Plotting bar charts on maps in R?

Question

I'm using plotrix in R to plot a state level map of the US. It has an excellent function floating.pie to put pie charts on each state.

I'm wondering if there are similar functions within the plotrix package to display bar charts in each state? (I took a look at the documentation and the functions that deal with bar charts don't seem to have this possibility, but I just wanted to be sure.) I prefer to work within the plotrix package if possible, but feel free to name other packages.

As an example, I would be interested in producing a map similar to this (but for the US):

For my US map, there would be 50 bar charts, one for each state.

I got this map from https://stackoverflow.com/questions/20465070/barplots-on-a-map, but it seems that ggsubplot doesn't work on my version of R (similar to what others have been saying on the post).

Answer 1

I know I'm really late with this, but I think I've found a rather simple solution.

If you have a look of the source code of floating.pie() (e.g. by calling getAnywhere(floating.pie)), you will notice that it uses a very simple yet effective approach: drawing the pie segments as polygons. If all you want from your bar charts is the bars (no labels, axes, etc.), you can follow the same approach and write your own function. Here's a quick and dirty version:

# the function 
mapbars <- function (x, xllc = 0, yllc = 0, barwidth=1, maxheight=10){
  # calculate how long each bar needs to be
  bars <- (x/max(x)) * maxheight
  # get some quick colors
  col <- rainbow(length(x))

  for(i in 1:length(x)){
    # figure out x- and y coordinates for the corners
    leftx   <- xllc + ((i-1) * barwidth)
    rightx  <- leftx + barwidth
    bottomy <- yllc
    topy    <- yllc + bars[i]
    # draw the bar
    polygon(x=c(leftx, rightx, rightx, leftx, leftx),
            y=c(bottomy, bottomy, topy, topy, bottomy),
            col=col[i])
  }
}

• x is for the values to be represented by the bars
• xllc and yllc specify the position of the lower left corner of the left bar in whatever coordinate system you are using
• barwidth and maxheight are used for scaling the size of the bars

Here's a demo with a basic sp-based plot. I don't think I've worked with plotrix before, but based on how floating.pie works, I'd assume that this should also work with plotrix.

library(sp)
library(maptools) # just for easy access to a background map
# load some country borders as a background
data("wrld_simpl")
plot(wrld_simpl)

# zoom on a bit …
mexico <- subset(wrld_simpl, NAME=="Mexico")
plot(mexico, axes=TRUE)

# data for the bars
x1 <- c(4, 7, 1, 2)

# plot
plot(mexico, axes=TRUE)
mapbars(x=x1, xllc=-110, yllc=20, barwidth=.5, maxheight=5)
legend(x="topright", pch=22, col="black", pt.bg=rainbow(x1), legend=c("foo", "bar", "baz", "foobar"))

# add another one:
x2 <- c(9, 21, 64, 45, 33, 43, 12, 7)
mapbars(x=x2, xllc=-100, yllc=25, barwidth=.2, maxheight=2)

The result looks like this:

Answer 2

ggplot2 and ggvis are two libraries which may help you to display the plots on the map. In ggplot2 you can plot bubbles on the map and then you just need to give aes() the coordinates which have nothing to do with the size and color of bubble. Regarding bar chart, you need to assign at least 2 pairs of x and y, one for bar chart location and the other for the height and width of the bar chart. In other words, you need to know the coordinates of 4 corners for a bar.