Skip to content

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

Deploy tab

From the Leaderboard, select the model to use for generating predictions and click the Deploy tab.

In most cases, before deployment, you should unlock holdout and retrain your model at 100% to improve predictive accuracy. DataRobot automatically runs Feature Impact for the model (this also calculates Prediction Explanations, if available). You can access Prediction Explanations via the predictionExplanations route for the model.

If you have not already initiated the process in which DataRobot prepares the model for deployment, best practice recommends that you click Prepare for deployment. DataRobot runs feature impact, retrains the model on a reduced feature list, trains on a higher sample size and then the full sample size (latest data for date/time partitioned projects).

Note

For binary classification models, which use a prediction threshold, changing the threshold value via the Deploy tab does not persist through the deployment preparation process. This is because any new models added to the Leaderboard are assigned the default value 0.5. When deploying the prepared model, if you want it to use a value other than the default, set the value after the model has the Prepared for Deployment badge applied.

Once a model is deployed, the Leaderboard description indicates as much:

Add a deployment

Note

The Deploy tab behaves differently in environments without a dedicated prediction server, as described in the section on shared modeling workers, below.

The Deploy Model page provides an at-a-glance status of all currently deployed models and allows you to create a new deployment for the selected model:

To deploy a new, prepared model, click Deploy model. If using a binary classification model, set the prediction threshold before proceeding. (You can also deploy models that have not been run through the preparation process.)

Add deployment information

When you create a new deployment, you are redirected to the deployment information page.

The deployment information page outlines the capabilities of your current deployment based on the data provided. It populates fields for you to provide details about the training data, inference data, model, and your outcome data.

The progress bar at the top of the screen displays which deployment capabilities are enabled based on the data (and details about that data) you provide.

Under the Inference header, complete the fields, choose a prediction time stamp (non-time series projects), and select whether to allow data drift. Indicate if an association ID is required for making prediction requests with this deployment.

Field Description
Inference data DataRobot stores a deployment's inference data when a deployment is created. It cannot be uploaded separately.
Execution endpoint A prediction server that you will use to make predictions.
Prediction timestamp Determines the method for time-stamping prediction rows. Use the time of the prediction request or use a date/time feature (e.g., forecast date) provided with prediction data to determine the timestamp. Forecast date time-stamping is set automatically for time series deployments. It allows for a common time axis to be used between training data and the basis of data drift and accuracy statistics. This setting cannot be changed after the deployment is created and predictions are made.
Association ID toggle The column name that contains the association IDs in the prediction dataset for the model. The association ID functions as an identifier for the prediction dataset so you can later match up outcome data (also called "actuals") with those predictions. Read more about how association IDs enable accuracy tracking for a deployment here.
Data Drift settings Configures DataRobot to track target and feature drift in a deployment. Once enabled, you can activate additional features: prediction row storage, challenger models, and segmented analysis.
Segmented analysis toggle A toggle that, when enabled, allows DataRobot to perform segment analysis on prediction data.

Note

Time series projects have an additional option to set a prediction interval, described here.

When you have added all the available data and your model is fully defined, your deployment is ready to be created. Click Create deployment at the top of the screen.

When deployment creation completes, you are taken to the new deployment's Overview tab.

Once you have activated an instance, you can deploy to additional instances or (depending on your organization's configuration) send to a new dedicated server instance.

Using shared modeling workers

If you don't have a dedicated prediction server instance available, you can use a node that shares workers with your model building activities. In this case, the page has a different interface.

Click Show Example to generate and display a usage example:

When using the sample code, specify your API token (1). The project and model IDs (2) are available in the sample, as is the shared instance endpoint (3). The DataRobot Python client uses the API token for authentication and so no key or username is required. To execute the file, follow the instructions in the commented section of the snippet.


Updated October 27, 2021
Back to top