Is there a way to create borders around rectangles in ggplot2 with geom_raster?

Question

I'm creating a visualization of missing data by slightly tweaking some of the code from the missmap function in the Amelia package. I want to draw borders around my rectangles, but I can't figure out a way to do that in ggplot2.

I found the function "borders()" but that appears to be related to map work. I also tried using geom_rect, but it seems like that would require me to specify mins and maxes. Geom_raster seems to be doing exactly what I need, but I can't figure out how to specify borders.

This example code creates the visualization that I'm imagining, but I have more variables in the "real" version and I'd like to be able to outline each variable (var1, var2, etc.) with a line (border).

#Dataset
missmap_data_test <- data.frame(var1 = c(11, 26, NA, NA, 15),
                                var2 = c(NA, NA, 0, NA, 1))

#Create Function
ggplot_missing <- 

function(x){
  x %>% 
    is.na %>%
    melt %>%
    ggplot(data = .,
           aes(x = Var2,
               y = Var1)) +
    geom_raster(aes(fill = value)) +
    scale_fill_grey(name = "",
                    labels = c("Present","Missing")) +
    theme_minimal() + 
    theme(axis.text.x  = element_text(angle=90, hjust=1)) + 
    labs(x = "Variables in Dataset",
         y = "Observations")
}

#Feed the function my new data
ggplot_missing(missmap_data_test)

Use `geom_tile` instead. `geom_raster` is faster, but more limited. — Axeman, May 30 '19 at 14:28

score 8 · Accepted Answer · edited Jun 20 '20 at 09:12

As @Axeman suggests, geom_tile does the job. I've updated your code to give an example below. Here, colour defines the colour of the border, while size define the thickness.

#Dataset
missmap_data_test <- data.frame(var1 = c(11, 26, NA, NA, 15),
                                var2 = c(NA, NA, 0, NA, 1))

# Load libraries
library(dplyr)
library(ggplot2)
library(reshape2)

#Create Function
ggplot_missing <- function(x){
    x %>% 
      is.na %>%
      melt %>%
      ggplot(data = .,
             aes(x = Var2,
                 y = Var1)) +
      geom_tile(aes(fill = value), colour = "#FF3300", size = 2) +
      scale_fill_grey(name = "",
                      labels = c("Present","Missing")) +
      theme_minimal() + 
      theme(axis.text.x  = element_text(angle=90, hjust=1)) + 
      labs(x = "Variables in Dataset",
           y = "Observations")
  }

#Feed the function my new data
ggplot_missing(missmap_data_test)

^{Created on 2019-05-30 by the reprex package (v0.3.0)}

If you're getting notches in the top left corner (discussed here and apparent in the plot above), you may want to update to the development version of ggplot2. That is, devtools::install_github("tidyverse/ggplot2"). For example, compare the plot above with the plot below:

Update

I assume this is a toy example, so I've tried to come up with a generic solution. Here, I've created a function called boxy that will make a data frame for geom_rect based on the original data frame.

#Dataset
missmap_data_test <- data.frame(var1 = c(11, 26, NA, NA, 15),
                                var2 = c(NA, NA, 0, NA, 1))

# Function for making box data frame
boxy <- function(df){
  data.frame(xmin = seq(0.5, ncol(df) - 0.5),
             xmax = seq(1.5, ncol(df) + 0.5),
             ymin = 0.5, ymax = nrow(df) + 0.5)
}

# Load libraries
library(dplyr)
library(ggplot2)
library(reshape2)

#Create Function
ggplot_missing <- function(x){
  df_box <- boxy(x)
  df_rast <- x %>% is.na %>% melt
     
  ggplot() +
  geom_raster(data = df_rast,
              aes(x = Var2,
                  y = Var1,
                  fill = value)) +
  geom_rect(data = df_box, 
            aes(xmin = xmin, xmax = xmax,
                ymin = ymin, ymax = ymax),
            colour = "#FF3300", fill = NA, size = 3) + 
  scale_fill_grey(name = "",
                  labels = c("Present","Missing")) +
  theme_minimal() + 
  theme(axis.text.x  = element_text(angle = 90, hjust = 1)) + 
  labs(x = "Variables in Dataset",
       y = "Observations")
}

#Feed the function my new data
ggplot_missing(missmap_data_test)

^{Created on 2019-05-30 by the reprex package (v0.3.0)}

If you add a third variable (i.e., column) to your data frame, you get something like this:

Thank you @Lyngbakr ! This is very close to what I'm after. However, I don't want the lines around each cell, but rather around each variable. Is there a way to do that? In other words, one large red rectangle around the var1 "column" and a similar large red rectangle around the var2 "column," but no horizontal lines between the top and bottom of the viz. — Emily Bovee, May 30 '19 at 15:46

Is there a way to create borders around rectangles in ggplot2 with geom_raster?

1 Answers1

Update