Skip to content

Click in-app to access the full platform documentation for your version of DataRobot.

Deployment for segmented modeling

Availability information

Deployment for time series segmented modeling, available as a public preview feature, is off by default. Contact your DataRobot representative or administrator for information on enabling the feature.

Feature flag: Enable Time Series Segmented Deployments Support

To fully leverage the value of segmented modeling, you can now deploy combined models as you can deploy any other time series models. After selecting the champion model for each included project, you can deploy the combined model to bring predictions into production. Creating a deployment allows you to use DataRobot MLOps for accuracy monitoring, prediction intervals, and challenger models.

Note

Time series segmented modeling deployments do not support data drift monitoring, prediction explanations, or retraining.

Create a combined model

Availability information

Segmented modeling is available for multiseries regression projects.

When using segmented modeling, DataRobot creates a project for each segment, runs Autopilot (full or quick), and then selects (and prepares) a recommended model for deployment. DataRobot also marks the recommended model as the Segment Champion, although you can reassign the champion.

Note

Although DataRobot creates a project for each segment, these projects are not available from the Project Control Center. Instead, they are investigated and managed from within the Combined Model, which is available in the Project Control Center.

DataRobot combines all segment champions to create a Combined Model, a single project containing each segment. You can then deploy the Combined Model to create a "one-model" deployment for multiple segments; however, the individual segments in the deployed combined model still have their own segment champion models running in the deployment behind the scenes.

For more information on the segmented modeling process and creating a combined model, see Segmented modeling for multiseries.

Deploy a combined model

When segmented modeling completes, you can deploy the resulting combined model to bring predictions into production and leverage MLOps accuracy monitoring, prediction intervals, and challenger models.

To deploy a combined model, take the following steps:

  1. Once Autopilot has finished, the Model tab contains one model. This model is the completed Combined Model.

  2. Click the Combined Model, and then click Predict > Deploy.

  3. On the Deploy tab, click Deploy model.

    Note

    You can also click Add to Model Registry and then deploy the Combined Model from there.

  4. Add deployment information and create the deployment.

  5. Monitor, manage, and govern the deployed model in DataRobot MLOps.

Modify and clone a deployed combined model

When a combined model is deployed, to change the segment champion for a segment, you must clone the deployed combined model and modify the cloned model. This process is automatic and occurs when you attempt to change a segment's champion within a deployed combined model. The cloned model you can modify becomes the Active Combined Model. This process ensures stability in the deployed model while allowing you to test changes within the same segmented project.

Note

Only one combined model in a project can be the Active Combined Model (marked with a badge).

To modify and clone a deployed combined model, take the following steps:

  1. Once a Combined Model is deployed, it is labeled Prediction API Enabled.

  2. Click the active and deployed Combined Model, and then in the Segments tab, click the segment you want to modify.

  3. Reassign the segment champion.

  4. In the dialog box that appears, click Yes, create new combined model.

  5. On the segment's Leaderboard, you can access and modify the Active Combined Model.

Tip

For a short time, in the Combined Model updated toast, you can click Go to Combined Model to return to the segment's combined models in the Leaderboard.


Updated June 3, 2022
Back to top