I'm running a long MCMC chain that is generating samples for 22 variables.
I have each run of the chain as a row in a matrix.
So: Chain[,1] is the column with all the samples for variable one.
Chain[,2] is the column with all the samples for variable 2, etc.
I'd like to fit all 22 on a single page to print a nice summary. It is
OK if the graphs are small, I just need to show the overall shape and
convergence.
Using par(mfrow=(11,2)) gives me the error: "figure margins too
large"
when I try to draw a plot
I looked at the Lattice package, which seems very promising, but I can't
figure out how to have it plot each column in a separate box.
I need to do this same "one page" summary about 50 times, so it would
be
a nightmare to make over 1,000 plots by hand.
Ideally, I'd create a loop for each of the 50 runs that would generate a
single page containing the 22 plots.
In pseudocode:
for( i in 1:50){
par(mfrow(11,2))
for(j in 1:22){
plot(Chain[,j], type="l")
}
}
BUT, this doesn't work.
Does anyone have any ideas about an easy way to do this?
Thanks!
Hi, Try with: par(mfrow=c(11,2)) It should work better! Remember that mfrow is an argument of the function par(), not a function itself. One other tip: think about using pdf, ps, png, or SVG devices, I find it easier and nicer. HTH, Ivan Le 6/1/2010 10:02, Noah Silverman a ?crit :> I'm running a long MCMC chain that is generating samples for 22 variables. > > I have each run of the chain as a row in a matrix. > > So: Chain[,1] is the column with all the samples for variable one. > Chain[,2] is the column with all the samples for variable 2, etc. > > I'd like to fit all 22 on a single page to print a nice summary. It is > OK if the graphs are small, I just need to show the overall shape and > convergence. > > Using par(mfrow=(11,2)) gives me the error: "figure margins too large" > when I try to draw a plot > I looked at the Lattice package, which seems very promising, but I can't > figure out how to have it plot each column in a separate box. > > I need to do this same "one page" summary about 50 times, so it would be > a nightmare to make over 1,000 plots by hand. > > Ideally, I'd create a loop for each of the 50 runs that would generate a > single page containing the 22 plots. > > In pseudocode: > > for( i in 1:50){ > par(mfrow(11,2)) > for(j in 1:22){ > plot(Chain[,j], type="l") > } > } > > BUT, this doesn't work. > > Does anyone have any ideas about an easy way to do this? > > Thanks! > > ______________________________________________ > R-help at 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. > >-- Ivan CALANDRA PhD Student University of Hamburg Biozentrum Grindel und Zoologisches Museum Abt. S?ugetiere Martin-Luther-King-Platz 3 D-20146 Hamburg, GERMANY +49(0)40 42838 6231 ivan.calandra at uni-hamburg.de ********** http://www.for771.uni-bonn.de http://webapp5.rrz.uni-hamburg.de/mammals/eng/mitarbeiter.php
Hi,
You could use melt from the reshape package to create a long format
data.frame. This is more easy to plot with lattice or ggplot2, and you
can then use facetting to arrange several plots on the same page. The
dummy example below produces 10 pages of output with 10 graphs per
page.
library(ggplot2)
dl <- replicate(10, as.data.frame(matrix(rnorm(1e3), ncol=10)),
simplify=FALSE)
names(dl) <- paste("column", seq_along(dl), sep="")
# dummy list of data.frames
str(dl)
# a function to plot one data.frame
plotone <- function(d, ...)
qplot(seq_along(value), value, data=melt(d)) +
facet_wrap(~variable, scales="free")
# call the function for each data.frame
pl <- llply(dl, plotone)
# print to a file
pdf("test.pdf")
l_ply(pl, print)
dev.off()
HTH,
baptiste
On 1 June 2010 10:02, Noah Silverman <noah at smartmediacorp.com>
wrote:> I'm running a long MCMC chain that is generating samples for 22
variables.
>
> I have each run of the chain as a row in a matrix.
>
> So: ?Chain[,1] is the column with all the samples for variable one.
> Chain[,2] is the column with all the samples for variable 2, etc.
>
> I'd like to fit all 22 on a single page to print a nice summary. ?It is
> OK if the graphs are small, I just need to show the overall shape and
> convergence.
>
> Using par(mfrow=(11,2)) gives me the error: ?"figure margins too
large"
> when I try to draw a plot
> I looked at the Lattice package, which seems very promising, but I
can't
> figure out how to have it plot each column in a separate box.
>
> I need to do this same "one page" summary about 50 times, so it
would be
> a nightmare to make over 1,000 plots by hand.
>
> Ideally, I'd create a loop for each of the 50 runs that would generate
a
> single page containing the 22 plots.
