Interesting geom_area problem ...

[Geoff Russell]

Consider the following ...

data.csv .....

Year,Category,Diff

1990,Elec,-1

1991,Elec,-2

1992,Elec,-3

1990,Land,-1

1991,Land,-2

1992,Land,-3

1990,Water,-1

1991,Water,-2

1992,Water,-3

1990,Sheep,0

1991,Sheep,4

1992,Sheep,9

library(ggplot2)

df=read.csv("data.csv",header=T)

ggplot(aes(x=Year,y=Diff),data=df)+geom_area(aes(fill=Category));

The (dummy) data represent changes in greenhouse emissions in a category.  ie. in 1992 there were 3 less tonnes of GHG generated by electricity and so on, 3 less from

Land Clearing and so on. Extra sheep added 9 extra tonnes by 1992.

I want the increases (positive numbers) to be stacked and shown above zero and the decreases to be stacked and shown above zero. So you 

can see every category and what's happened in that category. But the current geom combines them...

Any ideas?   I could split the data into two panels ... but that seems a bit clumsy.

[Geoff Russell]

Found a good answer ... split the data into positive and negative Categories and then

create two layers, one for the positive Categories and one for the Negative.