5 Visualizing data in R

In this chapter you will learn the basics to build insightful graphics in R.

5.1 Basic plot instructions

5.1.1 Scatterplot

Your starting point is the construction of a scatterplot. You’ll work with the Journals data from the AER package.

# load the 'Journals' data set in the AER package
data("Journals")
# scan the data
head(Journals)
                                                     title
APEL                     Asian-Pacific Economic Literature
SAJoEH           South African Journal of Economic History
CE                                 Computational Economics
MEPiTE MOCT-MOST Economic Policy in Transitional Economics
JoSE                            Journal of Socio-Economics
LabEc                                     Labour Economics
                    publisher society price pages charpp citations foundingyear
APEL                Blackwell      no   123   440   3822        21         1986
SAJoEH So Afr ec history assn      no    20   309   1782        22         1986
CE                     Kluwer      no   443   567   2924        22         1987
MEPiTE                 Kluwer      no   276   520   3234        22         1991
JoSE                 Elsevier      no   295   791   3024        24         1972
LabEc                Elsevier      no   344   609   2967        24         1994
       subs             field
APEL     14           General
SAJoEH   59  Economic History
CE       17       Specialized
MEPiTE    2      Area Studies
JoSE     96 Interdisciplinary
LabEc    15             Labor
names(Journals)
 [1] "title"        "publisher"    "society"      "price"        "pages"       
 [6] "charpp"       "citations"    "foundingyear" "subs"         "field"       
# e.g. get variable 'price' 
Journals$price
  [1]  123   20  443  276  295  344   90  242  226  262  279  165  242  905  355
 [16]  375  135  171  284  242  371  115  355  355  835  223  172   62  191  411
 [31]  274  130  100   80  235  392  410  464  650  558  317  495  535  123  717
 [46]  481   54  379  168   82  355   95  240  448  255  448  392  475   85  108
 [61]  394  336  565  255  165   99  203  318  476  473  186  170  824  805  132
 [76]   50  424  130   90  805   96  448  130  595  474  410  395  437  270  265
 [91]  899  133  262  506 1140  211  799  760  442  296  272   45  614  436  481
[106]   95  357  280  142  710  490  870 1147  743  759   36  224   82  160 1170
[121]   90  742  575  163  175  120  590 2120  205  128 1539  346 1046   97  686
[136]  914   85  206 1154   45 1146   95  138  115  640  122  110  923 1000 1234
[151] 1492  810   90  177   74  113  145  590 1154 1450 1431   47   45   47   81
[166] 1010  334  190  180 1893 1400  301 1339   90  310  226  148  159  178   47
summary(Journals$price)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
     20     134     282     418     541    2120 
# focus on price of journal per citation
Journals$citeprice <- Journals$price/Journals$citations

In the next block of code, we construct a scatterplot of the number of subscriptions versus the price per citation.

with(Journals, plot(log(subs), log(citeprice)))
with(Journals, rug(log(subs)))  # adds ticks, thus visualizing the marginal distributions of
            # the variables, along one or both axes of an existing plot.        
with(Journals, rug(log(citeprice), side = 2))

# avoid "attach()" and "detach()"
plot(log(subs) ~ log(citeprice), data = Journals)

R has many plotting options that allow you to flex a graph. For example, pch for the plotting character, col for the color of the plotting characters, xlim and ylim to adjust the limits on the x- and y-axis of the scatterplot. To add a legend, you can use the legend() function.

plot(
  log(citeprice) ~ log(subs),
  data = Journals,
  pch = 19,
  col = "blue",
  xlim = c(0, 8),
  ylim = c(-7, 4),
  main = "Library subscriptions"
)
rug(log(Journals$subs))
rug(log(Journals$citeprice), side = 2)
# subset data, look at journal entitled "Econometrica"
journal <- "Econometrica"
journal_info <- subset(Journals, title == journal)
x.val <- log(journal_info$subs)
y.val <- log(journal_info$citeprice)
text(x.val, y.val, journal, pos = 2)
legend("topright", "Individual observations", col = "blue", pch = 19)

You can even use your own images as plotting characters! In case you want to find out how you do this, just go to http://www.statisticstoproveanything.com/2013/09/using-custom-images-as-pch-values-in-r.html

5.1.2 Graphical parameters

As briefly mentioned above, you have a lot of arguments to change the appearance of your plot. To see which ones you can adjust, go to the help file of par

?par

Below are some illustrations that show which ones you can use. When you want to adjust something in your plot and can’t immediately find it, just Google it and you will find lots of examples.

x <- 1:10
y <- exp(1:10)
plot(x, y, type = "l", lwd = 3, lty = 2, col = "red")

plot(x, y, bty = "n", cex = 3, cex.axis = 1.5, cex.lab = 2, las = 1, cex.main = 3, main = "Random plot")

When we want to adjust these graphical parameters globally, we can adjust them with the par function. Further, at the beginning of each session, it’s always a good idea to save the initial parameter settings.

