Individual (Observable) user models calculate a set of attributes and their corresponding
event count for the user for the past 10 days. When the user deviates outside the predicted range or
uses a new value, the ML model determines with a high level of confidence that the event is
anomalous.