This seems like a good start. I look forward to your updating the solution as it could use 8.2 ( not 8.1 ). I had planed to restructure my data so that I could generate one set of lattice plots for each of the filtered subsets. But I have just not gotten this to work properly ( at least yet ). Here's one way to describe the problem: First, I think William Cleveland would say that you don't want to try to put too many points on individual plots, and instead use systems of lattice plots ( "trellis plots" ) and other kinds of graphic techniques. These techniques are apt to be used together with filtering and dimensionality reduction on the back end where now one would be apt to make effective use of server clusters to do that work. Having said that, it's not clear that all the pieces are in place for SAS Studio / SAS VA this at this point to take this approach in general. My guess is that a general solution would be to be use SAS VA to call macros written in SAS Studio from SAS. The macros written in SAS Studio would do some additional filtering, sampling/ stratified sampling), or dimensionality reduction in other ways. But I don't see how to do that now, that is use SAS DS2 code to create data subsets on the fly that can then plotted in SAS VA. So at this point I still plan to do this in SAS Studio, using SGPANEL/SGPLOT functions wrapped by macros. Using this kind of approach would essentially give you the "R" approach to programming things using the "lattice" and "ggplot2" packages of R. And, that would be okay. But I think a more general solution making greater use of SAS VA would be more helpful. But if you could continue to evolve your current approach, a document it using SAS VA 8.2, perhaps that gives another way to make use of SAS VA rather than SAS Studio. So if you can continue to document your own approach so far, that would be quite good. Thanks, rh
... View more