Salesforce Docs Portal

Select a Predictive Model

Select the predictive model that you deployed to get predictions for your use case.

Required Editions

Available in: Lightning Experience
User Permissions Needed
To select a model: Customize Application
  1. From Setup, in the Quick Find box, enter Industries Einstein, and then select AI Accelerator.
  2. Click Edit for your use case, and then go to the Use Case Model section.
  3. Click Add Model.
    A row with a Prediction Platform field and Model field is added.
  4. Select the Prediction Platform.
    1. Einstein Discovery: Select this option if you want to use Salesforce's Einstein Discovery tool for predictive analytics. By default, this option is selected.
    2. Data Cloud BYOM (Bring Your Own Model): Select this option if you want to integrate a machine learning model built in Amazon SageMaker into your use case.
    3. Einstein on Data Cloud: Select this option if you want to use Salesforce's Einstein capabilities on Data Cloud for your predictive model.
  5. Select a machine learning model.
    You can select multiple machine learning models for a use case.
    Tip
    Tip If no models are available to select, no unused models exist in your org. Create an Insights & Predictions model, deploy the predictive model, and then retry.
  6. If you’ve selected Einstein on Data Cloud and Data Cloud BYOM (Bring Your Own Model), complete these steps:
    1. Enter Usecase Model Name to identify the particular use case model. Based on the Usecase Model Name entered, the Usecase Model API Name is auto-filled.
    2. Select Prediction Mode. You can select either Generate in real time with AI Accelerator or Already generated. If you select Generate in real time with AI Accelerator, you can select the Model. If you select Already Generated, you can select a Data Space.
    3. Select a score field that displays the score and any additional variables selected. The options are listed based on the selected model.
    4. To include other fields in the prediction card, click Add Input Field. Then, enter the field label and the source field to map.
    5. Select a Reference Record field that is associated with the primary key of the record. This allows the model to be queried using this primary key and displays all fields on the card.
    6. To specify the prediction sorting order and set the display order in the scorecard, click Add Sort Preference. Then, select the source field and sort order.
    7. Create a template that defines how to display the selected fields on the prediction card. You can add variables such as %%Score%%, which represents the likelihood of asset failure. If you select multiple fields, you can write text that includes these variables, and they will be shown on the prediction card.
  7. Save your changes.
    Note
    Note If a model selected for a use case is deleted, the use case becomes invalid.