[R] How to wrap or split labels on plot

[Marc Schwartz]

On Fri, 2004-12-03 at 15:00 -0500, Heather J. Branton wrote:
> Dear R gurus,
> 
> I want to wrap labels that are too long for a plot. I have looked at 
> strsplit(), substr(), nchar(), and strwrap(). I think it's some 
> combination but I'm having difficulty trying to figure out the right 
> combo. I think I need to create some new matrix containing the labels 
> already split, though I'm not sure if maybe there is a quick and dirty 
> way to address this without my wandering around the block.
> 
> I am using R 1.9.1, Windows XP.  (Note:  we are currently in the midst 
> of a big project and probably won't upgrade to R 2.0.1 for another 
> couple of weeks -- *unless* that's what I need to do to address these 
> issues.)
> 
> Here is my script with two label options at the bottom that are not working.
> 
> # Settings
> win.graph(width=8, height=8, pointsize=10)
> 
> # Read in data
> test <- matrix(data=c(2.52,9.5,3.07,2.5,1.99,8.95), nrow = 6, byrow=TRUE)
> 
> # Read in and attach labels (names) to data
> rownames(test) <- c("Mount Pleasant","Jordan","Oil City","Pleasant 
> Valley","Village of Lake Isabella","Rosebush")
> 
> # Set plot limits:
> xmax <- nrow(test)
> nvec <- ncol(test)
> ymax <- ceiling(max(test))
> yinc <- 1
> 
> # Generate Pareto order
> test <- test[order(test[,1],decreasing=TRUE),]
> 
> # Set color palette
> MyCols <- rep(c("lightcyan","cornsilk","lavender"), each = xmax)
> 
> # Adjust the margins
> par(mar = c(7, 5, 6, 3))
> 
> # Bar graph
> mp <- barplot(test, beside = TRUE,
>     col = MyCols,
>     axisnames = FALSE,
>     names.arg = rep(names(test),nvec),
>     las = 2,
>     cex.names = 0.75,
>     ylab = "IXYV",
>     ylim = c(0,ymax),
>     yaxt = "n")
> 
> # Set up the y axis tick marks and labels
> ifelse (ymax<=10,decpt <- 2,decpt <- 0)
> ticks <- seq(0, ymax, yinc)
> axis(2, at = ticks, las = 1,
>      labels = formatC(ticks, format = "f", digits = decpt))
> 
> # Draw a box around the whole thing
> box()
> 
> # Draw the x axis labels
> mtext(side = 1, at = rowMeans(mp)-.2, line = .5, las=2, text = 
> strsplit(names(test)," "))
> mtext(side = 1, at = rowMeans(mp), line = .5, las=2, text = 
> strwrap(names(test),7))
> mtext(side = 1, line = 5.5, text = "Division")
> 
> # Draw titles
> title(main="Central", outer=F, font.main=4, line=4)
> title(main="IXYV by Division", outer=F, font.main=2, line=2.5)


Heather,

There is likely to be more than one approach, but the one that I
generally use is to explicitly put a newline character "\n" into the
plot labels where required. So, in this case, you could do something
like:

names(test) <- c("Mount\nPleasant","Jordan","Oil City",
                 "Pleasant\nValley",
                 "Village of\nLake Isabella",
                 "Rosebush")


Also, there are some confusing things in your code, which I suspect may
tie back to your test data versus the actual data you are using. If I am
missing something here, you might want to clarify that, since things
like your colors and other things don't entirely make sense.

Here is something of a simplified approach using the test data as you
have it:

# Test can be a vector
test <- c(2.52, 9.5, 3.07, 2.5, 1.99, 8.95)

names(test) <- c("Mount\nPleasant","Jordan","Oil City",
                 "Pleasant\nValley",
                 "Village of\nLake Isabella",
                 "Rosebush")

# Use sort here
test <- sort(test, decreasing = TRUE)

ymax <- ceiling(max(test))

par(mar = c(7, 5, 6, 3))

# Note that you can use the names here for names.arg
# As a result of the "\n", the titles will print on two lines
mp <- barplot(test, names.arg = names(test),  
              cex.names = 0.8, ylab = "IXYV", 
              yaxt = "n", ylim = c(0, ymax))

ticks <- seq(0, ymax, 1)
axis(2, at = ticks, las = 1,
     labels = formatC(ticks, format = "f",
                      digits = ifelse(ymax <= 10, 2, 0)))

box()

mtext(side = 1, line = 3.5, text = "Division")

# you can combine the two title() calls into one mtext() call
mtext(side = 3, text = c("Central", "IXYV by Division"),
      font = c(4, 2), line = c(4, 2.5))



If your actual data is a more complex matrix, adjust the above
accordingly.

HTH,

Marc Schwartz