Hi:
This isn't very difficult if you use a little imagination. We want three
separate plots of monthly means by variable with attached error bars. This
requires faceting, so we need to create a factor whose levels are the
variable names. We also need to generate enough data to summarize by mean
and standard deviation per month in order to generate the point/line/error
bar plots. Here's one attempt.
# Generate fake data: if you don't have an example, make one up! :)
# 20 obs. per month, simulate from normal distribution with a
# given vector of monthly means and standard deviations
########## Set up time variable, monthly means and SDs ##########
# I'm lazy: I just used the three letter month abbreviations in month.abb()
# to use as my 'time' variable
Month <- factor(rep(month.abb, each = 20), levels = month.abb)
# monthly means
mmeans <- c(12, 15, 18, 20, 24, 30, 26, 23, 18, 15, 12, 10)
# monthly sd's
msd <- rep(c(3, 5, 4, 3), each = 3)
################# Generate data ###########################
# Simulate 240 observations for each of the three variables
C <- rnorm(240, m = rep(mmeans, each = 20), s = rep(msd, each = 20))
K <- rnorm(240, m = rep(mmeans+2, each = 20), s = rep(msd+1, each = 20))
S <- rnorm(240, m = rep(mmeans-2, each = 20), s = rep(msd+2, each = 20))
# Combine into a data frame
d <- data.frame(Month, C, K, S)
library(ggplot2) # plyr and reshape get loaded with ggplot2
#### Process data: get monthly means/SDs for each variable ######
# Use ddply to get monthly means and sd's for each variable
# (Yes, there are more efficient ways to do this, but there are only
three...)
md <- ddply(d, .(Month), summarise, C_avg = mean(C), C_stdev = sd(C),
K_avg = mean(K), K_stdev = sd(K), S_avg = mean(S), S_stdev =
sd(S))
# Melt the data from 'wide' to 'long' - the idea is to stack C, K and S
values
# and use their names as a factor variable. This is a very useful trick for
# faceting or grouping.
# grep() is used to select variable names that end in 'avg' or 'stdev'; the
$
# sign in a regular expression indicates that action.
# Thanks to Kohske Takahashi for the clue to melting multiple groups of
variables
# in a post on the ggplot2 list.
dmelt <- data.frame(
melt(md, id = 'Month', measure = c(grep('avg$', names(md)))),
sd = melt(md, id = 'Month', measure = c(grep('stdev$',
names(md))))$value
)
# Some housecleaning: change value to Mean in names and create a new
variable that
# only uses the variable name (C, K, S) as a factor level, to be used for
labeling
# the facets. This reduces the amount of ggplot() code we need to write.
names(dmelt)[3] <- 'Mean'
dmelt$Variable <- substring(dmelt$variable, 1, 1)
# Should be straightforward - scale code is used to avoid overlapping labels
in a
# confined graphics space.
g <- ggplot(dmelt, aes(x = Month, y = Mean))
g + geom_point() + geom_line(aes(group = 1)) +
geom_errorbar(aes(ymin = Mean - sd, ymax = Mean + sd), width = 0.4) +
facet_wrap(~ Variable, nrow = 1) +
scale_x_discrete(breaks = levels(Month),
labels = substring(month.abb, 1, 1))
# Just for the heck of it, here's a monthly plot of each variable's means
(no error bars) as an alternative:
g + geom_point(aes(colour = Variable), size = 3) +
geom_line(aes(colour = Variable, group = Variable), size = 1)
Notice that by limiting the aesthetics in g, I was able to insert additional
aesthetics ymin and xmin into geom_errorbar() in the first plot and colour +
group in the second plot and still use the same g as a foundation.
HTH,
Dennis
On Thu, Nov 4, 2010 at 7:32 AM, ashz <a...@walla.co.il> wrote:
>
> Dear Thierry,
>
> Your solution looks very elgant but I can not find a proper example.
>
> Can you provide me one?
>
> Thx
>
> --
> View this message in context:
> http://r.789695.n4.nabble.com/ggplot-output-tp3027026p3027108.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
