# View model insights

> View model insights - Feature Impact shows which features are driving model decisions the most. It
> is calculated using permutation or SHAP.

This Markdown file sits beside the HTML page at the same path (with a `.md` suffix). It summarizes the topic and lists links for tools and LLM context.

Companion generated at `2026-10-09T14:05:25.095082+00:00` (UTC).

## Primary page

- [View model insights](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/model-insights.html.md): Full documentation for this topic (Markdown sidecar).

## Sections on this page

- [Next steps](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/model-insights.html.md#next-steps): In-page section heading.

## Related documentation

- [NextGen UI documentation](https://docs.datarobot.com/en/docs/workbench/index.html.md): Linked from this page.
- [Registry](https://docs.datarobot.com/en/docs/workbench/registry/index.html.md): Linked from this page.
- [Models](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/index.html.md): Linked from this page.
- [DataRobot](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/register-dr-models.html.md): Linked from this page.
- [custom models](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/register-cus-models.html.md): Linked from this page.
- [external models](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/register-ext-models.html.md): Linked from this page.
- [Sliced insights?](https://docs.datarobot.com/en/docs/reference/pred-ai-ref/sliced-insights.html.md): Linked from this page.
- [Feature Impact](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/feature-impact.html.md): Linked from this page.
- [Individual Prediction Explanations](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/shap-predex.html.md): Linked from this page.
- [SHAP Distributions: Per Feature](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/shap-distribution.html.md): Linked from this page.
- [Lift Chart](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/lift-chart.html.md): Linked from this page.
- [ROC Curve](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/roc-curve.html.md): Linked from this page.
- [Residuals](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/residuals.html.md): Linked from this page.
- [Confusion matrix](https://docs.datarobot.com/en/docs/workbench/build-workbench/experiments/experiment-insights/confusion-matrix.html.md): Linked from this page.
- [Deploy registered models](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/deploy-models.html.md): Linked from this page.
- [Generate compliance documentation](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/compliance-doc.html.md): Linked from this page.
- [Manage key values](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/key-values.html.md): Linked from this page.

## Documentation content

On the Registry > Models tab, the Insights tab is available for [DataRobot](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/register-dr-models.html.md) and [custom models](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/register-cus-models.html.md) (not agent-monitored [external models](https://docs.datarobot.com/en/docs/workbench/registry/model-directory/register-ext-models.html.md)). The following insights are supported:

**Feature Impact:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-1.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-1.png)

**Individual Prediction Explanations:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-2.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-2.png)

**SHAP Distributions: Per Feature:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-3.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-3.png)

**Lift Chart:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-4.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-4.png)

**ROC Curve:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-5.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-5.png)

**Residuals:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-6.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-6.png)

**Confusion Matrix:**
[https://docs.datarobot.com/en/docs/images/registry-model-insights-7.png](https://docs.datarobot.com/en/docs/images/registry-model-insights-7.png)


| Insight | Description | Problem type | Sliced insights? |
| --- | --- | --- | --- |
| Feature Impact | Shows which features are driving model decisions. | All | ✔ |
| Individual Prediction Explanations | Estimates how much each feature contributes to a given prediction, with values based on difference from the average. | Binary classification, regression | ✔ |
| SHAP Distributions: Per Feature | Displays, via a violin plot, the distribution of SHAP values and feature values to aid in the analysis of how feature values influence predictions. | Binary classification, regression | ✔ |
| Lift Chart | Depicts how well a model segments the target population and how capable it is of predicting the target. | All | ✔ |
| ROC Curve | Provides tools for exploring classification, performance, and statistics related to a model. | Binary classification | ✔ |
| Residuals | Provides scatter plots and a histogram for understanding model predictive performance and validity. | Regression | ✔ |
| Confusion matrix | Compares actual with predicted values in multiclass classification problems to identify class mislabeling. | Classification, time-aware |  |

> [!NOTE] Compute insights
> When you first access the Insights tab for a registered model version, to view the table of predictions for the Individual Prediction Explanations insight and the violin chart for the SHAP Distributions: Per Feature insight, click Compute.

## Next steps

After reviewing a registered model version's insights, deploy the model, generate compliance documentation, or manage its key values.

- Deploy registered models : Deploy the registered model version you've been reviewing to a prediction environment to start making predictions.
- Generate compliance documentation : Create a compliance report that documents the model, including its insights, for governance and audit purposes.
- Manage key values : Add key-value metadata to the registered model version to track additional context alongside its insights.