Mental Model
Visualization is choosing the right chart to answer a specific question. Chart type depends on data relationship, not aesthetics.
Chart Selection
| Question | Chart Type |
|---|---|
| Trends over time? | Line chart |
| Compare categories? | Bar chart |
| Show distribution? | Histogram, box plot |
| Relationship between variables? | Scatter plot |
| Parts of whole? | Pie, stacked bar |
| 2D patterns? | Heatmap |
| Financial data? | Candlestick, OHLC |
Anti-Patterns (NEVER)
- Don't use Chinese characters anywhere in charts - use English for labels, titles, legends, data labels
- Don't use Chinese characters in filenames (ASCII only)
- Don't pick wrong chart type for the question
- Don't overload with data → aggregate or sample
- Don't forget labels, title, legend
- Don't use poor colors (colorblind-safe palettes)
Chart language: Always use English (titles, axes, legends, labels) to avoid font rendering issues.
Output Formats
- PNG: Static, high-quality for reports
- HTML: Interactive (zoom, pan, hover)
- SVG: Vector for editing
Filename: {chart_type}_{timestamp}.{ext} (ASCII only)
Workflow
- Ask: What's the question? What story to tell?
- Load data from CSV/JSON (or data-analysis output)
- Choose chart type based on question
- Create Python script and execute using virtual environment:
.venv/bin/python script.py - Return file path to user
Python Environment
Auto-initialize virtual environment if needed, then execute:
# Navigate to skill directory
cd skills/data-viz
# Auto-create venv if not exists
if [ ! -f ".venv/bin/python" ]; then
echo "Creating Python environment..."
./setup.sh
fi
# Execute script
.venv/bin/python your_script.pyThe setup script auto-installs: matplotlib, seaborn, plotly, pandas with Chinese font support.
References (load on demand)
For chart APIs and code templates, load: references/REFERENCE.md, references/templates.md