[scatter-embedding] t-SNE and UMAP Embedding Visualization

[MarkusNeusinger]

Description

A scatter-style visualization showing high-dimensional data projected into 2D space using t-SNE or UMAP dimensionality reduction. Points are colored by cluster or class label, revealing groupings and structure in the data. This is a standard tool in machine learning for exploring embeddings, single-cell RNA-seq data, and NLP document clustering.

Applications

• Visualizing clusters in single-cell RNA-seq data (bioinformatics)
• Exploring word or document embeddings from NLP models
• Inspecting latent space structure of autoencoders or VAEs
• Quality check after K-means or DBSCAN clustering

Data

• x (float) — first embedding dimension
• y (float) — second embedding dimension
• label (categorical) — cluster or class assignment
• Size: 500–5000 points typical

Notes

• Show cluster labels as colored point groups with legend
• Optionally annotate cluster centroids
• Perplexity / n_neighbors parameter note in subtitle

[claude]

✅ Spec ID: scatter-embedding

Rationale: The ID captures that this is a scatter-style plot specifically for dimensionality reduction embeddings (t-SNE, UMAP), distinguishing it from generic scatter-categorical and the PCA-focused biplot-pca. Valid and useful plot type — a standard visualization in ML and bioinformatics with no existing duplicate.

[github-actions]

✅ Specification Ready: scatter-embedding

PR: #5261

---

specification.md

scatter-embedding: t-SNE and UMAP Embedding Visualization

Description

A scatter plot displaying high-dimensional data projected into 2D space using non-linear dimensionality reduction techniques such as t-SNE or UMAP. Points are colored by cluster or class label, revealing groupings and latent structure in the data. This is a standard visualization in machine learning for exploring embeddings, single-cell RNA-seq data, and NLP document clustering, helping practitioners verify that learned representations capture meaningful distinctions.

Applications

• Visualizing cell-type clusters in single-cell RNA-seq data after dimensionality reduction in bioinformatics workflows
• Exploring word or document embeddings from NLP models to verify semantic groupings and detect outliers
• Inspecting latent space structure of autoencoders or variational autoencoders (VAEs) to assess representation quality
• Quality-checking clustering results from K-means or DBSCAN by overlaying cluster assignments on the 2D projection

Data

• x (float) — First embedding dimension (e.g., t-SNE 1 or UMAP 1)
• y (float) — Second embedding dimension (e.g., t-SNE 2 or UMAP 2)
• label (categorical) — Cluster or class assignment for coloring points
• Size: 500–5000 points typical
• Example: Synthetic clustered data with 5–10 groups projected via t-SNE or UMAP

Notes

• Color each cluster/class with a distinct, colorblind-accessible color and include a legend mapping colors to labels
• Optionally annotate cluster centroids with the cluster label text
• Use moderate point size with slight transparency (alpha) to handle overlapping points in dense regions
• Include a subtitle noting the algorithm and key parameter (e.g., "t-SNE (perplexity=30)" or "UMAP (n_neighbors=15)")
• Axes represent embedding dimensions and typically should not have tick labels, as the coordinates are not directly interpretable

---

Next: Add approved label to merge this specification to main.

---

🤖 spec-create workflow

[github-actions]

🚀 Specification Merged!

Specification: scatter-embedding
PR: #5261 (merged)

The specification is now in main and synced to PostgreSQL.

---

Next: Add generate:{library} labels to start generating implementations.

Available libraries: matplotlib, seaborn, plotly, bokeh, altair, plotnine, pygal, highcharts, letsplot

Or use workflow dispatch for bulk operations.

---

🤖 spec-create workflow