Reduce the amount of events required to build a model

Coveo Machine Learning (Coveo ML) models learn from data. If fewer than 100 events are available for your model, recommendations can’t be provided just yet.

Members with the required privileges can activate the Test configuration mode for a model when very few analytics are available to train a model.

To provide relevant output, Coveo ML models filter out the following by default:

  • Infrequent queries

  • Queries that were not followed by sufficient clicks

  • User context keys that were not seen frequently enough

  • Browsing patterns that were not performed frequently enough

When the test configuration mode is activated, the model reduces the amount of analytics data it needs to provide recommendations. This is ideal to test a model in a sandbox environment.

Example

In a non-production organization (sandbox), you create a model for testing purposes. Since you don’t want to generate a hundred events and want the model to provide recommendations immediately, you activate the test configuration mode for this model.

To activate test configuration mode

Follow the steps for your model type: