Datavis

Data Visualization Skill

Create beautiful, mathematically elegant, emotionally resonant data visualizations.

Philosophy: "Life is Beautiful"

Every visualization should:

1. Reveal truth through data
2. Evoke wonder through design
3. Respect the viewer through accessibility
4. Honor complexity through elegant simplification

Core Capabilities

1. Visual Encoding

Scale Selection:

Perceptual Honesty - Area scales with square of radius, so use sqrt:

// WRONG: Linear radius exaggerates large values
const badScale = d3.scaleLinear().domain([0, max]).range([0, maxRadius]);

// RIGHT: Sqrt maintains perceptual accuracy
const goodScale = d3.scaleSqrt().domain([0, max]).range([0, maxRadius]);

2. Color Design

Palette Types:

• Categorical - Distinct hues for nominal data (max 8)
• Sequential - Single hue gradient for ordered data
• Diverging - Two hues meeting at meaningful midpoint

Colorblind-Safe Palette (8 colors):

const colorblindSafe = [
  '#332288', '#117733', '#44AA99', '#88CCEE',
  '#DDCC77', '#CC6677', '#AA4499', '#882255'
];

Always use redundant encoding - don't rely on color alone:

node.attr('fill', d => colorScale(d.category))
    .attr('d', d => symbolScale(d.category)); // Shape too!

3. D3.js Patterns

Force Simulation:

const simulation = d3.forceSimulation(nodes)
  .force('charge', d3.forceManyBody().strength(-300))
  .force('link', d3.forceLink(links).id(d => d.id))
  .force('center', d3.forceCenter(width/2, height/2))
  .force('collision', d3.forceCollide().radius(d => d.r + 2));

Responsive SVG:

const svg = d3.select('#chart')
  .append('svg')
  .attr('viewBox', `0 0 ${width} ${height}`)
  .attr('preserveAspectRatio', 'xMidYMid meet');

Touch-Friendly (44x44px minimum):

node.append('circle')
  .attr('class', 'hit-area')
  .attr('r', Math.max(actualRadius, 22))
  .attr('fill', 'transparent');

4. Narrative Structure

Three Acts:

1. Invitation - What draws viewer in? Why should they care?
2. Discovery - What patterns emerge? What surprises?
3. Reflection - What should they feel/understand/do?

Progressive Disclosure:

Level 1: Overview → Level 2: Exploration → Level 3: Detail → Level 4: Context

5. Data Pipeline

Structure:

scripts/
├── 01_fetch_raw.py    # API calls with caching
├── 02_clean_data.py   # Transformation
├── 03_validate.py     # Quality checks
└── 04_export.py       # Final format

Source Documentation (every dataset needs):

• URL, access date, update frequency
• License and confidence level
• Field descriptions and limitations

Scripts

Generate Color Palette

scripts/color-palette.py --type sequential --hue blue --steps 9
scripts/color-palette.py --type categorical --count 6 --colorblind-safe
scripts/color-palette.py --type diverging --low red --high blue

Analyze Data Distribution

scripts/analyze-distribution.py data.csv --column value
# Outputs: min, max, skew ratio, recommended scale

Scaffold D3 Project

scripts/d3-scaffold.py my-viz --type force-network
scripts/d3-scaffold.py my-viz --type timeline
scripts/d3-scaffold.py my-viz --type choropleth

Anti-Patterns to Avoid

• 3D charts (distorts perception)
• Pie charts with >6 categories
• Dual y-axes
• Rainbow color scales (perceptually uneven)
• Truncated y-axes without disclosure
• Animation without purpose

Quality Checklist

• Scale choice justified for data distribution
• Color palette is colorblind-safe
• Minimum 44x44px touch targets
• Clear entry point for viewer
• Sources documented
• Responsive on mobile