[R] adding bwplot to existing bwplot

[Deepayan Sarkar]

On 3/27/08, Karin Lagesen <karin.lagesen at medisin.uio.no> wrote:
>
>
>  Hello.
>
>  I have made many normal boxplots where I have added a new boxplot to
>  an existing one. When I have done this, I have used the at command to
>  move the boxplots a bit so that they could fit next to eachother, like
>  this:
>
>  boxplot(data......, at = number_of_categories-0.15)
>  boxplot(data......, at = number_of_categoreis+0.15, add =TRUE)
>
>  Now I am wondering if it is possible to do the same in some way with
>  bwplot. The data I want to plot is like this:
>
>  > operonthings[1:5,]
>           phylum   pid type no_clust no_seqs
>  1  Acidobacteria 15771   5S        1       1
>  2  Acidobacteria 12638   5S        1       2
>  3 Actinobacteria 16321   5S        2       6
>  4 Actinobacteria    92   5S        2       2
>  5 Actinobacteria    87   5S        1       5
>  >
>
>  where phylum and types are the factors I would like to plot no_clust
>  and no_seqs against.I basically want these in the same plot:
>
>  bwplot(no_clust~type|phylum, data = operonthings)
>  and
>  bwplot(no_seqs~type|phylum, data = operonthings)
>
>  Any thoughts on how to do this?

One simple option is to have a conditioning variable distinguishing
the two variables:

foo <-
    data.frame(x = gl(3, 1, 100),
               y1 = rnorm(100),
               y2 = runif(100))
bwplot(y1 + y2 ~ x, foo, outer = TRUE)


However, this will not put them side by side. To do that, you need to
reshape the data, e.g., with


foo2 <- reshape(foo, direction = "long",
                varying = c("y1", "y2"),
                v.names = "y")
bwplot(y ~ factor(time) | x, foo2, layout = c(3, 1))


-Deepayan