CS448B Visualization
When to Use Each Reference
| Reference | Use When |
|---|---|
| encoding-perception.md | Choosing how to map data to visual properties, or evaluating if encodings are effective |
| chart-design.md | Deciding which chart type fits the data, or configuring axes/scales |
| d3-patterns.md | Writing D3.js code for bindings, scales, axes, transitions |
| interaction-animation.md | Adding brushing, filtering, tooltips, or animated transitions |
| color.md | Selecting color palettes or ensuring accessibility |
| networks-text.md | Visualizing graphs, hierarchies, or text/document data |
Critique Checklist
When reviewing any visualization:
- Expressiveness - Does it show all the data? Only the data? No misleading elements?
- Effectiveness - Is the most important data on the most accurate encoding (position > length > area > color)?
- Zero baseline - Do bar charts start at zero?
- Accessibility - Works for colorblind viewers (~8% of males)?
- Data-ink ratio - Can any non-data elements be removed?
- Aspect ratio - Are line charts banked so slopes are ~45°?
Encoding Decision Order
When mapping data fields to visual channels:
- Most important quantitative → Position (x or y)
- Second quantitative → Position (other axis) or Length
- Categories (≤7) → Color hue
- Categories (>7) → Position or small multiples
- Magnitude/importance → Size (but expect ~30% underestimation)
Chart Selection Logic
- One variable, distribution → Histogram
- One variable, categories → Bar chart
- Two variables, both quantitative → Scatterplot
- Two variables, time + quantitative → Line chart
- Two variables, both categorical → Heatmap
- Hierarchy → Treemap or node-link tree
- Network (sparse) → Force-directed layout
- Network (dense) → Matrix diagram
Animation Decision
- Presentation context → Use animation (faster, engaging)
- Analysis context → Use small multiples (more accurate)
- State transitions → Animate to maintain object constancy
- Duration: 200-300ms quick feedback, 500-700ms standard, 1000ms+ complex