Set up deployment for a batch or streaming model

For model deployment, you can select Batch Prediction or Real-time Streaming Predictions for deployment types.

  1. Open a project and select a model.
  2. From the Deployments tab, click Add Deployment.
  3. On the Create Deployment page select a continuous learning deployment type by clicking one of the following tabs:

    • Batch Prediction
    • Real-time Streaming Predictions
  4. Give the deployment a name and description. In the Name box, type a name and in the Description box, type a description.

  5. Select an Apache Spark instance that is associated with the current project.
  6. Select a source. You must retrieve your credentials from IBM Cloud. For a Batch Prediction, this might be a file in your IBM Cloud Object Storage or a connection and table from IBM Db2 Warehouse on Cloud. For a Real-time Streaming Prediction this is a streaming data source from an Event Streams instance.
  7. Select a target. You must retrieve your credentials from IBM Cloud. For a Batch Prediction, this might be a file in your IBM Cloud Object Storage or a connection and table from IBM Db2 Warehouse on Cloud. For a Real-time Streaming Prediction this is a streaming data target from an Event Streams instance.
  8. For a Batch Prediction, you must select how you want the target to be updated.
  9. Click Save.

Next steps

For more information about continuous learning and model evaluation, see Continuous learning and model evaluation