How to customize base R plots and subplots in detail? - r

There is this figure in an economics paper:
I want to style my plots just like these -- with invisible top axis, Y-axis values on the right hand side, axis labels on the top of it and aligned horizontally, subplot titles aligned to the left and each subplot taller than wider and thus emphasizing change along Y-axis. I primarily use MATLAB and I tried to fiddle with it to reproduce something like this but in vain. I then reached out to one of the authors of the paper asking if he could tell which application he used for plots and if he could share how to style my plots after theirs. He responded saying he didn't have the code but he thought it was done in R.
I have not seen plots like this being done in R and even after hours of internet trawling, I didn't find anything R-generated which looked even remotely similar. Will appreciate if you folks have any thoughts/advice on whether it is indeed possible to do it in R.
EDIT
Courtesy of inputs by Isabella Ghement and after whuber's comments, I tried plotting one of the 4 by 2 panels in the figure in my question. Here is how it looks:
Ruefully, this is quite different from the panels in the question. Presumably, if I can get one panel right, I can then prepare a figure containing m by n subplots. That said, this figure has only two elements that I want -- invisible top axis and subplot title aligned to the left. But its tick marks are outside, X-axis doesn't meet the two Y-axes and axis labels are still printed vertically beside them. Here's the code that produces the above plot (taken from https://www.statmethods.net/advgraphs/axes.html, thanks Isabella Ghement for the suggestion):
# specify the data
x <- c(1:10); y <- x; z <- 10/x
# create extra margin room on the right for an axis
par(mar=c(5, 4, 4, 8) + 0.1)
# plot x vs. y
plot(x, y,type="b", pch=21, col="red",
yaxt="n", lty=3, xlab="", ylab="", bty="n")
# add x vs. 1/x
lines(x, z, type="b", pch=22, col="blue", lty=2)
# draw an axis on the left
axis(2, at=x,labels=x, col.axis="red", las=2)
# draw an axis on the right, with smaller text and ticks
axis(4, at=z,labels=round(z,digits=2),
col.axis="blue", las=2, cex.axis=0.7)
# add a title for the right axis
mtext("y=1/x", side=4, line=3, cex.lab=1,las=2, col="blue")
# add a main title and bottom and left axis labels
title("(a) Some Variable", xlab="X values",
ylab="Y=X", adj=0)
I was hoping for simple way of generating such plots but seems like it's a lot of handiwork.
FURTHER EDIT
Though one of the authors wrote back saying he thought plots were in done in R, I do suspect at this point, like #iayork, that probably they weren't done in R. I looked at online appendix of another paper in which one of the authors of the paper in this question is a coauthor and that paper too has plots of similar style. Look at this plot for example:
When I looked at file properties of that plot (it's a PDF file that comes with online appendix in a zipped folder but contains no code), I saw this:
After seeing this, I immediately thought it was generated by S-PLUS but after spending hours on end, I didn't come across anything online on S-PLUS site or any other paper which bore any resemblance. And it was then that I thought it might not be that even though it seems like so by looking at file properties. And then as a last resort, I tried to contact the authors but couldn't get anything useful yet.

You've already deactivated the yaxt, so deactivate xaxt as well. In axis we can omit the labels with labels=FALSE. May I then introduce mtext with which you can create labels independently from the axis. Yes the inward ticks are done with tck=-something. To obtain the axes touch each other we start at 0 and end slightly higher as the maximum value, overlap should be hidden automatically. Add legend. Finally it's wise to use png device to obtain desired aspect ratio. I think that's it, in a nutshell?
# specify the data
x <- c(0:10); y <- ((x^2)-20)/100; z <- (100/x-3)/100
# helper variables
t.adj <- .03
y.seq <- c(-.5, seq(0, 1.5, length.out=4))
png("ecn.plot.png", width=400, height=500)
# margins
par(mar=c(4, 4, 4, 5) + 0.1)
# plot x vs. y
plot(x, y,type="l", pch=21, col="red", xlim=c(0, 10), ylim=c(-.42, max(y.seq)),
yaxt="n", xaxt="n", lty=1, xlab="", ylab="", bty="n", lwd=2)
# add x vs. 1/x
lines(x, z, type="l", pch=22, col="blue", lty=2, lwd=2)
# add y-zero line
abline(h=0, lwd=2)
# draw axes
axis(1, at=(-1:6)*2,labels=FALSE, col.axis="black", tck=t.adj, lwd=2)
mtext((0:5)*2, 1, 0, at=(0:5)*2,col.axis="black", font=2)
axis(2, at=y.seq, labels=FALSE, col.axis="red", las=2, tck=t.adj, lwd=2)
axis(4, at=y.seq, labels=FALSE, col.axis="red", las=2, tck=t.adj, lwd=2)
mtext(formatC(sort(c(0, y.seq)), digits=1, format="f"), 4, 2,
at=sort(c(0, y.seq)), col="black", las=2, font=2, adj=1)
mtext("pct.", 4, 0, at=max(y.seq)+.15, las=2, adj=1, font=2, cex=.9)
# add title
mtext("(a) Some Variable", padj=-2, adj=0, cex=1.2, font=2)
# add legend
legend(x[2], max(y.seq), c("Home", "Foreign"), lty=c(1, 2),
col=c("red", "blue"), bty="n", cex=.8)
dev.off()
Result
Ah, for the arranging of multiple plots you may want to look at this answer.

