Use Case
Research & Analysis
Operators react to what's happening now. Researchers ask why it happened, what will happen next, and how to prove it. Tetrapus supports both — with the same data, same tools, different workflows.
Operators vs. Researchers
Industry Operators
- Goal: Keep things running. Respond to events in real time.
- Time orientation: Now. What is happening right this second?
- Workspace: Scene3D + Control pane + Dashboard KPIs. Fixed layout, always open.
- Data flow: Live telemetry → visual rules → alerts → commands
- Automation: TAI learns their reactions and automates them
Researchers & Analysts
- Goal: Understand patterns. Test hypotheses. Produce evidence.
- Time orientation: Past and future. What happened? What correlates? What will happen if?
- Workspace: Plot + Timeline + Inspector + Table. Layouts change per investigation.
- Data flow: ClickHouse queries → replay → analysis → computed fields → export
- Automation: Computed fields, anomaly detection, statistical overlays
Research workflows
Historical Replay & Investigation
Load any time window from ClickHouse and replay it at 0.5x to 10x speed. Scrub to any moment, step frame by frame, overlay annotations. Every pane — Scene3D, Plot, Table, Inspector — synchronizes to the replay clock.
- Timeline pane with scrubber, speed control, and frame stepping
- ClickHouse temporal queries with configurable time ranges
- All panes sync to the replay clock — globe, charts, and tables show the same moment
- Annotations persist across sessions for collaborative investigation
Statistical Analysis & Anomaly Detection
Layer statistical tools on top of live or historical data. Plot panes support trend lines, standard deviation bands, regression, and dual-axis comparison. Server-side Z-score anomaly detection and Pearson correlation run entirely in ClickHouse.
- Linear regression trend lines on any time series
- Standard deviation bands for noise characterization
- Z-score anomaly detection computed server-side in ClickHouse CTEs
- Cross-domain Pearson correlation over configurable windows
- Computed fields: RATE(), SLOPE(), STDDEV_WINDOW() for derived metrics
Spatial Analysis & Pattern Discovery
Use H3 hexagonal heatmaps to visualize geographic density patterns. Voronoi tessellation partitions space by nearest entity. KDE surfaces reveal continuous density distributions. Combine with the filter expression language to isolate specific populations.
- H3 hexagonal aggregation from resolution 3 (~70 km) to 9 (~175 m)
- Voronoi tessellation for coverage analysis
- KDE heatmaps for continuous density surfaces
- Filter expression language for isolating entity subsets
- Range rings, distance measurement, and area calculation tools
Hypothesis Testing with Computed Fields
Define derived metrics using the expression language — arithmetic, group aggregations, windowed temporal functions, and ranking. Test whether a relationship holds by plotting the computed field against raw data. Export results for external statistical packages.
- Expression engine: AVG(), MIN(), MAX(), SUM(), COUNT() over entity groups
- Temporal windows: RATE(), SLOPE(), STDDEV_WINDOW() over time
- RANK() within groups for relative positioning
- Export to CSV, JSON, or XLSX for R, Python, MATLAB integration
- Saved views: name and recall specific analysis configurations
Collaborative Analysis & Export
Share workspace layouts as YAML files so colleagues can reproduce your exact analysis setup. Export any pane configuration as a template. Copy-to-clipboard on every section, table, and code block for AI-assisted workflow.
- Workspace layouts persist and can be exported as YAML
- Pane templates: save and share configurations
- Copy-to-clipboard on all documentation sections
- Classification-marked exports for regulated research environments
- Git-backed configuration versioning for experiment reproducibility
Recommended research layout
This layout gives researchers temporal control (left), spatial context (centre), and data access (right). Save it as a workspace template and share with your team.
When Tetrapus fits your research
Geospatial data
Your data has latitude/longitude — vehicles, sensors, aircraft, satellites, weather stations, seismic networks.
Time-series at scale
You need to analyse millions of time-series records with sub-second query response.
Multi-entity correlation
You want to compare behaviour across entities, groups, or regions simultaneously.
Reproducible workflows
Your analysis setup needs to be versioned, shared, and reproduced by collaborators.
Live + historical
You switch between real-time monitoring and historical investigation on the same platform.
Export for external tools
You need to extract filtered, classified datasets for R, Python, MATLAB, or publication.
Ready to see the full picture?
One docker pull. Full stack in minutes. No build tools required.