How to make a conditioning plot?

Question

I'd like to make a conditioning plot just like coplot in R. Is there a pre-made solution or do I have to make a bunch of plots individually?

Edit: This is an example from r screenshots. It is a way of looking at multivariate data. Each individual graph represents the respective bar in the conditioning variable or factor. In this case you see ozone vs solar for different levels of wind and temp.

Edit2: The R-integration looks interesting. I just found a StackExchange question about returning the plot to an output cell in Mathematica, so that works. I just thought there would be a built in function as these plots seem pretty popular in statistics. In r-code I would just type coplot(a~b|c) to see a vs b for levels of c.

Please link to definition of "conditioning plot" is. Also "what" are you plotting? Is it data, functions? - give the specifics. — Vitaliy Kaurov, Jul 27 '13 at 16:09
Please check out slide 5, bottom section from the notebook that Dillon Tracy presented. Mathematica for Data Science. It uses the iris flower data set to develop a scatterplot similar to the one above. It is fully coded within the notebook. http://www.wolfram.com/events/virtual-conference/spring-2013/resources.html — Zviovich, Jul 27 '13 at 18:29
@PatoCriollo That is called a pairs plot in R. http://gettinggeneticsdone.blogspot.ca/2011/07/scatterplot-matrices-in-r.html — brian, Jul 27 '13 at 19:15
@PatoCriollo Thanks for the interesting link. I think the iris example actually needs improvement, though: the frame labels are cut off because there isn't enough ImagePadding... another common Mathematica plot issue. Anyway, I also can't say I'm familiar with the R terminology for plots, so a definition would be appreciated. — Jens, Jul 27 '13 at 19:21
Here are coplots of the iris data in R http://rrubyperlundich.blogspot.ca/2011/06/r-conditional-plot.html Better formatting and colors are possible of course. — brian, Jul 27 '13 at 19:26

score 7 · Accepted Answer · answered Jul 27 '13 at 21:16

7

Using the airquality data set already included in R. Don't know R commands so I used an example of coplot from the web.

Needs["RLink`"];
InstallR[];
mathematicaRPlotWrapper = RFunction["function(filename, plotfun){
        pdf(filename)
        plotfun()
        dev.off()
        }"];
Clear[getRPlot];
getRPlot[plotFun_RFunction] := 
  With[{tempfile = FileNameJoin[{$TemporaryDirectory, "temp.pdf"}]}, 
   If[FileExistsQ[tempfile], DeleteFile[tempfile]];
   mathematicaRPlotWrapper[tempfile, plotFun];
   If[! FileExistsQ[tempfile], Return[$Failed]];
   Import[tempfile]];
First@getRPlot@
  RFunction[
   "function(){coplot(Ozone~Temp|Solar.R, data=airquality)}"]

enter image description here

answered Jul 27 '13 at 21:16

Zviovich

9,308
1
30
52

Thanks. I don't know what the 'First@' is needed for, I deleted it and everything works. I tried your code and adding color but it doesn't work in PDF files so I switched to png. I then got an error saying "First::normal: Nonatomic expression expected at position 1 in First" so I deleted it. To add color you must escape the quotes. I'll try to add your code with my changes in an answer. – brian Jul 27 '13 at 22:46
I accepted your answer but I'm still curious if there's a native solution. I guess it means writing your own function in Mathematica. – brian Jul 27 '13 at 23:08

brian · Answer 2 · 2013-07-28T02:25:22.460

Needs["RLink`"];
InstallR[];

mathematicaRPlotWrapper = RFunction["function(filename, plotfun){
           png(filename, width=600,height=600)
           plotfun()
           dev.off()
           }"];
Clear[getRPlot];
getRPlot[plotFun_RFunction] := 
  With[{tempfile = FileNameJoin[{$TemporaryDirectory, "temp.png"}]}, 
   If[FileExistsQ[tempfile], DeleteFile[tempfile]];
   mathematicaRPlotWrapper[tempfile, plotFun];
   If[! FileExistsQ[tempfile], Return[$Failed]];
   Import[tempfile]];

getRPlot@RFunction[
  "function(){coplot(Ozone~Solar.R|Temp*Wind,data=airquality,number=4, col=\"red\")}"]

getRPlot@RFunction[
  "function(){pairs(iris[1:4],main = \"Iris Data (cm)\",
         pch = 21, bg = c(\"red\", \"green3\",\"blue\")[unclass(iris$Species)])
  par(xpd=TRUE)
  legend(0, 1, as.vector(unique(iris$Species)),
         fill=c(\"red\",\"green3\", \"blue\"))}"]

I put blank lines where I divided the cells. This way you just run the last cell whenever you need an R-plot. Note you have to escape the quotes on "red"

Just to show how easy it is to make the previous image in R and why I'd like the functionality in Mathematica. Even that pairs scatterplot linked to earlier is just one line now.

https://i.stack.imgur.com/ausmq.png

How to make a conditioning plot?

2 Answers2