Related

R: two axes and grid lines

I would like to plot two graphs in the same plotting region with horizontal grid lines. Each side of the grid lines should give the value for one graph or the other. There should be no y-axis.
The grid() function allows me to simply set the number of bins using the ny= argument. How do I get the corresponding labels to the grid lines? Usually, I would use axis(..., lwd=0) to get the labels. However, the function requires label positions with at=c() and does not feature a ny= argument. Is there a way to automatically set the locations from the number of bins?
Based on Miff's hint below, this should solve the problem.
plot(1:10, axes=FALSE, ylim=c(0,10), ylab="")
par(yaxp=c(0, 10, 5))
axis(2, lwd=0, col.axis="gray")
par(new=TRUE)
plot(60:50, axes=FALSE, ylim=c(50,60), ylab="")
par(yaxp=c(50, 60, 5))
axis(4, lwd=0, col.axis="gray")
grid(NA, NULL)
grid() gets its locations for gridlines from axTicks(), which in turn uses numbers from par("yaxp"). If you modify this parameter (rather than explicitly passing it to grid), the result will then apply to both the grid drawn and the axis. For example:
plot(1:10, axes=FALSE)
axis(2) #Default 4 sections between ticks
par(yaxp=c(par("yaxp")[1:2], 7)) #Lets have seven instead
axis(4)
grid() #Grid now matches with right rather than left
Obviously similar works for the x axis.

minor.tick() position in R

I've extensively searched for a solution without luck. I'd like to plot functions in the usual way (axes cross at (0,0)) using the default R plotting facilities.
Setting the axes position to 0 as in the example below solve partially the problem. However when adding minor ticks the default position is retained resulting in the weird plot below.
plot.new()
x <- seq(-10,10, 0.1)
f <- ((x+2)*(x-5))/((x-3)*(x+1))
plot.window(xlim=c(-10,10), ylim=c(-30,30))
axis(side=1, at=seq(-10,10,2), pos=0, las=0)
axis(2)
library(Hmisc)
minor.tick(nx=5, ny=10, tick.ratio=0.5)
One solution would be to still use the axis function, and just specify the locations and size (tck) of the minor ticks.
plot.new()
x <- seq(-10,10, 0.1)
f <- ((x+2)*(x-5))/((x-3)*(x+1))
plot.window(xlim=c(-10,10), ylim=c(-30,30))
axis(side=1, at=seq(-10,10,2), pos=0, las=0)
axis(side=1, at=seq(-10,10,0.5), pos=0, las=0, tck=-0.01, labels=FALSE)
axis(2)
This allows you to retain complete control over the plot.

R: combine mtext() with points() - correct way to specify coordinates?

