Session 4 — Visualisation and basic statistics with ggplot2

Language Analytics in R

Author

Peter Gilles

0.1 Learning objectives

By the end of this session you will:

  • Build plots with ggplot2: data, aesthetics (aes()), and geoms.
  • Use layers: add geoms, scales, labels, and themes.
  • Choose appropriate geoms for the task (e.g. bar, point, line, boxplot, histogram).
  • Add basic statistical summaries to plots (e.g. means, error bars, smooth trends).
  • Customize axes, labels, and themes for clear, publishable figures.
  • Apply ggplot2 to language and corpus data (e.g. frequencies, distributions, model results).

1 Grammar of graphics

  • Data → Aesthetics (mapping variables to x, y, color, fill, etc.) → Geometries (how to draw).
  • ggplot(data, aes(x = ..., y = ...)) + geom_...()

[Add a minimal example: e.g. bar chart of word counts or scatter plot.]

2 Core geoms

  • geom_bar() / geom_col() — counts vs pre-computed values.
  • geom_point() — scatter plots.
  • geom_line() — time series or trends.
  • geom_histogram() / geom_density() — distributions.
  • geom_boxplot() / geom_violin() — group comparisons.

[Add one or two code chunks per geom using course-relevant data.]

3 Aesthetics and layers

  • aes(): map variables to x, y, color, fill, size, shape.
  • Faceting: facet_wrap(), facet_grid().
  • Scales: scale_* for axes, colors, labels.

[Add an example with faceting or custom scales.]

4 Labels and themes

  • labs() — title, subtitle, axis labels.
  • theme_minimal(), theme_bw(), or custom theme().
  • Saving: ggsave() or Quarto figure options.

[Add a polished plot example and ggsave.]

5 Adding basic statistics to plots

  • stat_summary() for means and confidence intervals/error bars.
  • geom_smooth() for trend lines (e.g. linear model or loess).
  • Combine group summaries with faceting to compare patterns across categories.

[Add one example with means + error bars and one example with geom_smooth().]

6 Hands-on exercises

  • Reproduce a simple plot from the session (e.g. bar chart, boxplot).
  • Build a plot from your own or course data; add labels and a theme.

7 Summary and further reading