I am a new and unexperienced user of R and got so far as to know how to produce boxplots. I have no experience of messing with function code, so presently I do not know how to create a boxplot with group means instead of group medians. If somebody could help me either replace the median with the mean or superimpose the mean onto the existing boxplot, it would be appreciated.
This will add triangles at the mean value. To change the behavior of
boxplot() to draw means instead of medians would involve rewriting the
bxp() code I believe. You could change the points in the code below to
segments. See ?segments. The archives have quite a few examples of
people modifying the bxp() code so look there and see ?bxp.
# Load some data
data(OrchardSprays)
# Make the boxplot and save the boxplot object
rb <- boxplot(decrease ~ treatment, data = OrchardSprays)
# Compute the means
mean.value <- tapply(OrchardSprays$decrease,
OrchardSprays$treatment, mean)
# Add them as triangles.
points(seq(rb$n), mean.value, pch = 17)
HTH, Andy
On Thu, 2004-01-22 at 12:11, Johannes Ullrich wrote:> I am a new and unexperienced user of R and got so far as to know how to produce > boxplots. I have no experience of messing with function code, so presently I do > not know how to create a boxplot with group means instead of group medians. If > somebody could help me either replace the median with the mean or superimpose > the mean onto the existing boxplot, it would be appreciated.You are probably better doing the second option, since boxplots are premised on the visual display of quantiles. Here is a quick example of adding means: # Create two groups of continuous data # and bind them together A <- data.frame(Group = "A", Measure = rnorm(50, 5)) B <- data.frame(Group = "B", Measure = rnorm(50, 7.5)) Data <- rbind(A, B) attach(Data) # Now create a boxplot with two groups using # the formula method boxplot(Measure ~ Group) # Get the group means means <- by(Measure, Group, mean) # Plot symbols for each mean, centered on # x = 1 and x = 2, which are the default center # values. points(1:2, means, pch = 23, cex = 0.75, bg = "red") # Now label the means, formatting the values # to one decimal place. Place the values to the # left of each group plot. text(1:2 - 0.4, means, labels = formatC(means, format = "f", digits = 1), pos = 2, cex = 0.9, col = "red") # Clean up detach(Data) See ?boxplot, ?par, ?points, ?text and ?formatC for more information. You might want to review R News (http://cran.r-project.org/doc/Rnews/) Vol 3 Number 2 (October 2003), where there is an article providing an introduction to R's base graphics. HTH, Marc Schwartz
On Thu, 22 Jan 2004 19:11:08 +0100 Johannes Ullrich <ullrichj at mailer.uni-marburg.de> wrote:> I am a new and unexperienced user of R and got so far as to know how to > produce boxplots. I have no experience of messing with function code, so > presently I do not know how to create a boxplot with group means instead > of group medians. If somebody could help me either replace the median > with the mean or superimpose the mean onto the existing boxplot, it > would be appreciated.You might want to also consider extended box plots or box-percentile plots to show more information, plus the mean. See the panel.bpplot function in the Hmisc package. There are many examples in the help file. --- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University