[R] Multiple density curves
Gabor Grothendieck
ggrothendieck at gmail.com
Fri Aug 11 20:40:00 CEST 2006
Here is one more solution. This one uses lattice. Its a bit shorter
than the classic graphics solution. In the classic graphics version we used
shading and color to distinguish the bars; however, grid, and therefore
lattice, do not easily support shading (its possible to simulate it using low
level vector graphics but that's beyond the scope of this) so we use
width (lwd), style (lty) and colour (col) to distinguish them. Also note
that the for loop iterates over the groups since the lattice histogram function
does not use the groups= argument of lattice's xyplot.
library(lattice)
# data
DF <- structure(list(SEQ = structure(c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
11, 12, 13), .Label = c("A", "B", "C", "D", "E", "F", "G", "H",
"I", "J", "K", "L", "M"), class = "factor"), A1 = c(532.5, 25.5,
265.2, 245.55, 546.52, 243.25, 452.55, 15.14, 543.4, 54.4, 646.5,
645.4, 646.54), A2 = c(554.5, 35.5, 522.2, 521.56, 141.52, 32.56,
635.56, 16.54, 646.56, 654.5, 64.54, 614.46, 634.46)), .Names = c("SEQ",
"A1", "A2"), class = "data.frame", row.names = c("1", "2", "3",
"4", "5", "6", "7", "8", "9", "10", "11", "12", "13"))
histogram(~ unlist(DF[,-1]), type = "density",
panel = function(x, breaks, ...)
for(j in 2:ncol(DF)) {
panel.histogram(DF[,j], border = j, lwd = j, lty = j,
breaks = breaks, col = "transparent", ...)
panel.densityplot(DF[,j], col = j, ...)
})
On 8/11/06, Gabor Grothendieck <ggrothendieck at gmail.com> wrote:
> The code below was missing the breaks= argument to hist.
> I had not noticed because coincidentally both give the same
> breaks anways thus the following corrected version gives the
> same plot in this case but might not in other cases.
>
> # data
> DF <- structure(list(SEQ = structure(c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
> 11, 12, 13), .Label = c("A", "B", "C", "D", "E", "F", "G", "H",
> "I", "J", "K", "L", "M"), class = "factor"), A1 = c(532.5, 25.5,
> 265.2, 245.55, 546.52, 243.25, 452.55, 15.14, 543.4, 54.4, 646.5,
> 645.4, 646.54), A2 = c(554.5, 35.5, 522.2, 521.56, 141.52, 32.56,
> 635.56, 16.54, 646.56, 654.5, 64.54, 614.46, 634.46)), .Names = c("SEQ",
> "A1", "A2"), class = "data.frame", row.names = c("1", "2", "3",
> "4", "5", "6", "7", "8", "9", "10", "11", "12", "13"))
>
> # determine breaks and y limits of the combined plot
> breaks <- hist(c(DF$A1, DF$A2), plot = FALSE)$breaks
> ymax1 <- max(hist(DF$A1, breaks = breaks, plot = FALSE)$intensities)
> ymax2 <- max(hist(DF$A2, breaks = breaks, plot = FALSE)$intensities)
> ylim <- c(0, max(ymax1, ymax2))
>
> # draw the two histograms and two densities
> hist(DF$A1, ang = 45, col = "red", ylim = ylim,
> breaks = breaks, freq = FALSE, density = 10)
> lines(density(DF$A1), col = "red")
> hist(DF$A2, ang = -45, col = "blue", add = TRUE,
> breaks = breaks, freq = FALSE, density = 10)
> lines(density(DF$A2), col = "blue")
>
>
>
> On 8/11/06, Gabor Grothendieck <ggrothendieck at gmail.com> wrote:
> > From your description I assume you want both histograms
> > and the densities all on the same chart. With existing R
> > graphics I am not sure that there really is a simple way to
> > do that.
> >
> > That aside, note that the hist function returns a list of
> > components that includes
> >
> > - breaks, defining the breakpoints of the histogram
> > - intensities defining the heights of the histogram bars
> >
> > We can use these two to determine the breaks and y limits
> > of the combined plot and then use the breaks= and ylim=
> > arguments of hist to specify them so that both histograms
> > can be drawn on the same chart. We also use freq=FALSE
> > in the hist calls to draw intensities rather than counts. On
> > the second hist call we use add=TRUE to cause it to be drawn
> > on the existing plot.
> >
> > The other problem is to distinguish the superimposition of
> > the bars and that can be handled by using shading lines of
> > different colors and angles using the col= and angle= and
> > density= arguments of hist.
> >
> >
> > # data
> > DF <- structure(list(SEQ = structure(c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
> > 11, 12, 13), .Label = c("A", "B", "C", "D", "E", "F", "G", "H",
> > "I", "J", "K", "L", "M"), class = "factor"), A1 = c(532.5, 25.5,
> > 265.2, 245.55, 546.52, 243.25, 452.55, 15.14, 543.4, 54.4, 646.5,
> > 645.4, 646.54), A2 = c(554.5, 35.5, 522.2, 521.56, 141.52, 32.56,
> > 635.56, 16.54, 646.56, 654.5, 64.54, 614.46, 634.46)), .Names = c("SEQ",
> > "A1", "A2"), class = "data.frame", row.names = c("1", "2", "3",
> > "4", "5", "6", "7", "8", "9", "10", "11", "12", "13"))
> >
> > # determine breaks and y limits of the combined plot
> > breaks <- hist(c(DF$A1, DF$A2), plot = FALSE)$breaks
> > ymax1 <- max(hist(DF$A1, breaks = breaks, plot = FALSE)$intensities)
> > ymax2 <- max(hist(DF$A2, breaks = breaks, plot = FALSE)$intensities)
> > ylim <- c(0, max(ymax1, ymax2))
> >
> > # draw the two histograms and two densities
> > hist(DF$A1, ang = 45, col = "red", ylim = ylim, freq = FALSE, density = 10)
> > lines(density(DF$A1), col = "red")
> > hist(DF$A2, ang = -45, col = "blue", add = TRUE, freq = FALSE, density = 10)
> > lines(density(DF$A2), col = "blue")
> >
> > On 8/10/06, Davendra Sohal <dsohal at gmail.com> wrote:
> > > Hi,
> > >
> > > I am new to R...a recent convert from SAS.
> > > I have a dataset that looks like this:
> > >
> > > SEQ A1 A2
> > > A 532.5 554.5
> > > B 25.5 35.5
> > > C 265.2 522.2
> > > D 245.55 521.56
> > > E 546.52 141.52
> > > F 243.25 32.56
> > > G 452.55 635.56
> > > H 15.14 16.54
> > > I 543.4 646.56
> > > J 54.4 654.5
> > > K 646.5 64.54
> > > L 645.4 614.46
> > > M 646.54 634.46
> > >
> > > I want to make a histogram each for A1 and A2, with density curves, on the
> > > same plot so that I can see how they overlap.
> > >
> > > Please let me know some simple code for this.
> > >
> > > I looked at ldahist but it was complicated. Anything simpler?
> > >
> > > Thanks a lot,
> > > -DS.
> > >
> > > [[alternative HTML version deleted]]
> > >
> > > ______________________________________________
> > > R-help at stat.math.ethz.ch 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.
> > >
> >
>
More information about the R-help
mailing list