5.3 Advanced plotting with ggplot2

ggplot2 takes data visualization to the next level with its advanced plotting capabilities. Building on the grammar of graphics, it offers a flexible framework for creating complex, multi-layered plots that reveal deeper insights into your data.

From distribution plots to heatmaps, ggplot2 provides a wide array of tools to visualize data in meaningful ways. You'll learn to customize aesthetics, combine multiple layers, and create multi-panel plots, empowering you to tell compelling data stories.

Grammar of Graphics and ggplot2 Structure

Fundamentals of the Grammar of Graphics

• The grammar of graphics is a framework for creating statistical graphics that separates the components of a plot into layers, scales, and coordinate systems
• This framework provides a structured and systematic approach to building complex visualizations
• The grammar of graphics allows for the creation of a wide range of plots by combining different components in a modular fashion
• Key components of the grammar of graphics include data, aesthetics (visual properties), geometric objects (points, lines, bars), statistical transformations, scales, and coordinate systems

ggplot2 Implementation and Syntax

• ggplot2 is an implementation of the grammar of graphics in R, providing a powerful and flexible tool for data visualization
• The basic structure of ggplot2 code includes the ggplot() function, which initializes the plot, followed by layers defined using the + operator
• Layers in ggplot2 include geom_ functions that specify the type of plot (geom_point(), geom_line(), geom_bar()), aes() for mapping variables to plot aesthetics, and stat_ functions for statistical transformations
• Scales control the mapping between data values and visual properties, such as color, size, and shape, and are automatically generated or can be customized using scale_ functions
• Coordinate systems (coord_cartesian(), coord_polar()) define the mapping between data coordinates and the 2D plane of the plot, allowing for transformations like zooming, panning, or polar coordinates

Advanced Plots with ggplot2

Visualizing Distributions

• Boxplots (geom_boxplot()) display the distribution of a continuous variable, showing the median, interquartile range (IQR), and potential outliers
  • They are useful for comparing the distribution of a variable across different categories or groups
  • Outliers are typically defined as values that fall below Q1 - 1.5 * IQR or above Q3 + 1.5 * IQR, where Q1 and Q3 are the first and third quartiles, respectively
• Violin plots (geom_violin()) combine the information of a boxplot with a kernel density plot, displaying the probability density of the data at different values
  • They provide a more detailed view of the distribution shape compared to boxplots
  • The width of the violin at each point represents the density of observations at that value
• Density plots (geom_density()) show the distribution of a continuous variable by estimating the probability density function
  • They are a smooth alternative to histograms and can be used to compare the distribution of multiple groups or variables
  • The area under the density curve represents the probability of observing a value within a given range

Heatmaps and Ridgeline Plots

• Heatmaps (geom_tile() or geom_raster()) are used to visualize 2D data, where the color of each cell represents the value of a variable
  • They are commonly used to display relationships between two variables or to show patterns in large datasets
  • The color scale can be customized using scale_fill_gradient() or scale_fill_viridis_c() to choose appropriate color palettes
• Ridgeline plots (geom_density_ridges() from the ggridges package) are useful for comparing the distribution of a variable across multiple categories
  • They display the density curves of each category as overlapping ridges, making it easier to compare the shapes and positions of the distributions
  • Ridgeline plots are a space-efficient alternative to faceting when there are many categories to compare

Combining Multiple Layers

• ggplot2 allows for the creation of complex and informative plots by combining multiple geom_ layers
• Examples of combining layers include:
  • Adding points (geom_point()) or lines (geom_line()) to a boxplot or violin plot to show individual observations or trends
  • Overlaying a density plot (geom_density()) on top of a histogram (geom_histogram()) to show the distribution shape and individual bins
  • Combining a scatterplot (geom_point()) with a smoothed line (geom_smooth()) to visualize the relationship between two variables and the overall trend

Customizing ggplot2 Aesthetics

Built-in Themes and Theme Customization

• ggplot2 provides a wide range of functions to customize plot aesthetics, including colors, fonts, axis labels, and legends
• Theme functions, such as theme_bw(), theme_minimal(), and theme_classic(), control the overall appearance of the plot, including background color, gridlines, and axis formatting
  • These built-in themes provide a quick way to change the look and feel of a plot
  • For example, theme_bw() creates a plot with a white background and gray gridlines, while theme_minimal() removes most of the background elements for a cleaner look
• The theme() function allows for fine-grained control over individual plot elements, such as axis titles, legend positions, and plot margins
  • Each plot element can be customized by providing the appropriate argument within theme(), such as axis.title, legend.position, or plot.margin
  • For example, theme(axis.title = element_text(size = 14, face = "bold")) sets the font size and weight of the axis titles

