Filter the Data for the Machine Learning Model

Duplicate and inactive requests negatively impact the quality of predictions.

If you want to clean up the data before it’s fed to the Machine Learning API, create exclusion rules for data ingestion. These rules help to filter out the duplicate or inactive data, so they’re excluded from being ingested by the ML model.

1. In the Setup and Maintenance work area, go to the following:
   • Offering: Service
   • Functional Area: Communication Channels
   • Task: Manage Email Configuration, Registration, and Validation
2. In the Inbound Email Configuration and Validation page, navigate to the SR Classification area.
3. In the Ingestion Exclusion Rule area, click Add.
4. In the Object field, select one of the following: Service Request, Internal Service Request, or HR Help Desk Request.
5. In the Attribute field, select the object attribute based on which you want to filter the data.
6. In the Operator field, select a suitable operator such as Contains, Equals, or Is Not Null, depending on the rule you’re creating.
7. In the Value field, specify the value for the exclusion rule.
   Here are some examples:

   • Requests created by messages from WhatsApp: Here, you could select the attribute as Channel Type, operator as Equals or Contains, and the value as WhatsApp.
   • Requests created by a user whose email pattern contains test: These requests aren’t to be considered for training the ML model, because this would be test data.
8. Click Save.