R help - Oct 2009 - Code improvement

Dear R-Users,

I would like to have the opinion of the list on the following matter. I 
have this generic function that creates multiple lattice scatterplots 
per page based upon different subsets of the same dataset. The use of 
different line/point colors/symbols  in each plot is based upon a 
'group' variable', which is the same for all plots. My goal is to
create
a main legend per page:
1- because the 'group' variable' is the same for all plots, one
legend
per page is sufficient;
2- because the subset of data used for a particular scatterplot does not 
have to contain all unique elements of the group variable in the whole 
dataset, the graphical settings need to be adjusted for each plot based 
upon a general list of settings created from the whole dataset prior to 
the creation of the plots;
3- for the same reason, I cannot use the key argument of xyplot to 
create a legend from the first scatterplot.

Another piece of info is that this generic function could be used to 
create very different categories of graphs, which each require a 
different legend design.

At the moment, I am creating the legend based upon the general graph 
settings and the category of plot. Prior to the creation of the graphs, 
I store grid objects (text, point, line, or rectangle) into a list, with 
also some information about the number of lines that the legend will 
use, and the category of the plot. After the plots are printed to the 
device, I open a  viewport at the bottom of my page and split it 
according to the type of the graph and the number of lines it needs. 
Finally, I just loop through the content of the list and draw the grid 
objects.

Overall, this is all fine but a bit 'brute force'. Also the final 
'open-a-viewport-and-draw-inside' step is a big messy code within a 
generic graph function, because one specific piece of code is needed per 
category of plot, and because of additional subtleties (conditional 
argument, etc... )

Based upon the behavior of the xyplot function, I understand that it 
would be possible to store the fully formatted legend directly at its 
creation and then just 'print' the stored object within my legend 
viewport. Could anybody advise me on the process to follow to accomplish 
that or maybe a few functions to look at?

As always, any help would be greatly appreciated.

Sebastien

Hi,

I don't know if it helps, but looking at the output of xyplot you can
extract the legend (a grid.frame) as follows,

library(grid)
library(lattice)

p = xyplot(x~y, group=x,data=data.frame(x=1:10,y=1:10),
auto.key=list(space="right"))
legend = with(p$legend$right, do.call(lattice:::drawSimpleKey, args))
grid.draw(legend)

and lattice::draw.key might also help if you need to customize the legend.


HTH,

baptiste


2009/10/22 Sebastien Bihorel <sebastien.bihorel at
cognigencorp.com>:
> Dear R-Users,
>
> I would like to have the opinion of the list on the following matter. I
have
> this generic function that creates multiple lattice scatterplots per page
> based upon different subsets of the same dataset. The use of different
> line/point colors/symbols ?in each plot is based upon a 'group'
variable',
> which is the same for all plots. My goal is to create a main legend per
> page:
> 1- because the 'group' variable' is the same for all plots, one
legend per
> page is sufficient;
> 2- because the subset of data used for a particular scatterplot does not
> have to contain all unique elements of the group variable in the whole
> dataset, the graphical settings need to be adjusted for each plot based
upon
> a general list of settings created from the whole dataset prior to the
> creation of the plots;
> 3- for the same reason, I cannot use the key argument of xyplot to create a
> legend from the first scatterplot.
>
> Another piece of info is that this generic function could be used to create
> very different categories of graphs, which each require a different legend
> design.
>
> At the moment, I am creating the legend based upon the general graph
> settings and the category of plot. Prior to the creation of the graphs, I
> store grid objects (text, point, line, or rectangle) into a list, with also
> some information about the number of lines that the legend will use, and
the
> category of the plot. After the plots are printed to the device, I open a
> ?viewport at the bottom of my page and split it according to the type of
the
> graph and the number of lines it needs. Finally, I just loop through the
> content of the list and draw the grid objects.
>
> Overall, this is all fine but a bit 'brute force'. Also the final
> 'open-a-viewport-and-draw-inside' step is a big messy code within a
generic
> graph function, because one specific piece of code is needed per category
of
> plot, and because of additional subtleties (conditional argument, etc... )
>
> Based upon the behavior of the xyplot function, I understand that it would
> be possible to store the fully formatted legend directly at its creation
and
> then just 'print' the stored object within my legend viewport.
Could anybody
> advise me on the process to follow to accomplish that or maybe a few
> functions to look at?
>
> As always, any help would be greatly appreciated.
>
> Sebastien
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>