Build Digital Twin

The causal graph you reviewed in Step 4 is a structural map — it shows which variables influence which others. A Digital Twin takes that structure and fits it with equations, giving you a runnable model you can interrogate with simulations.

Building the Digital Twin is largely automatic. Your main decisions are: which type of twin to train, which variables to include, and which training options to enable.

---

Navigate to Digital Twin

From the left sidebar, click Digital Twin. If this is your first twin, you'll see an empty list.

The Digital Twins list. Click + Create Digital Twin to begin.

Click + Create Digital Twin.

---

Configure the twin

The configuration page has three sections: Data View, Type, and Fields.

Configuration page. Select the Data View, choose a type, review the field list, then pick a training path.

Data View — select the 360 Data Table you built in Step 3. The name is auto-generated from the Data View name; you can edit it.

Type — choose how the model handles time:

Type	When to use
Static	Data without meaningful time ordering — customer attributes, cross-sectional snapshots, survey data
Temporal	Time-series data where variables influence each other across periods — trends, lags, forecasting

Temporal twins require a DateTime field in your Data View. If none exists, only Static is available.

Fields — all variables from the Data View are listed with their detected types. By default, all are included. Deselect any that are pure identifiers (customer ID, order ID) or metadata that has no causal meaning. When in doubt, include the field — the algorithm handles irrelevant variables better than missing drivers.

Training options:

• Confounder Modelling — detects and accounts for hidden variables that influence multiple observed variables. Recommended on.

• Equation Discovery — fits symbolic equations to describe each relationship precisely. Recommended on.

---

Start training

Two buttons appear at the bottom:

• Discover Relationships — runs causal structure discovery only. Use this if you want to review the graph before committing to a full training run.

• Full Training — runs the complete pipeline end to end. This is the normal path.

Click Full Training to proceed.

---

Training progress

Training runs five sequential stages.

Training progress. The five stages run in sequence; each can take from seconds to several minutes depending on dataset size.

Stage	What happens
Preparing Data	Variables standardised, missing values handled
Causal Discovery	Statistical tests identify cause-and-effect structure
Latent Confounder Modelling	Hidden common causes detected and modelled
Symbolic Equation Discovery	Equations fitted to quantify each relationship
Building & Evaluating Model	Final model assembled and quality metrics computed

Training time ranges from a few minutes for small datasets to longer for large ones. You can navigate away — training continues in the background. A badge on the Progress icon in the left sidebar tracks completion.

---

Review the trained twin

When training completes, you land on the Digital Twin overview.

The overview after training. Config summary on the right, model fit score in the Evaluation section, and suggested simulations ready to run.

The overview shows:

• Version — each training run creates a new version. Previous versions are retained and can be compared or rolled back to.

• Config — the Data View, twin type, variable count, relationship count, and algorithms used.

• Evaluation — overall model fit. A score above 60% is generally good; lower scores may indicate missing variables or data quality issues worth investigating.

• Suggested simulations — RootCause proposes simulation questions based on your model structure. These are a good starting point.

---

Next step

Your Digital Twin is trained and ready. Now you can run simulations to ask what-if questions, find key drivers, and test interventions.

Next step: Step 6: Run Simulations