The course, module by module
Foundations
Why visualisation works, and which chart for which message
Introduction to Visualization
Perception, Gestalt, data types, the seven stages, boxplot anatomy, correlation and cause.
12 figures · encoding channels, Anscombe, boxplot builder, confounder demo Chapter 2 · FoundationsTaxonomy of Visualisation Methods
The seven core visuals, form follows function, and the five communication purposes.
15 figures · waterfall, sankey, treemap, stream graph, choropleth trapDesign
Cognitive load, clutter, memory, encoding accuracy, colour
Effective Visuals Design
Clutter, the decluttering sequence, memory, pre-attentive attributes, colour, the Z pattern.
9 figures · timed counting test, colour-blindness simulator, 3D bar failure Chapter 4 · DesignVisual Best Practices
Measurement scales, Cleveland's operations, the encoding ranking, and visualising text.
10 figures · guess-the-ratio test, text pipeline, topic modelDashboards and storytelling
Who it is for, and how to make them act on it
Dashboards
The four types, the thirteen mistakes, the data-pixel ratio, usability and context.
9 figures · a live cross-filtering dashboard, the five second rule, timed Chapter 6 · StorytellingVisual Storytelling
Frameworks, narrative types, the seven story types, and every way a chart can mislead.
15 figures · the cholera map, Simpson's paradox, cherry picking sliderPython
Matplotlib, Seaborn, Bokeh
Matplotlib
Figure and Axes, the workflow, colour, colormaps, spines, plot types, annotation, styles.
12 figures · code beside every chart, viridis against jet, bins slider Chapter 8 · PythonMatplotlib in the Real World
Reading files, time series, subplots, GeoPandas, curve fitting and extrapolation.
7 figures · the subplots index grid, random walk, polyfit degree Chapter 9 · PythonSeaborn
Figure-level and axes-level, semantic mappings, faceting, the categorical family, regression.
10 figures · the whole catplot family, Anscombe with order=2 Chapter 10 · PythonBokeh
Glyphs, the free toolbar, clickable legends, ColumnDataSource, hover, widgets, layouts.
10 figures · a working pan and box-zoom toolbar you can dragBI tools
Four of the six sample-paper questions live here
Tableau
The workspace, dimensions and measures, live against extract, metadata, joins and blending.
8 figures · build a view by dropping pills, the four join types Chapter 12 · BI toolsTableau: Analysis and Dashboards
The analysis cycle, the moves that matter, aggregation traps, layering, dashboard layout.
9 figures · SUM against AVG on a percentage, COUNTD, five channels Chapter 13 · BI toolsPower BI and Google Data Studio
Connectors, the canvas, DAX measures, tree maps as filters, quadrants, and filter scopes.
10 figures · a live Power BI build, and the fixed-axis trap breakingHow the guide is built
Five ideas that come back in every chapter
| Idea | First appears | Returns as |
|---|---|---|
| Audience-centric, not data-centric | Chapter 1, context | Dashboard type (5), narrative type (6), user experience in every tool week |
| Encoding accuracy | Chapter 4, the ranking | Why hover tooltips exist (10), why bubbles need labels (12) |
| Ordering and decluttering | Weeks 1 and 3 | Seaborn's automatic ordering (9), removing axes in Power BI (13) |
| Correlation is not causation | Chapter 1 | Confounders (4), misleading charts (6) |
| Right aggregation for the data type | Chapter 4, measurement scales | Summing a percentage is wrong (11, 12, 13) |
How to use these pages
The chapters build on each other, and each one stands alone. Every page ends with its key points, and cross-references name the chapter they point to, so you can start wherever your question is.
Every segmented control, slider, toggle and clickable mark changes what is drawn. Where a figure has a control bar, the point of the figure is usually the difference between two of its states.
Chapters 7 to 13 put the code or the interface beside the result, so you can read the call that produced a chart and change it in your head before you change it in an editor.
The button in the header follows your system by default and remembers your choice. Every chart redraws with the new palette rather than being inverted.