R help - Sep 2011 - substitute games with randomForest::partialPlot

I'm having trouble calling randomForest::partialPlot programmatically.
It tries to use name of the (R) variable as the data column name.

Example:

  library(randomForest)
  iris.rf <- randomForest(Species ~ ., data=iris, importance=TRUE,
proximity=TRUE)
  partialPlot(iris.rf, iris, Sepal.Width)   # works
  partialPlot(iris.rf, iris, "Sepal.Width") # works
  (function(var.name)
    partialPlot(iris.rf, iris, var.name))("Sepal.Width") # fails

I can get around this with the following hack:

  (function(var.name)
    eval(substitute(partialPlot(iris.rf, iris, var.name),
                    list(var.name=var.name)))
  )("Sepal.Width") # works

But this seems like a horrible hack.  Is there a "nicer" way to do
this?

The relevant code is:

  function (x, pred.data, x.var, which.class, w, plot = TRUE, add = FALSE, 
      n.pt = min(length(unique(pred.data[, xname])), 51), rug = TRUE, 
      xlab = deparse(substitute(x.var)), ylab = "", main =
paste("Partial Dependence on",
          deparse(substitute(x.var))), ...) 
  {
  ...
      x.var <- substitute(x.var)
      xname <- if (is.character(x.var)) 
          x.var
      else {
          if (is.name(x.var)) 
              deparse(x.var)
          else {
              eval(x.var)
          }
      }
      xv <- pred.data[, xname]
  ...
  }

Thanks,
Johann

> Johann Hibschman wrote:
> I'm having trouble calling randomForest::partialPlot programmatically.
> It tries to use name of the (R) variable as the data column name.
You may want to consider looking at plotmo (in the plotmo package) which 
doesn't have the above issue.

library(randomForest)
library(plotmo)
plotmo.var <- function(var.name)
{
    plotmo(model, degree1=var.name, degree2=0)
}
model <- randomForest(Volume ~ ., data = trees)
plotmo(model)
plotmo.var("Girth")

Plotmo can be considered to be a "poor man's partial dependence
plot" --- it
simply holds the not-being-plotted variables at their median values, rather 
than averaging as in a partial dependence plot.  This may seem a bit 
simplistic, but an advantage is that plotmo plots are easier to understand 
than partial dependence plots (I believe). Or at least it is easier to 
understand what a plotmo plot is NOT telling you.  For additive models, the 
shape of the plotmo curve is identical to the partial dependence curve. 
Currently the quickest introduction to plotmo is pages 17-19 of 
http://www.milbo.org/rpart-plot/prp.pdf

Regards,
Steve
www.milbo.users.sonic.net