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(2,2), 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")

# PLOT 4:
emptyPlot(c(-100, 1000), c(-8,8), 
          h0=0, v0=0, 
          xmark=TRUE, ymark=c(-5,5), las=1,
          main="Plot 4: Simplified 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=1, las=2)
# add errorbars:
errorBars(b, avg, sds, border = TRUE)
# 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(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(80, 200, by = 5), zlim=c(80,200))


# 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")

The function plotsurface is recommended for data frame or vector input, which is how most experimental data is loaded in R. Here the data from the gradientLegend example is recycled to illustrate the application of the function plotsurface:

dat <- expand.grid(x=seq(0,1,by=.1), y=seq(0,1, by=.1))
dat$z <- dat$x * dat$y
# inspect the structure of the data:
head(dat)

##     x y z
## 1 0.0 0 0
## 2 0.1 0 0
## 3 0.2 0 0
## 4 0.3 0 0
## 5 0.4 0 0
## 6 0.5 0 0

par(mfrow=c(1,2), cex=1.1)
# PLOT 1: plot the default surface
plotsurface(dat, view=c("x", "y"), predictor="z")

## Warning in gradientLegend(zlim, n.seg = 3, pos = 0.875, dec = dec, color =
## color, : Increase right margin to fit labels or decrease the number of decimals,
## see help(gradientLegend).

# PLOT 2: customized color palette
plotsurface(dat, view=c("x", "y"), predictor="z", 
            color = c('gray25', 'white',  'red'), col=1,
            main="Customized interaction surface", labcex=1)

## Warning in gradientLegend(zlim, n.seg = 3, pos = 0.875, dec = dec, color =
## color, : Increase right margin to fit labels or decrease the number of decimals,
## see help(gradientLegend).

Function plot_error

Plot line with confidence intervals.

# Generate some data:
x <- -10:20
y <- 0.3*(x - 3)^2 + rnorm(length(x))
s <- 0.15*abs(100-y + rnorm(length(x)))

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

# PLOT 1: shaded confidence interval
emptyPlot(range(x), c(-25,100), h0=0, v0=0, main="Symmetric CI")
plot_error(x, y, s, shade=TRUE, lwd=2, col="steelblue")

# PLOT 2: Use of se.fit2 for asymmetrical error bar
cu <- y + 2*s
cl <- y - s
emptyPlot(range(x), c(-25,100), h0=0, v0=0, main="Asymmetric CI")
plot_error(x, y, s, shade=TRUE, lwd=2, col="steelblue")
plot_error(x, y, se.fit=cu, se.fit2=cl, col='red', shade=TRUE, density=30)

Function plot_image

The function plot_image allows to combine images in plots. The packages png, jpeg, or caTools need to be installed to load png, jpg, or gif images directly. For the example below the png image is also available as image object: a list with matrix called 'image' and a vector with colors called 'col'. The original image (© Silver Spoon) can be found on Wikipedia.

# 1 A. load png image directly -- only works with package png:
img <- system.file("extdata", "Netherlands_by_SilverSpoon.png", package = "plotfunctions")
plot_image(img=img, type='png')

# 1 B. load image object:
data(img)
plot_image(img=img, type='image')

By default the image is plotted covering the complete plot region. Alternatively, the image could be added to an existing plot,, to specific coordinates. Below some of the features of the function are illustrated.

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

# PLOT 1: replace colors
plot_image(img=img, type='image', replace.colors = list("#00000000"="#0000FF33", "#B.+"="#99DD99FF"),
           main="the Netherlands")
points(c(.45, .8), c(.6, .85), pch=15)
text(c(.45, .8), c(.6, .85), labels=c("Amsterdam", "Groningen"), pos=1)

# example data
x <- 1:100
y <- -0.01*(x - 30)^2+rnorm(100, mean=100)

# PLOT 2: add picture to existing plot, while keeping original picture size ratio
emptyPlot(100, c(50, 100), h0=0, 
          main="Example data plot")
lines(x, y, lwd=2, col='steelblue')
plot_image(img=img, type='image', add=TRUE, 
           xrange=c(30,70), yrange=c(50,80), adj=1, keep.ratio = TRUE,
           replace.colors = list("#B.+"="steelblue"),
           bty='n')