Chapter 4 Graphics with base R
Summarising your data, either numerically or graphically, is an important (if often overlooked) component of any data analysis. Fortunately, R has excellent graphics capabilities and can be used whether you want to produce plots for initial data exploration, model validation or highly complex publication quality figures. There are three main systems for producing graphics in R; base R graphics, lattice graphics and ggplot2. Each of these systems have their strengths and weaknesses and we often use them interchangeably. In this Chapter we’ll focus mostly on base R graphics with a sprinkling of lattice graphics for added variety. In the next Chapter we’ll introduce you to the
The base R graphics system is the original plotting system that’s been around (and has evolved) since the first days of R. When creating plots with base R we tend to use high level functions (like the
plot() function) to first create our plot and then use one or more low level functions (like
text() etc) to add additional information to these plots. This can seem a little weird (and time consuming) when you first start creating fancy plots in R, but it does allow you to customise almost every aspect of your plot and build complexity up in layers. The flip side to this flexibility is that you’ll often need to make many decisions about how you want your plot to look rather than rely on the software to make these decisions for you. Having said that, it’s generally very quick and easy to generate simple exploratory plots with base R graphics.
The lattice system is implemented in the
lattice() package that comes pre-installed with the standard installation of R. However, it won’t be loaded by default so you’ll first need to use
library(lattice) to access all the plotting functions. Unlike base R graphics, lattice plots are mostly generated all in one go using a single function so there’s no need to use high and low level plotting functions to customise the look of a plot. This can be a real advantage as things like margin sizes and plot spacing are adjusted automatically. Lattice plots also make a few more decisions for you about how the plots will look but this comes with a slight cost as customising lattice plots to get them to look exactly how you want can become quite involved. Where lattice plots really shine is plotting complex multi-dimensional data using panel plots (also called trellis plots). We’ll see a couple of examples of these types of plots later in the Chapter.