Data Visualization

Consolidated skill covering static plots, interactive charts, dashboards, infographics, and generative art. Routes to specialized references by task type.

Task Router

Need	Tool	Reference
Static plots, publication figures, subplots, styling	Matplotlib	`references/matplotlib.md`
Statistical plots (distributions, regressions, pair plots)	Seaborn + Matplotlib	`references/matplotlib.md`
Deep statistical: pairplot, clustermap, QQ, forest, survival	Seaborn + Matplotlib + lifelines	`references/statistical-plots.md`
Interactive charts, web-ready, hover/zoom/pan	Plotly Express / Graph Objects	`references/plotly.md`
Dashboards with callbacks, multi-page apps	Plotly Dash	`references/plotly.md`
Real-time / live dashboards, dcc.Interval, streaming	Plotly Dash	`references/realtime-dashboards.md`
Geographic maps, choropleth, scatter_map, Folium, GIS	Plotly / Folium	`references/geographic-maps.md`
Network graphs, Sankey, force-directed, NetworkX, treemap	Plotly + NetworkX	`references/network-graphs.md`
WCAG accessibility, color blindness, alt text, ARIA	Best practices	`references/accessibility-compliance.md`
Custom web viz, D3 force layout, React + D3	D3.js	`references/d3-custom-viz.md`
AI-generated infographics, visual data storytelling	Gemini MCP	`references/infographics.md`
Generative art, algorithmic art, flow fields, p5.js	p5.js	`references/generative-art.md`
Large-dataset server-side rendering	Bokeh / Datashader	Use Context7 for API docs
Excel charts, pivot tables, conditional formatting, data validation	openpyxl + pandas	`references/excel-charts.md`
Color themes, palette selection, professional styling	Theme factory	`references/color-themes.md`
Animated GIFs, PIL animation, Slack GIFs, easing functions	Pillow / matplotlib.animation	`references/animated-visualizations.md`

Decision Tree

What is the output?
  ART (generative, algorithmic, creative) --> p5.js (references/generative-art.md)
  INFOGRAPHIC (visual storytelling, poster-style) --> Gemini MCP (references/infographics.md)
  ANIMATED GIF / Slack GIF --> PIL + easing (references/animated-visualizations.md)
  EXCEL FILE (charts, pivot tables, dashboards) --> openpyxl (references/excel-charts.md)
  COLOR THEME / PALETTE --> references/color-themes.md
  DATA CHART/PLOT -->
    Need interactivity (hover, zoom, pan, callbacks)?
      YES --> Need full dashboard with state/callbacks?
        YES --> Plotly Dash
        NO  --> Plotly Express (px for quick, go for custom)
      NO  --> Need statistical semantics (distributions, regression, facets)?
        YES --> Seaborn (builds on matplotlib)
        NO  --> Matplotlib OO API
    Need 3D?
      Static 3D --> mpl_toolkits.mplot3d or Plotly go.Surface
      Interactive 3D --> Plotly scatter_3d / surface
    Need animation?
      Scientific/data --> matplotlib.animation (FuncAnimation, save to mp4/gif)
      Creative/Slack GIF --> PIL frame-by-frame (references/animated-visualizations.md)
      Web --> Plotly animation_frame or D3 transitions
    Need maps?
      Choropleth / scatter / GIS → references/geographic-maps.md
      Folium (layers, popups, clusters) → references/geographic-maps.md
    Need network/hierarchy?
      Sankey (flows) → references/network-graphs.md
      Treemap / Sunburst → references/network-graphs.md
      NetworkX force-directed → references/network-graphs.md
    Need live/streaming data?
      dcc.Interval polling → references/realtime-dashboards.md
      extendData streaming → references/realtime-dashboards.md
    Need statistical depth?
      pairplot, clustermap, lmplot → references/statistical-plots.md
      QQ plot, forest plot, survival → references/statistical-plots.md
    Need custom web viz?
      D3 force layout, React + D3 → references/d3-custom-viz.md
    Accessibility required?
      WCAG checklist, color blindness → references/accessibility-compliance.md

Matplotlib Essentials

Always use OO interface: fig, ax = plt.subplots(figsize=(10, 6), constrained_layout=True).

