Skip to main content

Regular monitoring of adaptive models

Regular inspection

Adaptive models learn from all customer interactions, adjusting to changing behavior over time. To confirm the continuing accuracy of your adaptive models, perform the following tasks regularly:

  • Check the performance and success rate of your models every two weeks.
  • Inspect predictors every two or three months.

The purpose of regular inspection is to detect factors that negatively influence the performance of the adaptive models and the success rate of the actions.

Identifying technical problems

Look for adaptive models with a success rate of zero. This means that the actions for these models do not have any positive responses.

Identifying actions for which the model is not predictive

Look for adaptive models with low performance. Consider adding additional data as predictors.

Identifying actions that have a low number of responses

Look for adaptive models with a low number of responses. Discuss the eligibility criteria set in the Next-Best-Action Designer with the business. Changing the exclusion settings may increase the number of responses.

Identifying actions that are offered so often that they dominate other actions

Look for adaptive models with a high number of responses. A high number of responses might be fine from the business point of view. However, if necessary, prioritization can be adjusted in the Next-Best-Action Designer.

Identifying actions with a low success rate

Look for adaptive models with a low success rate. If the model performance is high, the relevance to the customers is high, but the action is unattractive and should be discussed with the business.

Inspecting an adaptive model

Inspect your model after introducing a new action, adding or removing a predictor, or changing prioritization. Take note of the active and inactive predictors.

Inspecting predictors

Check the details of a predictor with a low performance score. A possible cause can be too many missing values for the predictor. Look at the top predictors and in the bins that have a particularly high or low success rate.

Identifying predictors that are never used

Because unused predictors have only a minor effect on model performance, you do not need to remove them from an adaptive model configuration; however, you can conduct an occasional cleanup as part of your maintenance activities. An unused predictor might still become relevant for a future action.

This Topic is available in the following Modules:

If you are having problems with your training, please review the Pega Academy Support FAQs.

Did you find this content helpful?

Want to help us improve this content?

We'd prefer it if you saw us at our best.

Pega Academy has detected you are using a browser which may prevent you from experiencing the site as intended. To improve your experience, please update your browser.

Close Deprecation Notice