Digital Twin
A Digital Twin is a causal model of your business or system. Unlike traditional predictive models that learn correlations, a Digital Twin understands cause and effect — enabling simulation, optimization, and counterfactual reasoning.
Building and using a Digital Twin covers Steps 4–6 of the seven-step workflow: Build Causal Graph, Build Digital Twin, and Run Simulations.
Building
Creating a Digital Twin — Select a Data View, configure model settings, and train the twin. This is where causal discovery runs.
Causal Graph — Read, interpret, and refine the discovered cause-and-effect graph with domain knowledge.
Using
Simulations — Run what-if scenarios, optimize decisions, find root causes, and predict outcomes.
Comparing
Model Comparison — Put two Digital Twin versions side by side to understand exactly what changed in structure and parameters.
The Digital Twin interface
Once created, you interact with a Digital Twin through five sections — panel twins add a sixth. The twin's Home shows the causal graph with a side panel of cards — Configuration, Versions, Simulations, Graph explorer, Model evaluation — and opening a card switches the view; the URL tracks the active section (?tab=…):
Home
The causal graph, model status, and the cards linking to every other section
Config
The data, fields, and algorithms behind the current version (read-only; Modify model opens the builder)
Relationships
View and edit causal relationships
Evaluation
Model quality metrics
Simulations
Run all simulation types
Environments (panel twins only)
Per-environment insights
Selecting a node or edge on the causal graph opens a details panel with Path Analysis and Model Probabilities views — plus Seasonal Trends on temporal twins and Environment Analysis on panel twins. Version History opens from the Versions card on Home, and two twins or versions can be compared side by side on the separate Compare page — see Model Comparison.
Last updated