Op = par(no.readonly = T) # Save initial parameter settings
par()$mar                 # Check current setting
[1] 5.1 4.1 4.1 2.1
par(mar = c(5, 5, 2, 2) + 0.1, mgp = c(4, 1, 0), cex = 1.5, cex.axis = 1.1, cex.lab = 1.25, las = 1, cex.main = 1.5)
plot(x, y, main = "Random plot")

par(Op)                   # Change back to default settings

When we want to have two plots side-by-side, we can adjust the mfrow parameter.

par(mfrow = c(1, 2))
plot(x, y)
plot(x, sqrt(x))

par(Op)

Alternatively, we can make use of the layout function.

m <- rbind(c(1, 1, 2, 2),
           c(4, 3, 3, 5))
layout(m)
layout.show(5)

plot(x, y)
plot(x, sqrt(x))
plot(x, log(x))
layout(1)

5.1.3 Exporting the scatterplot

It is often very useful to directly export your customized graph. Certainly when submitting the plot for a paper, we need to be able to export it to a file that they accept. Some of the basic formats that (base) R supports are BMP, JPEG, PNG, TIFF, PDF and PS files.

?png
?grDevices

We can, for example, save a plot as pdf using the following code.

path <- tempdir()
graph.path <- file.path(path, "myfile.pdf")
pdf(graph.path, height = 5, width = 6)
    plot(log(citeprice)~log(subs), data = Journals, pch = 19, col = "blue", xlim =   
           c(0, 8), ylim = c(-7, 4),
    main = "Library subscriptions")
    rug(log(Journals$subs))
    rug(log(Journals$citeprice),side=2)
    journal <- "Econometrica"
  journal_info <- subset(Journals, title==journal)
  x.val <- log(journal_info$subs)
  y.val <- log(journal_info$citeprice)
  text(x.val, y.val, journal, pos=2)
dev.off()

To save the plot as a PNG file, we use the png function. In this function, we can adjust the resolution of the plot by adjusting the res argument.

graph.path <- tempfile(fileext = ".png")
png(graph.path, height = 7.5e2 * 3, width = 3e3, res = 300)
par(cex.axis = 1.25, cex.lab = 1.5, cex.main = 1.75, cex = 1.5)
    plot(log(citeprice)~log(subs), data = Journals, pch = 19, col = "blue", xlim =   
           c(0, 8), ylim = c(-7, 4),
    main = "Library subscriptions")
    rug(log(Journals$subs))
    rug(log(Journals$citeprice),side=2)
    journal <- "Econometrica"
  journal_info <- subset(Journals, title==journal)
  x.val <- log(journal_info$subs)
  y.val <- log(journal_info$citeprice)
  text(x.val, y.val, journal, pos=2)
dev.off()
opendir(graph.path)

5.1.4 The curve() function

This function draws a curve corresponding to a function over the interval [from, to].

curve(dnorm, from = -5, to = 5, col = "slategray", lwd = 3, main = "Density of the standard normal distribution")
text(-5, 0.3, expression(f(x) == frac(1, sigma ~~ sqrt(2*pi)) ~~ e^{-frac((x - mu)^2, 2*sigma^2)}), adj = 0)

?plotmath

5.2 More fancy plots

R has many dedicated packages for advanced plotting. You will work with two of them in this Section.

5.2.1 Creating graphics with ggplot2

ggplot2 is a package created and maintained by prof. Hadley Wickham, it’s aim is to Create Elegant Data Visualisations Using the Grammar of Graphics. Here is the basic explanation of how ggplot2 works from (Grolemund and Wickham 2016).

With ggplot2, you begin a plot with the function ggplot(). ggplot() creates a coordinate system that you can add layers to. The first argument of ggplot() is the dataset to use in the graph. So ggplot(data = mpg) or ggplot(mpg) creates an empty graph.

You complete your graph by adding one or more layers to ggplot(). The function geom_point() adds a layer of points to your plot, which creates a scatterplot. ggplot2 comes with many geom functions that each add a different type of layer to a plot.

Each geom function in ggplot2 takes a mapping argument. This defines how variables in your dataset are mapped to visual properties. The mapping argument is always paired with aes(), and the \(x\) and \(y\) arguments of aes() specify which variables to map to the \(x\) and \(y\) axes. ggplot2 looks for the mapped variable in the data argument, in this case, mpg.

library(ggplot2)

# use default theme
ggplot(data = mtcars, mapping = aes(x = hp, y = mpg)) +
  geom_point(shape=1, alpha = 1/2)+
  geom_smooth() 
`geom_smooth()` using method = 'loess' and formula 'y ~ x'

# shorter
ggplot(mtcars, aes(x = hp, y = mpg)) +
  geom_point(shape=1, alpha = 1/2)+
  geom_smooth() 
`geom_smooth()` using method = 'loess' and formula 'y ~ x'

# use black and white lay-out
ggplot(mtcars, aes(x = hp, y = mpg)) + theme_bw() +
  geom_point(shape=1, alpha = 1/2)+ 
  geom_smooth() 
`geom_smooth()` using method = 'loess' and formula 'y ~ x'

