Create non-overlapping stacked area plot with ggpl

问题:

I have some data scraped and processed from the web in this form:

>head(dat)
  count  name          episode    percent
1   309   don 01-a-little-kiss 0.27081507
2   220 megan 01-a-little-kiss 0.19281332
3   158  joan 01-a-little-kiss 0.13847502
4   113 peggy 01-a-little-kiss 0.09903593
5   107 roger 01-a-little-kiss 0.09377739
6    81  pete 01-a-little-kiss 0.07099036

I'm trying to created a stacked area chart, similar to the one here: Making a stacked area plot using ggplot2

When I do a

require(RCurl)
require(ggplot2)
link <- getURL("http://dl.dropbox.com/u/25609375/so_data/final.txt")
dat <- read.csv(textConnection(link), sep=' ', header=FALSE, 
             col.names=c('count', 'name', 'episode'))

dat <- ddply(dat, .(episode), transform, percent = count / sum(count))

ggplot(dat, aes(episode, percent, group=name)) + 
     geom_area(aes(fill=name, colour=name), position='stack')

I get this bizarre chart.

I want the areas not to cross eachother, and to fill the entire canvas as the total percent for each episode factor equals 100%.

回答1:

That was interesting. You're missing a single row (Lane didn't appear in Tea Leaves...?), so

dat2 <- rbind(dat,data.frame(count = 0,name = 'lane',
                    episode = '02-tea-leaves',percent = 0))

ggplot(arrange(dat2,name,episode), aes(x = episode,y = percent)) + 
  geom_area(aes(fill=name,group = name), position='stack')

appears to work. But it had to be in the right order as well, and I'm not entirely sure why.