Systems and methods for data analysis
The method computes tessellation cells and identifies clustered data points based on density thresholds to analyze large data sets efficiently, reducing computational overhead and eliminating bias, thus enhancing data analysis.
Patent Information
- Application Number
- PCT/US2025/021316
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-03-25
- Filing Date
- 2025-03-25
- Publication Date
- 2025-10-02
AI Technical Summary
Existing methods for data analysis are computationally inefficient, limiting the insights that can be gleaned from large, unstructured data sets, and often require hyperparameters that introduce bias.
The method involves computing a tessellation cell for each input data point, removing unbounded cells, calculating instantaneous spatial density, identifying clustered data points based on a density threshold, and grouping them into clusters, followed by analyzing these clusters in continuous or discrete space to identify relationships, without the need for hyperparameters.
This approach reduces computational overhead and runtime, allowing for efficient analysis of complex data sets at all spatial scales with improved accuracy and eliminates bias by avoiding the use of hyperparameters.