3

My dataset has three variables: longitude (x), latitude (y), and a continuous variable that for these purposes I'm calling z.

I am trying to use the stat_binhex function in the hexbin library to visualize the sum of z, rather than the count of (x,y) points, by hex bin. I know that there is a straightforward workaround for small sample sizes if I coerce z to a factor, but obviously this is neither feasible nor desirable with a large, continuous z. 

library(ggplot2)
library(hexbin)

x <- runif(1000, -125, -65)
y <- runif(1000, 25, 50)
z <- runif(1000, 1, 30000000)
test <- data.frame(x=x, y=y, z=z)

p <- ggplot(test, aes(x=x, y=y, fill=z))

p <- p + stat_binhex(binwidth=c(.5,.5))

p

> Error in unit(tic_pos.c, "mm") : 'x' and 'units' must have length > 0
E. Vincenti
  • 377
  • 1
  • 6
  • 11

1 Answers1

11

I think the solution is in the manual to ggplot2. The function you may want is [stat_summary_hex][1]:

library(ggplot2)
library(hexbin)

x <- runif(1000, -125, -65)
y <- runif(1000, 25, 50)
z <- runif(1000, 1, 30000000)
test <- data.frame(x=x, 
                   y=y, 
                   z=z)

p <- ggplot(data = test,
            aes(x = x,
                y = y,
                z = z)) +
  stat_summary_hex(fun = function(x) sum(x))

print(p)

You'll end up with something like this: enter image description here

Andy Clifton
  • 4,926
  • 3
  • 35
  • 47