Skip to content

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

Add deployments

You can create new deployments in various ways, depending on the type of starting artifact you want to use.

Starting artifact Deployment method
DataRobot model Deploy a DataRobot model by selecting it from the Leaderboard and navigating to the Predict > Deploy tab. Alternatively, upload your model directly to the deployments inventory using the Add Deployment link.
Custom model Deploy a custom model from the Custom Model Workshop. Alternatively, upload your model directly to the deployments inventory using the Add Deployment link.
External model (monitored by the MLOps agent) Upload an external model's training data directly to the deployments inventory to create an external deployment. You can upload historical prediction data after the deployment is created. Alternatively, deploy an external model package.
Historical prediction data Upload the training data for a model that made predictions in the past directly to the deployments inventory to create an external deployment. Add historical prediction data post-deployment.

When you initiate the creation of a new deployment (using any of the methods described above), you are redirected to the deployment information page.

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

Deploy custom models

This section outlines how to create a deployment with a custom model by uploading it directly to the deployments inventory. You can also reference the workflow for deploying a custom model from the Custom Model Workshop.

Navigate to Deployments and click the Add Deployment link.

Under Model, select browse > Local file to upload a folder containing your custom model contents. You can also select Model registry if you instead want to create a deployment with a custom model package.

Note

Reference the documentation for preparing a custom model folder before proceeding.

After uploading your model, you are directed to the deployment information page. Fill out the fields under the Model header:

Field Description
Name The name of the custom model.
Model Content The files that make up the model. You can remove files or click Add Files to upload the necessary files for the custom model.
Prediction type The type of prediction the model is making, either binary classification or regression. For a classification model, you must also provide the positive and negative class labels and a prediction threshold.
Target The dataset column name the model will predict on.
Runtime environment The environment you want your model code to run on. You can choose from the default environments provided by DataRobot, or create your own custom environment.
Functional validation data A partition of the training data that is used to evaluate model performance. Click Browse to upload the data.

Under the Inference header, complete the fields and select optional configurations for the deployment:

Field Description
Inference data The inference data is being stored by DataRobot and is not required when uploading model code.
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.
Prediction environment The environment from which the deployed model makes predictions (either an internal, DataRobot environment or an external one).
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.
Enable data drift tracking Configures DataRobot to track target and feature data drift in a deployment.
Enable prediction row storage Store prediction request rows, allowing you to configure challenger models for the deployment.
Track attributes for segmented analysis A toggle that, when enabled, allows DataRobot to can perform segment analysis on prediction data.

DataRobot recommends you upload training data (if available) to your deployment to enable more capabilities. To do so, click Browse under the Learning header.

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.

Depending on the data and fields provided, DataRobot will indicate which deployment capabilities you have enabled:

Create deployments with training data

This section explains how to create an external deployment with training data. This deployment method allows you to upload historical predictions and analyze data drift and accuracy statistics in the past. You can also instrument the deployment with the MLOps Agent to monitor future predictions.

To create a deployment with training data, navigate to Deployments and click the Add Deployment link.

Under the Add a training dataset header, select browse to upload your XLSX, CSV, or TXT formatted training data. You can also select training data from the AI Catalog.

fter selecting your training dataset, provide information about the model that used the training data. Once completed, select Continue to deployment details to further configure the deployment.

Under the Inference header, complete the fields and select optional configurations for the deployment:

Field Description
Inference data The inference data is being stored by DataRobot and is not required when uploading model code.
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.
Prediction environment The environment from which the deployed model makes predictions (either an internal, DataRobot environment or an external one).
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.
Enable data drift tracking Configures DataRobot to track target and feature data drift in a deployment.
Enable prediction row storage Store prediction request rows, allowing you to configure challenger models for the deployment.
Track attributes for segmented analysis A toggle that, when enabled, allows DataRobot to can perform segment analysis on prediction data.

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, assign an importance level for the deployment, and then create the deployment.

Once you create an external deployment, there are two options for additional configuration. You can upload historical prediction data to the deployment to analyze data drift and accuracy in the past. You can also instrument the deployment with the MLOps Agent to monitor future predictions. To do so, navigate to the Predictions tabto access the monitoring snippet.

Add data post-deployment

Users with the Owner role can add historical prediction data to deployments if data drift is enabled. To do so, navigate to the Settings tab and select choose file under the Inference header to upload your prediction data in XLSX, CSV, or TXT format. You can also select prediction data from the AI Catalog.

Training data is a critical component for calculating data drift. If you did not include training data when you created a deployment, or if there was an error when uploading that data, add it from the Settings tab. You can also check for training data from the Data Drift tab.

The data must meet the following requirements:

  • Appended prediction data must have the same features as the original prediction dataset. After uploading new data, DataRobot prompts to confirm the addition, because you cannot later remove data from a deployment. To use different prediction data, create a new deployment.

  • An uploaded training dataset must include the same features as the prediction (scoring) dataset. You cannot replace training data. If you want a deployment to use different training data, create a new deployment with the appropriate data.


Updated November 16, 2021
Back to top