Determining impact of features on individual prediction of machine learning based models

A system trains a machine learning based model to predict the likelihood of an outcome for an entity, for example, a user. The system determines, for a particular prediction for a user, feature impact scores that indicate how each feature of the user impacted the prediction for that user. The feature impact scores are ranked to determine top driver features for the user that had the highest impact on the prediction. The system generates a human understandable description for the top driver features. The system provides the generated description for the top driver features for display, for example, via a user interface.