On 4/23/2012 9:24 AM, Matthias Rieber wrote:> Hello,
>
> I've some problem with the ggplot2. Here's a small example:
>
> --8<--
> library(ggplot2)
> molten<-
data.frame(date=c('01','01','01','01',
>
'02','02','02','02'),
>
channel=c('red','red','blue','blue',
>
'red','red','blue','blue'),
> product=c('product1','product2',
> 'product1','product2',
> 'product1','product2',
> 'product1','product2'),
> value=c(1,1,1,1,
> 1,1,1,1))
>
> str(molten)
> molten
>
> ggplot(molten, aes(date, weight = value, fill = channel)) +
> geom_bar(colour = I('black')) + facet_grid(product ~ .)
> -->8--
>
> This gives the expected result (at least I expect this, see attachment
> case1). When I change molten to:
>
> molten<-
data.frame(date=c('01','01','01','01',
>
'02','02','02','02'),
>
channel=c('red','red','blue','blue',
>
'red','red','blue','blue'),
> product=c('product1','product2',
> 'product1','product2',
> 'product1','product1',
> 'product1','product2'),
> value=c(1,1,1,1,
> 1,1,1,1))
>
> I get a strange result(see case2). I expect that for date=02 and
> product=product1 the bar should show 2 'red', but it's just 1.
So the
> total sum is 7 instead of 8.
>
> When I change molten again:
>
> molten<-
data.frame(date=c('01','01','01','01',
>
'02','02','02','02'),
>
channel=c('red','red','blue','blue',
>
'red','red','blue','blue'),
> product=c('product1','product2',
> 'product1','product2',
> 'product1','product1',
> 'product1','product2'),
> value=c(1,1,1,1,
> 1,2,1,1))
>
> I get the expected result, where the bar show 3 'red' and the total
sum
> is 9.
>
> Is it wrong to use geom_bar with that kind of data? I could avoid this
> issue when I cast the data.frame, but I like to avoid that solution.
There is nothing wrong with using bars with this sort of data. There is
a bug in the faceting code of 0.9.0 that will be fixed in 0.9.1 (see
https://github.com/hadley/ggplot2/issues/443 ) which caused duplicate
rows of data to be dropped when there was faceting. That is what you are
seeing in the second example; row 6 is identical to row 7 and is dropped
before plotting. One easy workaround until 0.9.1 comes out is to add
unique column to the data that is otherwise ignored:
molten <-
data.frame(date=c('01','01','01','01',
'02','02','02','02'),
channel=c('red','red','blue','blue',
'red','red','blue','blue'),
product=c('product1','product2',
'product1','product2',
'product1','product1',
'product1','product2'),
value=c(1,1,1,1,
1,1,1,1),
dummy=1:8)
> Matthias
--
Brian S. Diggs, PhD
Senior Research Associate, Department of Surgery
Oregon Health & Science University