Examples of the package plotfunctions

Jacolien van Rij

Loading library

library(plotfunctions)

Function add_bars

Add bars to a graph. See example Function errorBars.

Function check_normaldist

par(mfrow=c(1,2))
set.seed(123)

# PLOT1: t-distribution:
test <- rt(1000, df=5)
check_normaldist(test)

# PLOT2: skewed data, e.g., reaction times:
test <- exp(rnorm(1000, mean=.500, sd=.25))
check_normaldist(test)

The ideal normal distribution is displayed in gray, whereas the data is represented by the red line. Generally the distribution is checked by a QQ norm plot, and the function check_normaldist may facilitate interpretation.

par(mfrow=c(1,2))
set.seed(123)

# PLOT1: t-distribution:
test <- rt(1000, df=5)
qqnorm(test)
qqline(test)

# PLOT2: skewed data, e.g., reaction times:
test <- exp(rnorm(1000, mean=.500, sd=.25))
qqnorm(test)
qqline(test)

Function dotplot_error

Creating dotplots with error bars, and optionally grouping of the data points.

# example InsectSprays from R datasets
avg <- aggregate(count ~ spray, data=InsectSprays, mean)
avg <- merge(avg, 
    aggregate(count ~ spray, data=InsectSprays, sd),
    by="spray", all=TRUE)
# we could add the type of spray to the averages:
avg$type <- c(1,1,2,2,2,1)

# visualize output
dotplot_error(avg$count.x, se.val=avg$count.y, groups=avg$type, labels=avg$spray)

Function drawDevArrows

Facilitates connections between graphs. Precise information is found in the examples in the help file (help(drawDevArrows).

# 3 panels:
par(mfrow=c(1,3), cex=1.1)

# define x and y
x <- -5:20
y <- x^2

# PLOT 1:
plot(x, y, pch=16)
# convert arrow positions:
ap1 <- getArrowPos(x, y, units="coords")

# PLOT 2:
plot(x, -1*y)
# convert arrow positions:
ap2 <- getArrowPos(x, -1*y, units="coords")
ap3 <- getArrowPos(c(1,1,1), c(1,.5,0), units="prop")

# PLOT 3:
plot(x, -2*y+200, ylim=c(-600,200), pch=18)
abline(h=c(0,-400), lty=3, col='red2')
points(x, -1*y)
ap4 <- getArrowPos(c(0,0,0), c(.75,.5,.25), units="prop")

# DRAW ARROWS:
drawDevArrows(start=ap1, end=ap2, arrows="none", 
              col=alphaPalette("red2", f.seq=c(.1,1), n=length(x)))

drawDevArrows(start=ap3, end=ap4, arrows="end", col="red2", lwd=3, length=.1)

Function emptyPlot

Quickly setting up an empty plot, wrapper around plot(x, y, type='n').

par(mfrow=c(1,3), cex=1.1)

# PLOT 1:
emptyPlot(10,1)

# PLOT 2:
emptyPlot(c(-10, 10), c(-100,500), 
          h0=0, main="Plot 2", xlab="X", ylab="Y")

# PLOT 3:
emptyPlot(c(-100, 1000), c(-8,8), 
          h0=0, v0=0, eegAxis=TRUE,
          main="Plot 3: EEG axes")

Function errorBars

Add confidence intervals or other error bars to a plot.

# load example data:
data(chickwts)

# first calculate means and sd per feeding type:
avg <- with(chickwts, tapply(weight, list(feed), mean))
sds <- with(chickwts, tapply(weight, list(feed), sd))

# barplot:
b <- barplot(avg, beside = TRUE, ylim=c(0,400),
             col="steelblue", las=2)
# add errorbars:
errorBars(b, avg, sds)
# add average:
add_bars(b[length(b)]+diff(b[1:2]), mean(avg), 
         col="red", xpd=TRUE)
errorBars(b[length(b)]+diff(b[1:2]), mean(avg), se(avg), xpd=TRUE)
mtext("mean/SE", at=b[length(b)]+diff(b[1:2]), line=1, side=1, font=2, las=2)

Functions getCoords, getProps, and getFigCoords

These functions facilitate in converting the coordinates to proportions and back. Basically a wrapper around par()$usr.

emptyPlot(c(-10,100), c(-2,2), h0=0, v0=0)

# Proportions to coordinates:
x <- getCoords(c(0, .25, .5, .75, 1))
y <- getCoords(c(0, .25, .5, .75, 1), side=2)
points(x, y, col="red1", pch=1:5, lwd=2, xpd=TRUE)

