> For the complete documentation index, see [llms.txt](https://docs.rootcause.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.rootcause.ai/more-details/digital-twin/exploring-causal-model.md).

# Exploring the Causal Model

When you open a Digital Twin from the Digital Twins list, the causal graph fills the center and the right-hand panel opens on the **Home** tab. The graph is where the model lives — every variable, every relationship, every direction of cause and effect is visible at once.

This page covers the two things you do most often in this view: read the graph canvas, and drill into a single variable to see what drives it. The other views — model quality, configuration, version history, individual relationships — have their own pages.

For context on how a Digital Twin is built in the first place, see [Step 5: Build Digital Twin](/user-guide/creating-digital-twin.md).

***

## The graph canvas

The canvas lays the model out as a flowchart: variables positioned by their place in the causal chain, inputs toward the edges and outcomes downstream, with every edge an arrow of cause and effect. Each node carries its data-type badge.

<figure><img src="/files/JGBnqDjjkA3feX9ZukMT" alt="A trained Digital Twin open on its Home tab: the causal graph canvas fills the center with variables laid out left to right, and the right panel stacks the Model trained and ready banner with the Configuration, Versions, Simulations, Graph explorer, and Model evaluation cards"><figcaption><p>The twin Home. The causal graph fills the canvas; the card panel on the right summarizes the state of the model.</p></figcaption></figure>

Controls sit in the bottom-left corner: zoom in and out, fit the whole graph to the window, and lock the canvas against accidental dragging. **Recompute layout** redraws the arrangement after the graph changes. The model itself is untouched by any of these: they only change how it is drawn.

***

## The Home tab

The right-hand panel's Home view is a stack of cards summarizing the state of the twin:

* **Model status** — a banner reporting whether the twin is trained and ready to simulate.
* **Configuration** — the type, variable count, and relationship count behind the model. The [Config tab](/more-details/digital-twin/configuration.md) holds the detail, and **Modify model** starts a new version.
* **Versions** — the current version and its status; switch, compare, and manage saved versions via [Version History](/more-details/digital-twin/version-history.md).
* **Simulations** — recent runs. The [Simulations tab](/user-guide/simulations.md) runs and lists them.
* **Graph explorer** — how many relationships are in the model and how many candidate connections were tested and ruled out.
* **Model evaluation** — model fit at a glance. The [Evaluation tab](/more-details/digital-twin/model-quality.md) has the full metrics.

***

## Inspecting a single variable

Open a variable from the Relationships view (or click its node in the graph) and its full detail page opens. You see:

* A natural-language summary of how the variable is influenced, with how much of its behavior the model explains.
* A **Prediction quality** verdict on held-out data.
* The **distribution** the variable takes under the model's learned parameters.
* A ranked **Relationships** panel of its drivers, each marked direct or indirect with its relative strength.

<figure><img src="/files/2okpMDAKumgnBen5cawI" alt="The Variable Details page for Churn: an outcome driven by 4 variables with 26% of its behaviour explained, a Good fit verdict of 81% accuracy, 0.88 F1 and 0.84 AUC, the learned distribution chart, and a Relationships panel ranking Contract, InternetService, OnlineSecurity, and TechSupport as direct drivers"><figcaption><p>Variable Details for Churn. Contract, InternetService, OnlineSecurity, and TechSupport are its direct drivers; prediction quality and the learned distribution sit below the summary.</p></figcaption></figure>

***

## What to look for

* **The chain from input to outcome.** Pick the variable you care about — usually a business outcome — and trace backward through the graph. The strongest drivers in its Relationships panel tell you where intervention will have the most leverage.
* **Direct vs indirect drivers.** A variable can affect an outcome through several routes. The driver ranking shows the relative weight; the graph shows the actual path.
* **Surprises.** A driver you didn't expect in the top three. A relationship that runs the wrong way. A central variable that nothing else points to. These are usually the most interesting findings — and the most useful to validate with domain knowledge.

***

## Other Working with a Digital Twin pages

* [Reviewing Model Quality](/more-details/digital-twin/model-quality.md) — how trustworthy is the model?
* [Inspecting Causal Relationships](/more-details/digital-twin/causal-relationships.md) — individual edges and their statistics.
* [Configuration for Existing Twin](/more-details/digital-twin/configuration.md) — model settings, included variables, constraints.
* [Version History](/more-details/digital-twin/version-history.md) — multiple versions of the same twin.

See [Digital Twin overview](/more-details/digital-twin.md) — general overview.