R/bars.R

In pkollenda/pkgraph: Fancy Graphs with GGPlot

#' Create Bar Charts
#'
#' Create a bar chart based on ggplot.
#'
#' Use ggplot's geom_col geom to create bar charts that are visually appealing and in line with a common
#' theme and visualization style.
#'
#' @param data A data frame.
#' @param group The groups that are displayed as separate bars.
#' @param count The variable that contains the number of observations per group.
#' @param highlight_variable Default NA, indicate variable that is TRUE only for the highlighted group.
#' @param ordered Default alphabetically, alternative is count, then ordered in descending order.
#' @param first_group Default NA, indicate a group which should be displayed first (left or top).
#' @param flipped Default TRUE. If flipped is TRUE bars are horizontally (groups on y axis), otherwise bars are vertically (groups on x axis).
#'
#' @author Philipp Kollenda
#'
#' @import tidyverse
#' @importFrom cowplot theme_minimal_vgrid
#' @importFrom scales wrap_format
#'
#' @return A ggplot style bar graph.
#' @export bars
#'
#' @examples
#' \dontrun{
#'   bars(data = sdg %>% count(donor) %>% filter(n > 30000) %>% mutate(highlight = donor == "Germany"), count = n, group = donor, highlighted_var = highlight, ordered = "count", first_group = "Germany")
#'   }
bars <- function(
  data,
  count, group, highlight_variable = NA,
  ordered = "alphabetically", first_group = NA, flipped = TRUE, labelled = FALSE, show_percent = NA,
  label_nudge_y = -0.02*max(select(data, {{count}})), label_nudge_x = 0, label_size = 3, label_color = "white", label_hjust = "right",
  axis.title.x = element_blank(), axis.title.y = element_blank(),
  title = element_blank(),
  legend.position = "none",
  fontsize = 16) {

  # 0) Prepare data
  # If flipped, order for alphabetical needs to be reversed, such that a is on top.
  if(ordered == "alphabetically" & flipped == T){data <- data %>% mutate({{group}} := fct_rev({{group}}))}
  # . Order by frequency?
  if(ordered == "count") {data <- data %>% arrange({{count}}) %>% mutate({{group}} := fct_inorder(fct_drop({{group}})))}
  # . Order one in front, rest either frequency (see above) or alphabetically
  if(!missing(first_group) & flipped == F) {data <- data %>% mutate({{group}} := fct_relevel({{group}}, first_group))} else
    if(!missing(first_group) & flipped == T) {data <- data %>% mutate({{group}} := fct_relevel({{group}}, rev(first_group), after = Inf))}


  # 1) Start by specifying the mappings that should hold for all geoms.
  p <- ggplot(data = data, aes(x = {{group}}, y = {{count}}))

  # 2) Add all the relevant geoms that I want to have.
  # . Add geom_col for column, bar chart
  p <- p + geom_col()

  # . Add geom_text to label the bars with the number of observations, if desired.
  if(labelled){ p <- p + geom_text(aes(label = {{count}}, y = {{count}}), nudge_y = label_nudge_y, nudge_x = label_nudge_x,
                                   size = label_size, hjust = label_hjust, color = label_color)}

  # 3) Add highlighted variable if desired.
  if(!missing(highlight_variable)){
    p <- p + aes(fill = {{highlight_variable}}) + scale_fill_manual(name = NULL, values = c("#B0B0B0D0", "#D55E00D0"))
  }

  # 4) Flip the coordinate system, almost always desired for column bar plots.
  if(flipped){ p <- p + coord_flip()
  # # With flipped coordinate system, alphatically ordered factors have to be changed
  # p <- p + scale_x_discrete(limits = rev(levels(data$group)))
  # Rename x and y axis for better intuitive handling.
  temp = axis.title.x; axis.title.x = axis.title.y; axis.title.y = temp; rm(temp);
  }

  # 5) Adjust (if desired) the scales.
  p <- p + scale_x_discrete(labels = scales::wrap_format(width = 16))
  p <- p + scale_y_continuous(labels = scales::comma)

  # 6) Add all the labels.
  p <- p + labs(
    title = title, x = axis.title.x, y = axis.title.y)


  # 7) Set and adjust the main theme.
  p <- p + cowplot::theme_minimal_vgrid(font_size = fontsize)
  p <- p + theme(legend.position = legend.position, axis.title = element_text(face = "italic"),
                 plot.title = element_text(face = "bold", size = fontsize * 1.15))
  if(labelled && flipped){p <- p + theme(panel.grid = element_blank(), axis.text.x = element_blank())}

  return(p)
}

# bars(data = sdg %>% count(donor) %>% filter(n > 30000) %>% mutate(highlight = donor == "Germany"), count = n, group = donor, ordered = "count", first_group = "Germany", flipped = T)
# bars(data = emperors %>% count(cause) %>% mutate(highlight = cause == "Suicide"), first_group = c("Unknown", "Suicide"), ordered = "count", axis.title.x = "Hi", flipped = T, highlight_variable = highlight, count = n, group = cause)