# wrt figure region:
x <- getCoords(c(0.05, .25, .5, .75, .95), input="f")
y <- getCoords(c(0.05, .25, .5, .75, .95), side=2, input="f")
points(x, y, col="steelblue", pch=1:5, lwd=2, xpd=TRUE)

emptyPlot(c(-10,100), c(-2,2), h0=0, v0=0)
# get plot coordinates:
getFigCoords("p")
## [1] -14.40 104.40  -2.16   2.16
# get figure coordinates:
getFigCoords("f")
## [1] -49.69565 122.47826  -4.20000   3.80000
# get proportions:
getProps(c(20,60,100,500))
## [1] 0.2895623 0.6262626 0.9629630 4.3299663
getProps(c(-2,1,4), side=2)
## [1] 0.03703704 0.73148148 1.42592593

Function `gradientLegend

Add a gradient legend, with colors indicating the z-values in a plot. Note that a surface plot maybe more suited to visualize this 3-dimensional data, for example using the function plotsurface.

dat <- expand.grid(x=seq(0,1,by=.1), y=seq(0,1, by=.1))
dat$z <- dat$x * dat$y

emptyPlot(1, 1, xlab="X", ylab="Y")
points(dat$x, dat$y, col=topo.colors(100)[round(dat$z*99)+1], pch=16, cex=2)
gradientLegend(range(dat$z), color="topo", nCol=100,inside = FALSE, pos=.825)

Function legend_margin

Add a legend in the margins of the plot to save space. Wrapper around legend.

emptyPlot(1, 1, xlab="X", ylab="Y", bty='o')
legend("topright", legend=c("normal", "topright"), pch=21)
legend("center", legend=c("normal", "center"), pch=21)
legend("bottomleft", legend=c("normal", "bottomleft"), pch=21)

legend_margin("topright", legend=c("margin", "topright"), pch=21,
              col="red1", box.col="red1", text.col="red1")
legend_margin("center", legend=c("margin", "center"), pch=21,
              col="red1", box.col="red1", text.col="red1")
legend_margin("bottomleft", legend=c("margin", "bottomleft"), pch=21,
              col="red1", box.col="red1", text.col="red1")

Function marginDensityPlot

Adds a density distribution in the margins of a plot.

Example: the onset of a certain stimuli varies. One would like to indicate on the time axis what is the variation in onsets. This example is illustrated below with a made-up data set.

set.seed(1234)

# grand mean of data:
x <- 1:100
y <- -0.01*(x - 30)^2+rnorm(100, mean=100)
# stimulus onset values:
so <- runif(100, min=20, max=40)+rnorm(100, sd=2)


par(mfrow=c(1,2), cex=1.1)

# PLOT 1
emptyPlot(range(x), range(y), h0=0, 
          main="Data", xlab="Time", ylab="Y")
lines(x, y, lwd=2, col='steelblue')
# add mean of stimulus onset:
abline(v=mean(so), lwd=2)
# add density of stimulus onset in 
marginDensityPlot(density(so), side=1)


# PLOT 2
emptyPlot(range(x), range(y), h0=0, 
          main="More examples", xlab="Time", ylab="Y")
lines(x, y, lwd=2, col='steelblue')
# add mean of stimulus onset:
abline(v=mean(so), lwd=2)
# add density of stimulus onset on top of plot:
marginDensityPlot(density(so), side=3, scale=1, density=25)
marginDensityPlot(density(so), side=3, from=getCoords(0, side=2), scale=1)
# or on left side:
marginDensityPlot(density(y), side=2, col="steelblue")

Functions color_contour and plotsurface

Both functions create a colored surface plot. The function color_contour is a wrapper around the functions image and contour, and a more flexible alternative to filled.contour which cannot be used in subpanels (package graphics, see also the examples in the help file: help(color_contour)). Similar to image and contour, color_contour expects a matrix of values to be plotted. In contrast, the function plotsurface can create a surface of a vector of z-values in a data frame. Below both functions are illustrated.

data(volcano)
x <- 10*(1:nrow(volcano))
y <- 10*(1:ncol(volcano))

par(mfrow=c(1,3), cex=1.1)

# PLOT 1: image and contour
image(x, y, volcano, col = terrain.colors(100), 
      axes = FALSE, xlab="", ylab="")
contour(x, y, volcano, levels = seq(90, 200, by = 5),
        add = TRUE, col = "peru")

# PLOT 2: color_contour
color_contour(x, y, volcano, 
              color = terrain.colors(100), axes=FALSE,
              col="peru", levels=seq(90, 200, by = 5))


# PLOT 3: filled.contour (takes the complete device)
filled.contour(x, y, volcano, color.palette = terrain.colors, axes=FALSE)
# not possible to add contour lines:
contour(x, y, volcano, levels = seq(90, 200, by = 5),
        add = TRUE, col = "peru")