geom_area gives odd results for data with duplicated x values

[fmitha]

Call this file dat.csv.

"x","y"
0.1,0.9
0.2,0.8
0.3,0.8
0.3,0.8
0.4,0.7
0.5,0.8
0.6,0.8
0.7,0.8
0.8,0.75
0.8,0.76
0.9,0.8
1.0,0.7

Then

#!/usr/bin/Rscript
library(ggplot2)
stat = read.csv("dat.csv")
png(file = "graph.png")
stat = ggplot() + geom_area(data=stat, aes(x, y))
print(stat)
dev.off()

gives a plot that is a little curious. See displayed graph. The
problem is that the x value 0.3 is duplicated. This might be
considered an error, but could be handled more
gracefully. Additionally, I would like to make a plea for points like
(0.8,0.75) and (0.8,0.76) to be dealt with in a reasonable way. Such
values can easily be produced as a result of rounding errors or
simulations that are not using enough precision. I suspect that
whatever geom_area would do with (0.8,0.75) & (0.8,0.76) is masked by
the duplicate x value problem.

Tested on Debian squeeze with R 2.15.3 and 0.9.3.1.

[wch]

Cleaned-up example:

dat <- read.csv(header=T, text='
x,y
0.1,0.9
0.2,0.8
0.3,0.8
0.3,0.8
0.4,0.7
0.5,0.8
0.6,0.8
0.7,0.8
0.8,0.75
0.8,0.76
0.9,0.8
1.0,0.7
')

library(ggplot2)

# Stacked
ggplot(dat, aes(x=x, y=y)) + geom_area()

I think the problem is being caused by position="stack". This makes it look as expected:

ggplot(dat, aes(x=x, y=y)) + geom_area(position="identity")

[fmitha]

Hi Chang,

You're right. That does fix the problem. I should have looked again at the documentation for geom_area. So, I guess this is user errror. However. the document is a little sketchy. For example, for the page for geom_area, e.g. Area Plot, it is not clear what the different options for position are,
nor are there any examples given.

[fmitha]

Shall I close this?