beautiful-data-visualizations

name: beautiful-data-visualizations description: Agent skill for creating publication-ready, beautiful data visualizations with matplotlib, seaborn, plotly. Generates professional charts, dashboards, heatmaps, with proper styling, annotations, and presentation-quality outputs suitable for business presentations and scientific publications. license: MIT

Beautiful Data Visualizations Skill

When users ask you to create visualizations, analyze data with charts, build dashboards, or present data insights, use this skill to generate publication-ready, beautiful visualizations.

Core Competencies

1. Professional Styling & Color Schemes

• Color Palettes: Use seaborn palettes (viridis, husl, Set2, coolwarm) or custom matplotlib colormaps
• Consistency: Apply uniform style across all figures (same fonts, sizes, colors)
• Accessibility: Ensure colorblind-friendly palettes (Okabe-Ito, Paul Tol schemes)
• Typography: Use serif fonts (Times) for labels, sans-serif (Arial, Helvetica) for titles

2. Figure Quality Standards

• DPI: 300 DPI minimum for print/presentation
• Dimensions: 12x8 or 14x10 inches for main charts; 10x6 for subplots
• Aspect Ratio: 16:9 for presentations, 4:3 for papers
• Margins: Use tight_layout() or gridspec for proper spacing
• Fonts: Title 16-18pt, axes labels 12-14pt, tick labels 10-12pt

3. Specific Chart Types & Implementation

3.1 Correlation Analysis

Purpose: Show relationships between numerical variables

3.2 Distribution Analysis

Purpose: Show data distributions (histograms, KDE, violin plots)

3.3 Categorical Comparisons

Purpose: Compare values across categories (bar, box, violin plots)

3.4 Relationship & Trend Analysis

Purpose: Show relationships between continuous variables (scatter, trend)

3.5 Multi-Panel Dashboards

Purpose: Combine multiple charts into professional dashboards

4. Advanced Techniques

4.1 Annotation & Highlighting

• Highlight important points with markers
• Add value labels on bars
• Add statistical annotations

4.2 Color & Style Customization

• Define professional color schemes
• Use seaborn palettes
• Custom colormaps

4.3 Export & Output Quality

• High-quality PNG (300 DPI)
• Vector format (scalable) - PDF, SVG
• Interactive HTML (Plotly)

Workflow for Data Visualization Projects

Step 1: Data Exploration

1. Load data with pandas
2. Check dimensions, data types, missing values
3. Calculate summary statistics
4. Identify key variables for analysis

Step 2: Planning Visualizations

1. Understand the story you want to tell
2. Identify visualization types that support the narrative
3. Plan chart composition and layout
4. List specific metrics/statistics to display

Step 3: Implementation

1. Create individual charts with consistent styling
2. Add proper titles, axis labels, legends, and annotations
3. Ensure color accessibility and visual hierarchy
4. Test readability at presentation size

Step 4: Refinement

1. Check alignment and spacing in multi-panel layouts
2. Verify all text is legible (font sizes, colors)
3. Ensure consistent styling across all charts
4. Export at appropriate resolution (300 DPI for print)

Common Analysis Patterns

Pattern 1: Feature Importance Analysis

When: Identifying most impactful variables How: Bar charts sorted by magnitude, horizontal for readability

Pattern 2: Statistical Comparison

When: Comparing means/medians across groups How: Box plots with overlaid strip plots, statistical annotations

Pattern 3: Time Series with Multiple Metrics

When: Tracking multiple variables over time How: Dual/triple y-axes or subplots

Best Practices Checklist

• Clarity: One clear insight per chart
• Context: Include reference lines, means, benchmarks
• Consistency: Same fonts, colors, styles across all charts
• Completeness: All axes labeled, units specified, legends clear
• Color: Use colorblind-safe palettes, ensure sufficient contrast
• Annotations: Important values/insights clearly labeled
• Resolution: 300 DPI for print, appropriate file formats
• Simplicity: Remove chart junk, keep only essential information
• Accessibility: Large fonts, high contrast, no reliance on color alone
• Testing: Verify readability at actual presentation size

Required Libraries

pip install pandas numpy matplotlib seaborn scipy scikit-learn plotly

When to Use This Skill

• Creating charts for presentations or publications
• Exploring data relationships visually
• Building data analysis dashboards
• Comparing distributions across groups
• Highlighting trends and patterns
• Preparing publication-quality figures

Limitations & When to Use Alternatives

• Interactive dashboards: Use Plotly/Dash instead of static matplotlib
• Real-time monitoring: Use Grafana or similar tools
• 3D visualizations: Consider Plotly 3D or other specialized tools
• Geographic data: Use Folium or GeoPandas for maps