You can add a third variable to a two dimensional scatterplot by mapping it to an aesthetic. An aesthetic is a visual property of the objects in your plot. Aesthetics include things like the size, the shape, or the color of your points. You can display a point in different ways by changing the values of its aesthetic properties.

ggplot(mtcars, aes(x = hp, y = mpg))+
  geom_point(mapping = aes(color = gear))

Or you could have mapped this variable to the alpha aesthetic, which controls the transparency of the points, or the shape of the points.

ggplot(mtcars, aes(x = hp, y = mpg))+
  geom_point(mapping = aes(alpha = gear))

ggplot(mtcars, aes(x = hp, y = mpg))+
  geom_point(mapping = aes(size = gear))

You’ll now construct a boxplot of mpg per cyl using ggplot().

ggplot(mtcars, aes(factor(cyl), mpg))+
  geom_boxplot() + geom_jitter() + theme_bw()

Another way to code the same example

p <- ggplot(mtcars, aes(factor(cyl), mpg))
p + geom_boxplot() + geom_jitter() + theme_bw()

5.2.2 Fancy correlation plots

You use the package corrplot to visualize correlations between variables. For more examples, see corrplot.

library(corrplot)
corrplot 0.84 loaded
# get correlation matrix
M <- cor(mtcars)
str(M)
 num [1:11, 1:11] 1 -0.852 -0.848 -0.776 0.681 ...
 - attr(*, "dimnames")=List of 2
  ..$ : chr [1:11] "mpg" "cyl" "disp" "hp" ...
  ..$ : chr [1:11] "mpg" "cyl" "disp" "hp" ...
M
          mpg      cyl     disp       hp      drat       wt      qsec       vs
mpg   1.00000 -0.85216 -0.84755 -0.77617  0.681172 -0.86766  0.418684  0.66404
cyl  -0.85216  1.00000  0.90203  0.83245 -0.699938  0.78250 -0.591242 -0.81081
disp -0.84755  0.90203  1.00000  0.79095 -0.710214  0.88798 -0.433698 -0.71042
hp   -0.77617  0.83245  0.79095  1.00000 -0.448759  0.65875 -0.708223 -0.72310
drat  0.68117 -0.69994 -0.71021 -0.44876  1.000000 -0.71244  0.091205  0.44028
wt   -0.86766  0.78250  0.88798  0.65875 -0.712441  1.00000 -0.174716 -0.55492
qsec  0.41868 -0.59124 -0.43370 -0.70822  0.091205 -0.17472  1.000000  0.74454
vs    0.66404 -0.81081 -0.71042 -0.72310  0.440278 -0.55492  0.744535  1.00000
am    0.59983 -0.52261 -0.59123 -0.24320  0.712711 -0.69250 -0.229861  0.16835
gear  0.48028 -0.49269 -0.55557 -0.12570  0.699610 -0.58329 -0.212682  0.20602
carb -0.55093  0.52699  0.39498  0.74981 -0.090790  0.42761 -0.656249 -0.56961
            am     gear      carb
mpg   0.599832  0.48028 -0.550925
cyl  -0.522607 -0.49269  0.526988
disp -0.591227 -0.55557  0.394977
hp   -0.243204 -0.12570  0.749812
drat  0.712711  0.69961 -0.090790
wt   -0.692495 -0.58329  0.427606
qsec -0.229861 -0.21268 -0.656249
vs    0.168345  0.20602 -0.569607
am    1.000000  0.79406  0.057534
gear  0.794059  1.00000  0.274073
carb  0.057534  0.27407  1.000000
# visualize the correlation structure
corrplot(M, method = "circle")

corrplot(M, method = "square")

corrplot(M, method = "color")

corrplot(M, type = "upper")

corrplot(M, type = "upper", method = "square")

5.3 Exercises

Learning check

  1. Use the Danish fire insurance losses. Plot the arrival of losses over time. Use type= "l" for a line plot, label the \(x\) and \(y\)-axis, and give the plot a title using main.

  2. Do the same with instructions from ggplot2. Get inspiration from R for Data Science and use geom_line() to create the line plot.

  3. Use the data set ‘car_price.csv’ available in the documentation.

  • Import the data in R.
  • Explore the data.
  • Make a scatterplot of price versus income, use basic plotting instructions and use ggplot2.
  • Add a smooth line to each of the plots (using lines to add a line to an existing plot and lowess to do scatterplot smoothing and using geom_smooth in the ggplot2 grammar).
  1. Use the mpg data set. Work through the following steps. The data contains observations collected by the US Environment Protection Agency on 38 models of car.
  • Explore the data.
  • Plot displ, a car’s engine size, in litres on the \(x\)-axis and hwy, on the \(y\)-axis, that is the car’s fuel efficiency on the highway, in miles per gallon (mpg).
  • Now do the same but use different colors for the points, based on the class variable in mpg. Add a smooth line.