Security Behavior Analytics — Segmentation Parameters
Pre-computed analytical segmentation centroids and engine configuration for the Security Behavior Analytics UEBA platform.
Contents
| File | Description |
|---|
config.json | Engine version, cohort dimensions, framework metadata |
segmentation_params.json | K-Means, GMM, HDBSCAN, Hierarchical cluster parameters |
train_metrics.json | Evaluation summary (no ML training performed) |
Usage
These parameters are loaded automatically by the Gradio Space dashboard. No runtime training required.
Limitations
- Trained on 100% synthetic anonymized IAM behavior events
- Segmentation centroids are static after export
- Not intended for production deployment without re-calibration on real (anonymized) data
Intended Use
- Portfolio demonstration of UEBA analytics methodology
- Security data science education and benchmarking
- SOC/IAM team evaluation of behavior analytics approaches