Type	Method	When
Line	`ax.plot()`	Time series, trends
Scatter	`ax.scatter()`	Correlations
Bar	`ax.bar()` / `ax.barh()`	Categorical comparisons
Histogram	`ax.hist()`	Distributions
Heatmap	`ax.imshow()` + `colorbar`	Matrices, correlations
Box/Violin	`ax.boxplot()` / `ax.violinplot()`	Statistical spread
Contour	`ax.contour()`	2D scalar fields

Colormaps: viridis (default, accessible), cividis (colorblind-safe), coolwarm (diverging). Avoid jet.

Seaborn integration: import seaborn as sns; sns.set_theme() then use sns.histplot(), sns.pairplot(), sns.heatmap() for statistical plots with automatic aesthetics.

Save: plt.savefig('fig.png', dpi=300, bbox_inches='tight'). Close: plt.close(fig).

See references/matplotlib.md for layouts, 3D, animations, styling, and export details.

Plotly Essentials

Two APIs: px (high-level, DataFrames) and go (low-level, fine control). Combine: start with px, refine with fig.update_layout().

Key chart types: scatter, line, bar, histogram, box, violin, heatmap, sunburst, treemap, sankey, candlestick, choropleth, scatter_3d, surface.

Templates: plotly_white, plotly_dark, ggplot2, seaborn, simple_white.

Export: fig.write_html('chart.html') (interactive), fig.write_image('chart.png') (static, needs kaleido).

Dash: dcc.Graph(figure=fig) inside app.layout. Use @app.callback for interactivity.

See references/plotly.md for all chart types, subplots, interactivity, and Dash patterns.

Infographics Essentials

Use Gemini MCP for AI-generated infographics. Workflow: gemini_image generates, gemini_analyze scores, gemini_image_edit refines. Loop until quality threshold met.

Types: statistical, timeline, process, comparison, list, geographic, hierarchical. Styles: corporate, healthcare, technology, education, marketing, finance. Palettes: wong (recommended), ibm, tol.

See references/infographics.md for CLI usage, quality thresholds, prompt tips, and research integration.

Generative Art Essentials

p5.js for algorithmic/generative art. Always use seeded randomness (randomSeed(seed); noiseSeed(seed)) for reproducibility.

Core techniques: flow fields (Perlin noise vectors), particle systems, recursive structures, circle packing, Voronoi tessellation, L-systems.

Output: self-contained HTML artifact with inline p5.js. Include sidebar with seed navigation (prev/next/random) and parameter sliders.

See references/generative-art.md for philosophy framework, implementation patterns, templates, and parameter design.

Accessibility

Apply these rules to every visualization, not just when asked:

Color: Use colorblind-safe palettes (viridis/cividis for matplotlib, wong for infographics). Never encode meaning with color alone.
Redundancy: Add patterns, hatching, or markers alongside color. Use distinct line styles (solid, dashed, dotted).
Contrast: Minimum 4.5:1 for text, 3:1 for graphical elements (WCAG AA). Test with grayscale conversion.
Alt text: Every chart saved as an image needs a descriptive alt text string. Format: "[Chart type] showing [relationship] of [variables]. Key finding: [insight]."
Labels: Direct-label lines/bars when possible instead of relying on legends. Use fontsize>=12 for readability.
Screen readers: For web charts, add role="img" and aria-label to container divs. Plotly has built-in fig.update_layout(meta=...).

Performance

Large datasets (>100k points): Use rasterized=True in matplotlib, scattergl trace type in Plotly, or Datashader for server-side aggregation.
Web optimization: fig.write_html('chart.html', include_plotlyjs='cdn') cuts file size by ~3MB. Compress PNGs with pngquant.
Memory: Always plt.close(fig) after saving. In loops, use plt.close('all').
Animation: Prefer FuncAnimation with blit=True for matplotlib. For web, Plotly animation_frame handles transitions automatically.
Notebook rendering: Use %matplotlib inline for static, %matplotlib widget for interactive. Plotly: fig.show(renderer='notebook').

Choosing a Tool (Quick Matrix)

Criterion	Matplotlib	Seaborn	Plotly	Dash	Infographics	p5.js
Interactivity	Limited	Limited	Full	Full + callbacks	None	Full
Publication quality	Excellent	Excellent	Good	Good	Good	N/A
Learning curve	Moderate	Low	Low (px)	Moderate	None	Moderate
Web deployment	Manual	Manual	Native HTML	Native	Image embed	Native HTML
Best for	Papers, reports	Statistical EDA	Dashboards, exploration	Apps	Storytelling	Art

data-viz