Models and Insights models-and-insights
The performance of your models depends on the quality and completeness of the input data. See the top influencing factor for each of your likelihood AI models. Also, see the top factors resulting in higher/lower event registration, event attendance, or unsubscribes.
NOTE
Behaviors marked with (+) influence predictions positively (and vice versa).
Here’s how to assess your model’s health.
Navigate to the Models and Data Health section under Predictive Audiences in the Admin area of Marketo Engage. Here you’ll see all of your models and their statuses.
- Training Status: Indicates whether your model is actively training (improving predictions). Training automatically occurs every two weeks. Any models that are Processing could take up to 24 hours to finish. For any Failed models, please contact Marketo Support.
- Scoring Status: Indicates whether your model is actively calculating predictions (likelihood percentages) for program members.
- Performance: Categorization of your model’s health based on Data Completeness and Data Quality (see below).
- Data Completeness: Percentage of data attributes that are present/complete.
- Data Quality: Percentage of attributes that contain good, usable data.
- Last Trained: The date of the model that is the best out of the evaluation between the current model versus the new model that is trained every two weeks.
recommendation-more-help
94ec3174-1d6c-4f51-822d-5424bedeecac