data-viz

Mental Model

Chart Selection

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)

Output Formats

• PNG: Static, high-quality for reports
• HTML: Interactive (zoom, pan, hover)
• SVG: Vector for editing

{chart_type}_{timestamp}.{ext}

Workflow

1. Ask: What's the question? What story to tell?
2. Load data from CSV/JSON (or data-analysis output)
3. Choose chart type based on question
4. Create Python script and execute using virtual environment:

   .venv/bin/python script.py

5. Return file path to user

Python Environment

# 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.py

References (load on demand)

references/REFERENCE.md

references/templates.md