[Rd] termplot; failure to subset non-dataframe carriers (PR#6327)

john.maindonald at anu.edu.au john.maindonald at anu.edu.au
Sun Jan 4 03:07:25 MET 2004


termplot() does not carry subsetting over to carriers that
are in the environment but not in the data frame.  This
generates a "subscript out of bounds" error.

 > data(ToothGrowth)
 > logdose <- log(ToothGrowth$dose)
 > tooth.lm <- lm(len ~ logdose, data=ToothGrowth)
 > termplot(tooth.lm)     ## Works fine
 > toothVC2.lm <- lm(len ~ poly(dose,2), data=ToothGrowth,
+                   subset=ToothGrowth$supp=="VC")
 > termplot(toothVC2.lm)  ## Works fine
 > toothVC.lm <- lm(len ~ logdose, data=ToothGrowth,
+                  subset=ToothGrowth$supp=="VC")
 > termplot(toothVC.lm)
Error in xy.coords(x, y, xlabel, ylabel, log) :
	subscript out of bounds
 > traceback()
4: xy.coords(x, y, xlabel, ylabel, log)
3: plot.default(xx[oo], tms[oo, i], type = "l", xlab = xlabs[i],
        ylab = ylabs[i], xlim = xlims, ylim = ylims, main = main[i],
        col = col.term, lwd = lwd.term, ...)
2: plot(xx[oo], tms[oo, i], type = "l", xlab = xlabs[i], ylab =  
ylabs[i],
        xlim = xlims, ylim = ylims, main = main[i], col = col.term,
        lwd = lwd.term, ...)
1: termplot(toothVC.lm)

Because logdose is not in the data frame ToothGrowth, it was not  
subsetted.
Pending better suggestions, I offer the following fix:

Replace the lines:
     if (NROW(tms) < NROW(data))
         data <- data[dimnames(tms)[[1]], ]

by

     if (NROW(tms) < NROW(data)) {
         use.rows <- match(dimnames(tms)[[1]], dimnames(data)[[1]])
         data <- data[use.rows,  ]
     }
     else use.rows <- NULL

Then following

             xx <- carrier(cn[[i]])
add:

             if(!is.null(use.rows))xx <- xx[use.rows]

Alternatively, the (conditional) subsetting may be included
within the function carrier()

------------------------------------------------------------------------ 
-------------
This may be a good time to mention that I'd like to be able to
add a smooth to the points that appear when partial.resid=TRUE.
One mechanism would be to allow as argument a panel function that
can be set to panel.smooth.  (The default might be NULL).
I'll forward a more detailed suggestion to r-devel.

--please do not edit the information below--

Version:
  platform = powerpc-apple-darwin6.8
  arch = powerpc
  os = darwin6.8
  system = powerpc, darwin6.8
  status =
  major = 1
  minor = 8.1
  year = 2003
  month = 11
  day = 21
  language = R

Search Path:
  .GlobalEnv, package:methods, package:ctest, package:mva,  
package:modreg, package:nls, package:ts, Autoloads, package:base

John Maindonald             email: john.maindonald at anu.edu.au
phone : +61 2 (6125)3473    fax  : +61 2(6125)5549
Centre for Bioinformation Science, Room 1194,
John Dedman Mathematical Sciences Building (Building 27)
Australian National University, Canberra ACT 0200.



More information about the R-devel mailing list