Data visualisation with ggplot2

Data visualization

• Important aspect in data sciences -> Communicate information clearly and efficiently to the community.

• Powerful tool to discovers patterns in the data.

• It makes complex data more accessible -> reveal data.

• Bad graphics can be a reason for paper rejection!

• A picture is worth a thousand words.
  • Always, always, always plot the data!
  • When possible, replace tables with figures that are more compelling.

What is a good graph?

Data-ink ratio

The data-ink ratio is the proportion of ink that is used to present actual data compared to the total amount of ink used in the entire display.

\[ \verb|data-ink ratio| = \frac{\verb|Data-ink|}{\verb|Total ink used to print the graphic|} \]

The data-to-ink ratio should be keep as high as possible.

Examples

Examples

How to lie with graphs

It is easy to exaggerate effects or distort the reality with graphs.

Examples

Examples

Do not use 3D, ever!

3D graphics are very rarely useful:

1. First, they break the data-to-ink ratio rule.

2. Secondly, they can distort the reality.

Examples

What is the value of C?

Examples

The value of C is 3…

Examples

What is the value of y at z = low and x = t1?

Top 3 of bad graphs

ggplot2

ggplot2

• ggplot2 is an extremely powerful package based on the grammar of graphics to produce complicated graphics in an elegant manner.

• ggplot2 works best when you have tidy data.

• Graphics are built by combining layers.

• You can refer to the printed cheat sheet for an overview of the package’s functions.

ggplot2

ggplot2 is not part of base R, so it needs to be installed.

install.packages("ggplot2") # Install the package

library(ggplot2) # Load the package

The data

For the following examples we are going to use the data from the mpg dataset. This dataset contains a subset of the fuel economy data that the EPA makes available on http://fueleconomy.gov.

Basic structure

geoms

geoms is the short for geometric objects which are used to specify which type of graphic you want to produce (boxplot, barplot, scatter, …).

##  [1] "geom_abline"     "geom_area"       "geom_bar"       
##  [4] "geom_bin2d"      "geom_blank"      "geom_boxplot"   
##  [7] "geom_col"        "geom_contour"    "geom_count"     
## [10] "geom_crossbar"   "geom_curve"      "geom_density"   
## [13] "geom_density_2d" "geom_density2d"  "geom_dotplot"   
## [16] "geom_errorbar"   "geom_errorbarh"  "geom_freqpoly"  
## [19] "geom_hex"        "geom_histogram"  "geom_hline"     
## [22] "geom_jitter"     "geom_label"      "geom_line"      
## [25] "geom_linerange"  "geom_map"        "geom_path"      
## [28] "geom_point"      "geom_pointrange" "geom_polygon"   
## [31] "geom_qq"         "geom_qq_line"    "geom_quantile"  
## [34] "geom_raster"     "geom_rect"       "geom_ribbon"    
## [37] "geom_rug"        "geom_segment"    "geom_sf"        
## [40] "geom_sf_label"   "geom_sf_text"    "geom_smooth"    
## [43] "geom_spoke"      "geom_step"       "geom_text"      
## [46] "geom_tile"       "geom_violin"     "geom_vline"

One variable graphics

There are two main types of one variable graphics:

Histogram

# First create the mapping.
p <- ggplot(data = mpg, mapping = aes(x = cty))

# Then add one histogram layer (geom) with no parameters.
p + geom_histogram()

Histogram

# Specify a binwidth of 5 miles per gallon.
p + geom_histogram(binwidth = 5)

Barplot

ggplot(data = mpg, mapping = aes(x = class)) + 
  geom_bar()

Two variables graphics

Two variables graphics are more commmon.

Scatter plot

# First, create the mapping
p <- ggplot(mpg, aes(x = displ, y = hwy))

# Then, tell ggplot2 we want points for the relation between x and y.
p + geom_point()

Line plot

# First, create the mapping
p <- ggplot(mpg, aes(x = displ, y = hwy))

# Then, tell ggplot2 we want lines for the relation between x and y.
p + geom_line()

Boxplot

In descriptive statistics, a box plot or boxplot is a convenient way of graphically depicting groups of numerical data through their quartiles (Wikipedia).

To make a boxplot, we need to have a discrete/categorical variable on \(x\) and a continuous variable on \(y\).

Boxplot

ggplot(data = mpg, mapping = aes(x = class, y = cty)) + 
  geom_boxplot() # boxplot geom with no parameters

Working with colors

It can be useful to add colors in graphics. To change the colors of the points, we have to use the color parameter.

ggplot(mpg, aes(x = displ, y = hwy)) + 
  geom_point(color = "red")

Working with colors

If we want to set a color based on a variable, we have to use the aesthetic: aes(colour = variable).

ggplot(mpg, aes(x = displ, y = hwy)) + 
  geom_point(aes(color = class))

Working with size

In the same manner, the size of the dots can be based on a particular variable.

ggplot(mpg, aes(x = displ, y = hwy)) + 
  geom_point(aes(size = factor(cyl)))

Faceting

Faceting

Faceting is a very powerful of the ggplot2 library which allows to display additional categorical variables in facets.

There are two types of faceting: facet_grid() and facet_wrap().

1D facets

ggplot(mpg, aes(x = displ, y = cty)) + # Mapping
  geom_point() + # We want a scatter plot
  facet_wrap(~class) # We want 1 panel for each class

1D facets

ggplot(mpg, aes(x = displ, y = cty)) + # Mapping
  geom_point() + # We want a scatter plot
  facet_wrap(~class, scale = "free") # We want 1 panel for each class

2D facets

2D facet graphics are made using the facet_grid() function.

ggplot(mpg, aes(x = displ, y = hwy)) +
  geom_point() +
  facet_grid(cyl ~ class) # number of cyl as a function of class

Graphics appearance

Graphics appearance

There are a ton of options to modify the look and feel of your graphics and we can not go through them all in short period of time.

Book: ggplot2: Elegant Graphics for Data Analysis (Use R!)

Here I present the principal functions I usually use to make publication-ready graphics.

Graphics appearance

ggplot(mpg, aes(x = displ, y = cty)) + 
  geom_point(colour = factor(cyl)) + # Scatterplot color based on cyl
  
  stat_smooth(method = "lm") + # Add a linear smoother to the data
  
  labs(colour = "Number of\ncylinders") + # Title of the color legend
  
  xlab("Horsepower") + # Change x-axis title
  
  ylab("Miles per gallon") + # Change y-axis title
  
  ggtitle("This is my title") + # Add a title on top of the plot
  
  theme(legend.position = "top") + # Change legend position
  
  xlim(0, 8) # Change limits of x-axis

Saving your graphic

Saving your pretty ggplot2 graphics is pretty easy with the ggsave() function.

p <- ggplot(mpg, aes(x = displ, y = cty)) +
  geom_point()

# Vector formats
ggsave("path/to/myfile.pdf", p, width = 5.97, height = 4.79)
ggsave("path/to/myfile.eps", p, width = 5.97, height = 4.79)
ggsave("path/to/myfile.ps", p, width = 5.97, height = 4.79)

# Raster formats
ggsave("path/to/myfile.jpg", p, width = 5.97, height = 4.79)
ggsave("path/to/myfile.tiff", p, width = 5.97, height = 4.79)
ggsave("path/to/myfile.png", p, width = 5.97, height = 4.79)