Scales and Color Palettes

• Scales in ggplot2 control the mapping between data values and visual properties, such as colors, sizes, and shapes
• Scales can be customized using scale_ functions, which allow for fine-tuning of the appearance of plot elements
  • scale_color_manual() and scale_fill_manual() are used to set specific colors for discrete variables
  • scale_color_gradient() and scale_fill_gradient() create color gradients for continuous variables
  • scale_x_continuous() and scale_y_continuous() control the appearance of the x and y axes, including breaks, labels, and limits
• ggplot2 also provides several built-in color palettes that can be used to create visually appealing plots
  • scale_color_brewer() and scale_fill_brewer() use color palettes from the ColorBrewer library, which are designed for different types of data and color-vision deficiencies
  • scale_color_viridis_c() and scale_fill_viridis_c() use the viridis color palette, which is perceptually uniform and colorblind-friendly

Extensions and Add-on Packages

• Several extensions and add-on packages are available for ggplot2, providing additional functionality and pre-defined themes for enhancing plot aesthetics
• The ggthemes package offers a collection of pre-defined themes that mimic the styles of various publications and visualization tools, such as theme_economist() or theme_fivethirtyeight()
• The ggplot2 extensions package (ggplot2.ext) includes additional geoms, scales, and themes that extend the capabilities of ggplot2, such as geom_split_violin() for split violin plots or scale_color_material() for material design colors
• The gganimate package allows for the creation of animated plots by specifying how plot elements should change over time or across different categories, using functions like transition_states() or transition_reveal()

Multi-panel Plots with ggplot2

Faceting with facet_wrap() and facet_grid()

• Multi-panel plots (facets) are used to display subsets of data in separate panels based on one or more categorical variables
• The facet_wrap() function creates a grid of panels based on a single categorical variable, wrapping the panels into multiple rows if necessary
  • The formula argument in facet_wrap() is used to specify the variable to facet by, such as ~ variable for a single variable or ~ variable1 + variable2 for the interaction of two variables
  • The nrow and ncol arguments control the number of rows and columns in the facet grid
• facet_grid() creates a grid of panels based on two categorical variables, with one variable represented by rows and the other by columns
  • The formula argument in facet_grid() is used to specify the row and column variables, separated by a tilde (~), such as variable1 ~ variable2
  • If a dot (.) is used instead of a variable name on either side of the tilde, the facet grid will only split the panels by the specified variable

Customizing Facet Appearance

• Faceting allows for the comparison of patterns and relationships across different subgroups or categories within the data
• The scales argument in facet_ functions controls whether the scales are fixed (scales = "fixed") or allowed to vary independently (scales = "free") across the panels
  • scales = "free_x" and scales = "free_y" allow the x and y scales to vary independently, respectively
  • By default, the scales are fixed, meaning that the axis limits and breaks are the same across all panels
• Customization of facet labels, strip backgrounds, and spacing can be done using the labeller, strip.background, and panel.spacing arguments within facet_ functions
  • The labeller argument accepts functions that modify the facet labels, such as label_both for displaying both variable name and value, or custom labeller functions
  • strip.background and panel.spacing control the appearance of the facet strip (the area containing the facet labels) and the spacing between panels, respectively

Example Use Cases for Faceting

• Faceting is particularly useful when exploring relationships between variables across different subgroups or categories
• Some common use cases for faceting include:
  • Comparing the distribution of a variable (e.g., income) across different levels of another variable (e.g., education level) using facet_wrap(~ education)
  • Examining the relationship between two variables (e.g., height and weight) for different subgroups (e.g., gender) using facet_grid(gender ~ .)
  • Visualizing time series data for multiple entities (e.g., stock prices for different companies) using facet_wrap(~ company, nrow = 2)
  • Displaying geographic data (e.g., unemployment rates) for different regions or countries using facet_grid(rows = vars(region), cols = vars(year))