stat_summary_bin bug?

[eipi10]

In the plot below, you can see that an "extra" bin value is plotted to the right of the data values.

library(ggplot2)

# simulate an example of linear data 
set.seed(1)
N <- 10^4
x <- runif(N)
y <- x + rnorm(N)
dt <- data.frame(x=x, y=y)

ggplot(dt, aes(x, y)) + 
  geom_point(alpha = 0.1, size = 0.01) +
  stat_summary_bin(fun.y=mean, bins=10, size=5, geom="point", colour="red") 

It seems like there could be a bug in stat_summary_bin, because, as shown in the code below, two y values (from the two rows with the highest x-values) are excluded from the binning and end up in an NA bin, which is the one plotted to the right of the data bins in the plot above. I would have expected that by default ggplot would include all the data values in the binning procedure.

p1=ggplot(dt, aes(x, y)) + 
  geom_point(alpha = 0.1, size = 0.01) +
  stat_summary_bin(fun.y=mean, bins=10, size=5, geom="point") +
  stat_summary_bin(fun.y=length, bins=10, size=5, geom="point")

p1b = ggplot_build(p1)
p1b$data[[2]][9:11, c(1,3,5,6)]
p1b$data[[3]][9:11, c(1,3,5,6)]

The issue raised above is based on this SO question.

[thomasp85]

Possible duplicate of #1651

[hadley]

Simpler reprex:

library(ggplot2)

x <- seq(0, 1, length = 1e2)
y <- x + rnorm(length(x))
dt <- data.frame(x, y)

# NOT OK
ggplot(dt, aes(x, y)) + 
  geom_point(colour = "grey80") +
  stat_summary_bin(fun.y=mean, bins=10, geom = "point", colour = "red") 

# OK
ggplot(dt, aes(x)) + 
  geom_histogram(bins = 10, boundary = 0)

[hadley]

That's not an NA bin, it's the bin [1.0, 1.1). This just an unfortunately consequence of the fact that you can't exactly represent 1 / 10 = 0.1 as a floating point number.

[eipi10]

Thanks Hadley.