On 08-Sep-10 19:16:15, Bert Gunter wrote:
> Ted:
> ?layout
>
> Is this close to what you want?
>
> layout(matrix(1:2, nrow=2),wid=1,heigh=c(1,1), resp= TRUE)
> set.seed(54321)
> X0 <- rnorm(50) ; Y0 <- rnorm(50)
> plot(X0,Y0,pch="+",col="blue",xlim=c(-3,3),ylim=c(-3,3),
> xlab="X",ylab="Y",main="My Plot",asp=1)
> plot(X0,Y0,pch="+",col="blue",xlim=c(-3,3),ylim=c(-3,3),
> xlab="X",ylab="Y",main="My Plot",asp=1)
Thanks Greg, William, Marc & Bert for the suggestions. It seems
that Bert's suggestion of 'layout' may work best for my real example
(the one I posted was a simple one). It is also necessary to add
par(pty="s").
Greg: I must study squishplot() -- it looks useful.
William, Marc: I had tried pty="s" in my real example (I inadvertently
omitted it from the dummy example I posted), but it still did not
work -- see Case B of the real example below.
Anyway, here is an instance of the real problem. DISCLAIMER: This
is being done as an "exploration of concept" and does not necessarily
represent my true state of mind ...
## Code common to all approaches (Alterbative continuations A and B):
library(MASS)
M <- 10 ; N <- 2 ; plur <- "'s"
set.seed(54321)
mu <- (-3)+6*(0:(M-1))/(M-1)
mu <- rep(mu, N) ## M evenly spaced on (-3,3)
Y <- matrix(rnorm(M*N,mean=mu),ncol=N) ## N Y's for each mu
KDE <- kde2d(x=mu, y=Y, h=2.5, n=100, lims=c(-5,5,-5,5,-5.0,5.0))
S.x <- KDE$x ; S.y <- KDE$y ; S.z <- KDE$z
## For each value of Y in the KDE grid, find the mu with max KDE:
Mxs <- numeric(100)
for(j in (1:100)){ Mxs[j] <- S.x[which(S.z[,j]==max(S.z[,j]))] }
## A: Using Bert's 'layout' approach with pty="s" as well:
layout(matrix(1:2, nrow=2),wid=1,height=c(1,1), resp=TRUE)
par(cex=0.5,pty="s")
plot(mu,Y,pch="+",col="blue",asp=1,xlim=c(-5,5),ylim=c(-5,5),
xlab="True mu", ylab="Sampled Y's",
main=paste("KDE for (mu,Y) ",
M," mu's, ",N," Y",plur," per mu",sep=""))
contour(S.x, S.y, S.z, nlevels=7, drawlabels=FALSE, add=TRUE)
lines(S.x,S.y,col="green")
for(j in (1:100)){ Mxs[j] <- S.x[which(S.z[,j]==max(S.z[,j]))] }
lines(Mxs,S.x,col="red")
plot(S.x[1:100],Mxs[1:100],asp=1,pch="+",col="red",
xlim=c(-5,5),ylim=c(-5,5),
xlab="Future Observed Y",ylab="MLE of mu by KDE",
main="MLE of mu by KDE for this Sim vs Future Y")
lines(c(-3,3),c(-3,3),col="green")
lines(c(-3,-3),c(-3,3)) ; lines(c( 3, 3),c(-3,3))
text(0,3.5,"<-- Range of True mu used -->")
##C: Using par(pty="s") with par(mfrow) leaving out 'layout':
plot.new()
par(mfrow=c(2,1),mfg=c(1,1),cex=0.5,pty="s")
plot(mu,Y,pch="+",col="blue",asp=1,xlim=c(-5,5),ylim=c(-5,5),
xlab="True mu", ylab="Sampled Y's",pty="s",
main=paste("KDE for (mu,Y) ",
M," mu's, ",N," Y",plur," per mu",sep=""))
contour(S.x, S.y, S.z, nlevels=7, drawlabels=FALSE, add=TRUE)
lines(S.x,S.y,col="green")
for(j in (1:100)){ Mxs[j] <- S.x[which(S.z[,j]==max(S.z[,j]))] }
lines(Mxs,S.x,col="red")
par(mfg=c(2,1))
plot(S.x[1:100],Mxs[1:100],asp=1,pch="+",col="red",
xlim=c(-5,5),ylim=c(-5,5),
xlab="Future Observed Y",ylab="MLE of mu by KDE",
main="MLE of mu by KDE for this Sim vs Future Y")
lines(c(-3,3),c(-3,3),col="green")
lines(c(-3,-3),c(-3,3)) ; lines(c( 3, 3),c(-3,3))
text(0,3.5,"<-- Range of True mu used -->")
#################################################
Anyway, thanks! It has helped.
Ted.
--------------------------------------------------------------------
E-Mail: (Ted Harding) <ted.hard...@manchester.ac.uk>
Fax-to-email: +44 (0)870 094 0861
Date: 08-Sep-10 Time: 21:39:53
------------------------------ XFMail ------------------------------
______________________________________________
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.