>
> In pseudocode:
>
> for( i in 1:50){
> ? ?par(mfrow(11,2))
> ? ?for(j in 1:22){
> ? ? ? ?plot(Chain[,j], type="l")
> ? ?}
> }
>
> BUT, this doesn't work.
>
> Does anyone have any ideas about an easy way to do this?
>
> Thanks!
>
> ______________________________________________
> R-help at 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.
>
Noah Silverman <noah <at> smartmediacorp.com> writes:> > I'm running a long MCMC chain that is generating samples for 22 variables. > > I have each run of the chain as a row in a matrix.> So: Chain[,1] is the column with all the samples for variable one. > Chain[,2] is the column with all the samples for variable 2, etc.The previous 2 paragraphs seemed contradictory until I realized that in the first paragraph you are using "run" to mean what I would usually call a "sample" ...> I'd like to fit all 22 on a single page to print a nice summary. It is > OK if the graphs are small, I just need to show the overall shape and > convergence.How about for example> x <- matrix(runif(22000),ncol=22) > library(coda) > m <- as.mcmc(x) > xyplot(m) > xyplot(m,layout=c(4,6))
Another thought I had was to use Sweave. I could, in theory, use Sweave to create a plot only 7 inches high, and then a summary table 4 inches high below it, all on the same page of a Latex document. Could be messy, but might work. On 6/1/10 11:18 AM, Dennis Murphy wrote:> > > On Tue, Jun 1, 2010 at 10:51 AM, Noah Silverman > <noah@smartmediacorp.com <mailto:noah@smartmediacorp.com>> wrote: > > You are correct, > > I initially missed the "as.mcmc" step. Without it, R doesn't want > to squeeze so many plots onto a page. > > > I've had that problem before with lattice...which makes me wonder > whether you'll be able to put the plots > and table on a single page. Here are some haphazard thoughts: > > * The gplots package has functions textplot and sinkplot which IIRC > will allow you to add a table to a plot. > I think you need to use layout() in base graphics to set it up > properly, but you'd need to get 22 graphs > and a table on one graphics page... > * Given the number of graphs you want per page, perhaps sparkline > plots might be less wasteful > of space (see the YaleToolkit package for one approach to sparklines) > * The problem is the time it would take to coordinate all of this... > * In a similar vein, there's always grid... :) > > Hopefully Deepayan or Felix Andrews can share some of their Lattice > wisdom and show us how to > do what you want in their system. That would be the most convenient > solution. > > Dennis > > > Thanks! > > > On 6/1/10 10:48 AM, Dennis Murphy wrote: >> Hi: >> >> Ben's example worked for me as follows: >> >> x <- matrix(runif(22000),ncol=22) >> library(lattice) >> library(coda) >> m <- as.mcmc(x) >> xyplot(m, layout = c(2, 11)) >> >> layout = uses columns first, then rows...I always forget that >> :) It also has an optional >> third argument for pages, which might come in handy here. If you >> want the order of the >> panels to be reversed, use the additional option as.table = TRUE. >> >> HTH, >> Dennis >> >> On Tue, Jun 1, 2010 at 9:37 AM, Noah Silverman >> <noah@smartmediacorp.com <mailto:noah@smartmediacorp.com>> wrote: >> >> Hi, >> >> I used the term "run", as each iteration of the Gibbs sampler >> produces >> 22 variables (coefficients for Beta in a regression model) >> >> >> The example wont work >> >> >> On 6/1/10 5:54 AM, Ben Bolker wrote: >> > Noah Silverman <noah <at> smartmediacorp.com >> <http://smartmediacorp.com>> writes: >> > >> > >> >> I'm running a long MCMC chain that is generating samples >> for 22 variables. >> >> >> >> I have each run of the chain as a row in a matrix. >> >> >> > >> >> So: Chain[,1] is the column with all the samples for >> variable one. >> >> Chain[,2] is the column with all the samples for variable >> 2, etc. >> >> >> > The previous 2 paragraphs seemed contradictory until I >> realized >> > that in the first paragraph you are using "run" to mean >> what I would >> > usually call a "sample" ... >> > >> > >> >> I'd like to fit all 22 on a single page to print a nice >> summary. It is >> >> OK if the graphs are small, I just need to show the >> overall shape and >> >> convergence. >> >> >> > How about for example >> > >> > >> >> x <- matrix(runif(22000),ncol=22) >> >> library(coda) >> >> m <- as.mcmc(x) >> >> xyplot(m) >> >> xyplot(m,layout=c(4,6)) >> >> >> > ______________________________________________ >> > R-help@r-project.org <mailto: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. >> > >> >> ______________________________________________ >> R-help@r-project.org <mailto: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]]