[R] tapply() and barplot() help files for 1.8.1

[David Whiting]

Martin Maechler <maechler at stat.math.ethz.ch> writes:

> and I like to help you.
> As I keep installed `(almost) all released versions of R ever
> installed on our machines'
> I can easily run 1.8.1 (or 1.4.x or 1.0.x ...) for you.
> 
> The only difference
>  between the help page help(tapply)
> is an extra   "require(stats)" statement at the beginning of the
> `Examples' section in 1.9.0.
> 
> and the only change to  tapply() is 
>     group <- rep.int(one, nx)#- to contain the splitting vector
> instead of
>     group <- rep    (one, nx)#- to contain the splitting vector
> 
> which hardly should have adverse results.
> 
> In barplot, there's the new 'offset' option  --- not in NEWS (!!!!)
> 
> and another change that may be a problem.
> 
> Can you dig harder and if possible provide a reproducible (small..)
> example to make progress here...
> 

Last night I found I had a backup of the source of 1.8.0, built that
and tested an example and it worked as in 1.9.0.  I then started to
question my sanity (or at least my competence).

The code that follows should be a reproducible example.  It creates a
data frame that has the same structure as the data I am working with
(with a number of other columns dropped) and is followed by the
function that creates the barplot.  The changes I have had to make to
make it work as I thought it was working with 1.8.1 have ## NEW BIT
after them, i.e. those lines were not there in the version I ran with
1.8.1.  The important new lines are:

 x <- matrix(x)          ## NEW BIT

and 

 beside = TRUE,          ## NEW BIT



--- EXAMPLE ---

## Create some fake data.
x <- c(rep("", 926), 
        rep("All Other Perinatal Causes", 46), 
        rep("Anaemia", 3), 
        rep("Congenital Abnormalities", 1), 
        rep("Unsp. Direct Maternal Causes", 24))
y <- runif(length(x))
tempdat <- data.frame(smi=x, yllperdth=y)



## Define the function to make my barplot
bodShare <- function(x, fld, main = "", userpar = 18, xlimMult=1.3 ) {
  ###############################################
  # A horizontal barchart to display BoD shares #
  ###############################################
  z <- subset(x, as.character(x[,fld]) != "")
  z[, fld] <- factor(z[, fld])

  ## We need to change the parameters of the chart.
  ## First save the old settings.
  oldpar <- par("mar")
  newpar <- par("mar")

  ## Increase the size of the margin on the left so there 
  ## is enough space for the long text labels (which will 
  ## be displayed horizontally on the y-axis).
  newpar[2] <- userpar

  
  ## Reduce the top margin because I will use a \caption in LaTeX 
  ## instead.
  newpar[3] <- 1


  ## Now apply the new settings.
  par(mar = newpar)

  ## Calculate the % of YLLs for each group in the cause classification.
  x <- tapply(z$yllperdth, z[, fld], sum)
  totalYLLs <- sum(x)
  x <- x / totalYLLs * 100
  x <- sort(x)

  causeNames <- names(x)  ## NEW BIT
  x <- matrix(x)          ## NEW BIT
  

  ## Plot the chart. horiz = TRUE makes it a bar instead of 
  ## column chart.  las = 1 prints the labels horizontally.
  xplot <- barplot(x, 
##                   main = main,
                   horiz = TRUE, 
                   beside = TRUE,            ## NEW BIT
                   names.arg = causeNames,   ## NEW BIT
                   xlab = "Percent of YLLs",
                   xlim = c(0, max(x) * xlimMult), 
                   las = 1)
  
  text(x + (max(x) * .15), xplot, formatC(x, digits=1, format='f'))

  ## Reset the old margin parameters.
  par(mar = oldpar)
  
  ## Write data to a table for export.
  # First we need to remove newlines from labels.
  names(x) <- sub("\n", "", names(x))
  write.table(as.table(x), file = paste("tables/", fld, ".csv", sep=""), col.names=NA, sep="\t")
  names(x) <- causeNames
  x[length(x)]
}

## Create the barplot.
bodShare(tempdat, "smi")


-- 
David Whiting
Dar es Salaam, Tanzania