---
schema_version: "1"
id: 2c28d72ab01d732e4e77475d680e24e7e36ba07c32905c81a141bbd3a268b4f6
title: Configure a Feature Extractor
collection: admin
version: current
locale: en-us
portal_url: https://docs.salesforce.com/admin/current/en-us/ind/264-0-0/ai_accelerator_configure_feature_extractor.md
canonical_url: https://help.salesforce.com/s/articleView?id=ind.ai_accelerator_configure_feature_extractor.htm&release=264.0.0&type=5
source_site: help.salesforce.com
content_type: documentation
content_hash: 8803930fd94f7f103712da745bd9a04d80d6c81ae75c14459ccfbf4c37328463
description: For use cases that you create, configure feature extractors to compute features in real time or in batches, or upload sample features that the predictive model can use to generate predictions. For auto-generated use cases that require features computed in real time, you can extend the configuration of a partially suitable feature extractor by adding other feature extractors with your configured Apex classes. Or, you entirely update the configuration by using another feature extractor instead of the default feature extractor. The predictive model uses the features computed based on the configuration to generate predictions.
products:
  - Industry Solutions
  - Cross Cloud
release: "264"
---
# Configure a Feature Extractor

For use cases that you create, configure feature extractors to compute features in real time or in batches, or upload sample features that the predictive model can use to generate predictions. For auto-generated use cases that require features computed in real time, you can extend the configuration of a partially suitable feature extractor by adding other feature extractors with your configured Apex classes. Or, you entirely update the configuration by using another feature extractor instead of the default feature extractor. The predictive model uses the features computed based on the configuration to generate predictions.

### Required Editions

<table class="slds-table slds-table_bordered slds-max-medium-table_stacked slds-m-bottom_small edition"><colgroup><col></colgroup><tbody><tr class=""><td class="slds-cell-wrap" style="vertical-align:top;">Available in: Lightning Experience</td></tr></tbody></table>

<table class="slds-table slds-table_bordered slds-max-medium-table_stacked slds-m-bottom_small permission">
        <colgroup>
          <col style="width:50%">
          <col style="width:50%">
        </colgroup>
        <thead class="slds-text-title_caps ">
          <tr class="">
            <th scope="colgroup" class="slds-cell-wrap" colspan="2" style="text-align:center;vertical-align:top;" id="d6136e54">User Permissions Needed</th>
          </tr>
        </thead>
        <tbody>
          <tr class="">
            <td data-label="User Permissions Needed" class="slds-cell-wrap" style="vertical-align:top;" headers="d6136e54 ">To configure a feature extractor:</td>

            <td class="slds-cell-wrap" style="vertical-align:top;" headers="d6136e54 ">Customize Application</td>

          </tr>
        </tbody>
      </table>

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 Model Feature Extractor section.
3. Click **Add Feature Extractor**.
    
    You can configure multiple feature extractors for one use case model.
4. Select the predictive model that you want to compute features in real time or in batches for, or you want to upload sample features for.
5. Enter a unique **Extractor API Name** for the feature extractor.
6. Select a **Feature Input Type:**
    - If you want to use features from the supplemental dataset added when deploying the model, select Batch Input.
    - If you want to compute features in real time, select Real-Time Input.
    - If you want to upload sample features, select Sample Input.
7. Select a **Data Space**.
8. Select the **DMO** and the corresponding Record ID.
9. Select the Input Feature(s) and the corresponding field(s).
10. If you selected Real-Time Input as the feature input type, select an Apex class name for the model.
11. If you selected Sample Input as the feature input type, upload a CSV file that contains sample features.
    
    The CSV file can have only two rows, a row with the feature names and another row with the corresponding feature values.
    
    ![Tip](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-ind-ai_accelerator_scoring_framework-264-0-0-production-enus/579cbc22-5fe0-45e4-b0a0-608ab8690ae6/images/icon_tip.png)
    
    Tip After the sample features in the CSV are uploaded, to view the features, click View Uploaded File Data.
12. Click **Update**.
13. To specify a feature extractor as the default, select Default in the corresponding row.
    
    The default feature extractor can be replaced by another feature extractor when configuring the Einstein Predictions Using AI Accelerator scorecard or the AI Accelerator—Einstein Predictions & Recommendations scorecard. If you don’t specify any feature extractor while configuring the scorecards, the model uses the default feature extractor specified on the AI Accelerator Setup page.

![Example](https://sf-zdocs-cdn-prod.zoominsoftware.com/tdta-ind-ai_accelerator_scoring_framework-264-0-0-production-enus/579cbc22-5fe0-45e4-b0a0-608ab8690ae6/images/icon_example.png)

Example A mobile service provider wants to predict the likelihood of a customer switching to another service provider. The predictive model that the service provider uses requires features, such as the customer’s accounts, payment type, plan type, subscription period, and the total number of service requests for customers’ network data issues. All features, except the total number of customer service requests, which is computed in real time, are available in the customer’s records. So, the admin configures a feature extractor to calculate the sum of customer service requests for network data issues. The admin configures an Apex class and then configures a feature extractor by selecting the predictive model that gets the predictions and the configured Apex class.