Part VI: Specialized Domains

The first five parts of this book gave you a complete visualization toolkit: perception science, design principles, and four Python libraries spanning static and interactive output. Part VI puts that toolkit to work on data types that demand specialized treatment.

Not all data is a tidy table of numbers and categories. Time series data has autocorrelation, seasonality, and trend components that require specific visual forms to reveal. Text data lives in high-dimensional spaces that must be projected, clustered, and summarized before any chart can represent it. Scientific data carries uncertainty, error bounds, and reproducibility requirements that standard business charts ignore. And big data, truly large datasets with millions of rows, breaks the rendering assumptions that every plotting library makes by default. Each of these domains has its own visual vocabulary, its own pitfalls, and its own solutions.

The four chapters each address one of these domains:

  • Chapter 25: Time Series Visualization covers the techniques for showing trends, seasonality, and change over time, including rolling averages, decomposition plots, event markers, and the particular challenges of irregular and multi-scale temporal data.
  • Chapter 26: Text and NLP Visualization addresses how to make language visible, from word clouds and frequency distributions to topic model visualizations, sentiment timelines, and embedding projections.
  • Chapter 27: Statistical and Scientific Visualization focuses on publication-ready figures for research, including error bars with proper confidence intervals, regression diagnostics, multi-panel journal figures, and the formatting standards that peer reviewers expect.
  • Chapter 28: Big Data Visualization confronts what happens when your dataset is too large for a scatter plot: sampling strategies, density estimation, hexbin plots, datashader, and the architectural patterns that keep rendering fast when row counts climb into the millions.

These chapters draw on everything that came before. You will use matplotlib for precise scientific figures, seaborn for statistical summaries, Plotly for interactive time series exploration, and Altair for declarative approaches to text data. The choice of library in each situation is deliberate and explained, reinforcing the idea that the right tool depends on the task.

The progressive project thread continues through Part VI as well. The climate dataset, now a familiar companion, receives a proper time series treatment: trend decomposition, rolling means, and change-point detection rendered as publication-quality figures.

Part VII takes the visualizations you have built across all these domains and puts them into production: dashboards, automated reports, branded themes, and end-to-end workflows.

Chapters in This Part