Hi:
I think it would be tough to do that in qplot(), but it's easier in
ggplot(), even if you don't add the mean information to the data frame.
Here's one way - use the three person data frame (call it dat1) and the
mean.y data frame that you created from aggregate() without adding the
factor info as follows:
# Set up the framework of the plot:
g <- ggplot(dat1, aes(x = time, y = y, groups = ID, colour = ID))
# This associates colors with groups. Next, add the points and
# lines from dat1 and then add the mean data with a separate
# geom_line() call, where the mean line is about twice as thick:
g + geom_point(size = 2) + geom_line() +
geom_line(data = mean.y, aes(x = time, y = y, colour = 'mean'), size =
1.5)
# Notice how the name 'mean' that we associated with colour got into the
# legend. This is because we *mapped* the same aesthetic (color) in the
second
# geom_line() call to the one existing for IDs. ggplot2 is smart enough to
pick
# this up. [We just have to be smart enough to realize it :)]. To exert more
# control over line colors, add the following:
last_plot() + scale_colour_manual(values = c('1' = 'red', '2' = 'green',
'3' = 'blue', 'mean' = 'black'))
The LHS is the value of ID, the RHS the color to associate with it.
As usual, it took me about five iterations of scale_* to get it right :) The
line
thicknesses in the scale are all the same as the thickest, but I see that as
a
feature rather than a bug :)
One more comment below.
On Mon, Aug 23, 2010 at 6:20 PM, Kingsford Jones
<
[hidden email]>wrote:
> On Mon, Aug 23, 2010 at 6:19 PM, Dennis Murphy <
[hidden email]> wrote:
> >
> > This is an excellent idea - the only snag might occur if someone wants
> > the mean line to be thicker :)
>
> fortunately, with your lattice solution this is easily accomplished by
> passing a vector to lwd:
>
> i <- c(1, 1, 1, 3)
>
I was going to do that, too, but I used 1.5 instead of 3, saw no material
difference,
and gave up...should have kept trying, huh?
HTH,
Dennis
> mykey <- list(space = 'right',
> title = 'ID',
> cex.title = 1.2,
> text = list(levels(dat$ID), cex = 0.8),
> lines = list(lty = i, lwd = i, col = 1:4))
>
> xyplot(y ~ time, data = dat, lty = i, lwd = i, col.lines = 1:4, col = 1:4,
> groups = ID, type = c('g', 'p', 'l'), key = mykey)
>
>
> but I didn't have luck trying the same with qplot:
>
> > qplot(time, y, data = dat, group = ID, color = ID,
> + geom = c('point', 'line'), lty = i, lwd = i)
> Error in data.frame(colour = c(1L, 1L, 1L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, :
> arguments imply differing number of rows: 18, 4
>
> perhaps using the construct ggplot(...) + geom_line(...) would be more
> fruitful?
>
> King
>
>
>
>
> > Having said that, it's usually easier to
> > 'fix' the
> > problem externally in the data rather than to fiddle with graphics
> commands.
> >
> >>
> >> #the original data (no replicates within time points)
> >> dat <- structure(list(ID = structure(c(1L, 1L, 1L, 2L, 2L, 2L, 2L),
> >> .Label = c("1", "2"), class = "factor"),
> >> time = c(1, 2, 3, 1.5, 4, 5.5, 6),
> >> y = c(1.4, 2, 2.5, 2.3, 4.5, 1.6, 2)),
> >> .Names = c("ID", "time", "y"),
> >> row.names = c(NA, -7L), class = "data.frame")
> >>
> >> #adding another subject to introduce replicates
> >> id3 <- data.frame(ID=as.factor(rep(3, 4)),time = c(1, 1.5, 2, 5.5),
> >> y = c(1, 2.2, 3, 2))
> >> dat <- rbind(dat, id3)
> >> mean.y <- aggregate(formula = y ~ time, data = dat, FUN = mean)
> >> mean.y <- cbind(ID = as.factor('mean'), mean.y)
> >> dat <- rbind(dat, mean.y)
> >> dat
> >> library(ggplot2)
> >> qplot(time, y, data=dat, group = ID, color = ID, geom = c('point',
> >> 'line'))
> >
> > A lattice version with a legend is:
> >
> > mykey <- list(space = 'right',
> > title = 'ID',
> > cex.title = 1.2,
> > text = list(levels(dat$ID), cex = 0.8),
> > lines = list(lty = 1, col = 1:4))
> >
> > xyplot(y ~ time, data = dat, lty = 1, col.lines = 1:4, col = 1:4,
> > groups = ID, type = c('g', 'p', 'l'), key = mykey)
> >
> > Defining the key externally modularizes the problem, lets one define
> > the features one wants to contain, and simplifies the high-level
> > xyplot() call.
> >
> > There is a type = 'a' (shorthand for panel.average()) that can be
> > used to good effect in xyplot(), but it creates 'holes' where missing
> > data reside, so taking care of the problem externally at the data
> > level is much cleaner.
> >
> > HTH,
> > Dennis
> >
> >>
> >> best,
> >>
> >> Kingsford Jones
> >>
> >> ______________________________________________
> >>
[hidden email] 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]]
______________________________________________
[hidden email] mailing list
https://stat.ethz.ch/mailman/listinfo/r-helpPLEASE do read the posting guide
http://www.R-project.org/posting-guide.htmland provide commented, minimal, self-contained, reproducible code.
Locked
