spec: add linked-views-selection specification

[github-actions]

New Specification: linked-views-selection

Related to #3344

---

specification.md

linked-views-selection: Multiple Linked Views with Selection Sync

Description

Multiple coordinated plots where selecting data in one view automatically highlights or filters corresponding data points in all other views. This brushing-and-linking interaction pattern is fundamental for exploratory data analysis, allowing users to discover relationships across different visual representations of the same dataset. By selecting a cluster in a scatter plot, for example, users can instantly see how those points distribute in histograms, how they appear in parallel coordinates, or which categories they belong to.

Applications

• Exploratory data analysis dashboards where analysts investigate multivariate datasets by brushing across scatter plots, histograms, and categorical views
• Customer segmentation tools where selecting a group in one dimension reveals their characteristics across demographics, behavior, and transactions
• Scientific data exploration comparing different measurements or experimental conditions with synchronized highlighting
• Financial portfolio analysis linking asset allocation views with performance metrics and risk indicators

Data

• x (numeric) - Values for scatter plot horizontal axis
• y (numeric) - Values for scatter plot vertical axis
• category (categorical) - Group labels for bar chart or color encoding
• value (numeric) - Additional numeric dimension for histogram or secondary visualizations
• Size: 100-1000 points (interactivity works best with moderate dataset sizes)
• Example: Iris dataset or simulated multivariate data with numeric and categorical columns

Notes

• Implement at least 2-3 coordinated views (e.g., scatter plot + histogram + bar chart)
• Selection in any view should highlight corresponding points in all other views
• Use consistent color encoding across all views for the same data points
• Unselected points should be visually de-emphasized (reduced opacity or gray)
• Include a clear selection mechanism (brush, click, or lasso)
• Provide a reset/clear selection button to restore full view
• Consider showing selection count or summary statistics
• Libraries with native linked selection support (Altair, Bokeh, Plotly) should leverage those features

---

Next: Add approved label to the issue to merge this PR.

---

🤖 spec-create workflow