I want to combine a text I place in the margins with mtext() with a graphics object that I create either using points() or polygon(). The following example roughly works for me with the default plot settings:
plot(1)
mtext("This is a red dot:", side=1, line=2, cex=0.8)
par(xpd=T)
points(1.08, 0.512, pch=15, cex=1.5, col="red")
However, using plot(1:10) instead or prefixing it with windows(8,8) puts the dot in the wrong position, as points() takes user coordinates. Is there a way to get my dot placed correctly independent of plot limits or device size?
I think I have found an answer that works based on the suggestions of #koekenbakker. To make it look pretty, some minor adjustments are still necessary to get the dot in line with the text, but it seems to work well independent of plot size and axis limits (but you cannot resize the plot once created). To do the adjustments on the exact position of the point, I would recommend using fractions of strheight("O", cex=0.8) and strheight("O", cex=0.8) for this example (which uses cex=0.8 for demonstration).
plot(1)
mtext("This is a red dot:", side=1, line=2, cex=0.8, col="green") ## place sample text at bottom of figure
par(xpd=T) ## enable plotting outside plot region
textXPos <- mean(par("usr")[1:2]) ## x position is middle of plot
textYPos <- par("usr")[3] - strheight("O") * 4 ## y position is below bottom of plot (line 2 = 4 * height of letter O)
text(textXPos, textYPos, "This is a red dot:", cex=0.8, col="red", adj=c(0.5, 0)) ## this text overlaps - y position is correct
pointXPos <- textXPos + 0.5 * strwidth("This is a red dot:", cex=0.8) ## text is centred, so need to move half of the text width to the right
points(pointXPos, textYPos, pch=16, col="red") ## still needs minor adjustments to x and y position to make it look nice

Overlay custom axes to a plot created by symbols

Struggling with this one. I have 25 data items that I want plotted with bubbles in 5 columns.
The plot can be re-created thus:
xcord <- c(1,1,1,1,1,2,2,2,2,2,3,3,3,3,3,4,4,4,4,4,5,5,5,5,5)
ycord <- c(1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5,1,2,3,4,5)
zsize <- c(2,1,2,3,6,8,9,1,4,5,5,6,7,8,8,9,5,5,5,5,1,8,1,1,12)
Save the parameters before I change them:
op <- par()
dev.off()
Change the parameters:
par (mfrow=c(1,0), mar=c(2,2,2,2), oma=c(2,2,2,2))
Plot using symbols:
symbols(xcord, ycord, zsize, inches=0.3, fg="white", bg="red", xlab="Events", ylab="Diagnoses", tck=0, xaxt="n", yaxt="n")
mtext("Rheumatic diagnoses by cerebrovasular events", line=2, font=2, cex=1.2)
I am happy with the above plot and deliberately used tck=0, xaxt="n", yaxt="n" to clear the axes. I want to manually overlay custom text, controlled with custom co-ordinates (which work with the sysmbols plot), but have not been able to do it. I have tried some of the par arguments and the axes function.
I also tried leaving the axes on:
symbols(xcord, ycord, zsize, inches=0.3, fg="white", bg="red", xlab="Events", ylab="Diagnoses")
but do not know how to change the output (1,2,3,4,5) to my own custom axes labels.
Thank you.
You are looking for the axis function (see ?axis for details), e.g. to replace the 1:5 with A, B, C, D, E:
axis(side=1, at=1:5, labels=LETTERS[1:5])

R barplot show values

I am plotting a barchart and face the following problem:
The barchart consists of negative and positive values. I want to show these values at the end each bar. When I choose pos=4 the positive values show up nicely but the negative ones are not so nice. When I choose pos=2 the same problem occurs vice versa...
What can I do... Please find my lines attached:
par(pin=c(9,9), mar=c(1,12,1,1), col=c(rgb(126,126,126,maxColorValue=255)), xpd=F)
PBG <- barplot(Gewicht_tsr, beside=T, horiz=T, space= c(.5,1), las=2,
cex.name=0.7, xlim=c(min(Gewicht_d)*1.7, max(Gewicht_d)*1.7),
border=NA, axes=F, main="Sektoren Gewichtung",
col=c(rgb(206,165,90,maxColorValue=255),
rgb(180,169,162,maxColorValue=255),
rgb(0,116,77,maxColorValue=255)),
col.axis=c(rgb(126,126,126,maxColorValue=255)),
col.main=c(rgb(126,126,126,maxColorValue=255)), width=.8)
legend("bottomright", legend=c("Portfolio", "Benchmark", "Aktiv"),
ncol=3, pch=15, bty="n",
col=c(rgb(0,116,77,maxColorValue=255),
rgb(180,169,162,maxColorValue=255),
rgb(206,165,90,maxColorValue=255)),
cex=0.7)
y <- (as.matrix(Gewicht_tsr [,1:ncol(Gewicht_tsr)]))
text(y,PBG, labels=as.character(y), cex=0.62, pos=4)
All other things work fine.

Resources