This application provides an interactive visualization tool for exploring patterns discovered by the NclusterBox algorithm in Boolean and fuzzy tensors.
NclusterBox is an algorithm that summarizes n-way Boolean and fuzzy tensors using sub-tensors, represented as patterns. It identifies significant patterns within the data based on their density and area, providing insights into the underlying structure and relationships within the tensor.
- Interactive Scatter Plot: Visualize patterns as data points in a 2D scatter plot.
- Size of data points represents the pattern size.
- Color of data points represents the pattern density.
- Zoom and pan to explore different regions of the visualization.
- Tooltips display detailed information about each pattern.
- Highlight data points and their connections to visualize intersections between patterns.
- RSS Evolution Graph: Analyze the Residual Sum of Squares (RSS) evolution of the model as the number of patterns increases.
- Identify potential elbows in the graph to determine an optimal number of patterns for summarizing the tensor.
- Truncate the model by selecting a specific number of patterns.
- Pattern Summary View: Explore detailed information about individual patterns.
- View the dimensions, density, and size of the pattern.
- Inspect the specific values of each dimension.
Download the appropriate installation files for your operating system from the Releases page.
- Launch the application.
- Select the tensor and patterns files generated by the NclusterBox algorithm.
- Explore the visualization, RSS evolution graph, and pattern summary view to analyze the discovered patterns.
- Frontend: Angular
- Backend: Rust
- Visualization Library: D3.js
- Desktop Application Framework: Tauri