Predict allows you to automatically triage incoming Issues and assign relevant Labels to them. You can use the assigned Label to setup corresponding Automations, Smart Views, Advanced Search, and more.

You can create models:

  • At a domain level, where all Issues from all your support channels are processed and labeled using a single model. 
  • At an app level, where Issues from various apps, website, email, and web contact us form are processed and labeled using different models.  

To create a Predict Model for a Language:

  1. Navigate to the Predict Dashboard. This page gives you an overview of all the Models for different languages.

  2. Click the Create New Model button, and select the language in which you want to create a model from the dropdown.

  3. Next, you will be given the option to customize the Model Name, and select the types of Issues that the model should apply to. For Model Name, we recommend you to follow  “<Language> – <configuration name>” naming format.
  4.  The following configuration settings are available for each Model:
    • All <Language> Issues: Select this option if the Model should apply Labels to All Issues for that <Language>.
    • <Language> Issues created via Email: Select this option, if the Model should apply Labels to all the <Language> Issues that are filed via Email.
    • <Language> Issues created via Web Contact Us Form: Select this option, if the Model should apply Labels to all the <Language> Issues that are filed via Web Contact Us form.
    • <Language> Issues from specific Apps: Select this option, if the Model should apply Labels to the <Language> Issues filed via specific Apps (only applicable for In-App and Web Chat issues).
      For the app-specific models, select the option <Language> Issues from Specific Apps and pick the App(s) from the drop-down list.
      The Apps that you’ve selected for the Model will be listed in the Configuration. An app can only be part of one Model for a language.
  5. Once you’ve selected and saved the Model Configuration, follow the next steps to train the model. Provide details specific to Prepare Data, Upload Data, Verify Data and Train the Model. For more information, refer to How do I prepare my data for the Predict Model?
    You can change the Model configuration at any time once the Model is created.

Update Existing Configuration to Support Multiple Models

To  split and update an existing model:

  1. Navigate to the Model you want to update, and click on the gear icon ⚙️ at the top right corner of that model. Select Edit Configuration from the drop-down menu.
  2. On the Edit Configuration modal that pops up, select the types of Issues you want the model to apply to, and click Save.
  3. Once you Save the configuration, all Issues from then onwards will be processed and labeled based on the updated model configuration.
    It is important to note that changing the configuration of the Model might have an impact on the coverage and accuracy of the Model based on the new configuration. You might need to create other Model(s) to cover remaining sets of Issues for the given language.