Visualization¶
Make data legible. Make insight obvious.
Pages¶
- Chart Selection — pick the right chart for the question
- Design Principles — 9 principles, McCandless method, storytelling
- Dashboard Architecture — top-down layouts, "F" pattern, interactivity
- Tools — Tableau, Metabase, alternatives
Foundation¶
- Five-second rule — a viz should be clear, effective, convincing in ≤ 5 seconds. Don't overcomplicate.
- Focus — narrow down the data so it's not misleading
- Feedback loop — show drafts to your audience to improve
- Make it understandable — the audience should grasp your point
- Avoid misleading viz — scaling and proportion can deceive
- Self-question — which chart makes it easiest to make the point?
- Dashboards have a downside — viewers create their own narrative; could misinterpret
- Accessibility — color-blind, low-vision, mobile readers
Quick rules of thumb¶
| Goal | Chart |
|---|---|
| Compare values | Bar |
| Trend over time | Line |
| Part of whole | 100% stacked bar (avoid pie if > 4 slices) |
| Distribution | Histogram, box plot |
| Relationship | Scatter |
| Geographic | Map |
| Density / ranking | Heatmap |
References¶
- From Data to Viz — interactive chart decision tree
- Storytelling with Data — Cole Knaflic
- Information is Beautiful
- Data Viz Catalogue
- Edward Tufte — Visual Display of Quantitative Information