League games between Nov. 9 and Nov. 23, 2019. NBA game action includes free throws, so the action time exceeds the game time.<br>
        <br>
        FiveThirtyEight SOURCE: UNIVERSITY OF TEXAS AT AUSTIN SPORTS ANALYTICS COURSE"
      )


    alignTitles <- function(ggplot, title = NULL, subtitle = NULL, caption = NULL) {
      # grab the saved ggplot2 object
      g <- ggplotGrob(ggplot)


      # find the object which provides the plot information for the title, subtitle, and caption
      if(!is.null(title)) {
        g$layout[which(g$layout$name == "title"),]$l <- title
      }
      if(!is.null(subtitle)) {
        g$layout[which(g$layout$name == "subtitle"),]$l <- subtitle
      }
      if(!is.null(caption)) {
        g$layout[which(g$layout$name == "caption"),]$l <- caption
      }
      g
    }

    # align caption to y axis value labels
    gg2 <- alignTitles(gg, caption = 2)
    grid.draw(gg2)

    # add arrow
    x <- rev(c(0.25, 0.25, 0.28, 0.28))
    y <- rev(c(0.2, 0.15, 0.15, 0.15))
    grid.bezier(x, y, gp=gpar(lwd=1.5, fill="black"),
                arrow=arrow(type="open",length = unit(0.1, "inches")))