t-SNE Visualization Examples

These examples demonstrate the interactive t-SNE visualization capabilities of JSPlots.jl.

t-SNE (t-Distributed Stochastic Neighbor Embedding) is a dimensionality reduction technique that maps high-dimensional data to 2D while preserving local structure.

Interactive Features:

• Randomize: Reset to random initial positions
• Step: Run one iteration of the algorithm
• Run to Convergence: Automatically iterate until the specified threshold
• Drag nodes: Manually reposition any point, then continue optimization
• Color by: Change node coloring without affecting positions
• Perplexity: Controls local vs global structure (lower = more local)
• Learning Rate: Controls step size (higher = faster but less stable)
• Continuous Color Gradients: Numeric columns can display smooth color transitions